0%|          | 1/13317 [00:01<5:25:10,  1.47s/it]
                                                   
{'loss': 0.9494, 'grad_norm': 2.646820306777954, 'learning_rate': 7.507507507507508e-09, 'epoch': 0.0}

  0%|          | 1/13317 [00:01<5:25:10,  1.47s/it]
  0%|          | 2/13317 [00:02<3:25:03,  1.08it/s]
                                                   
{'loss': 0.865, 'grad_norm': 2.3400673866271973, 'learning_rate': 1.5015015015015016e-08, 'epoch': 0.0}

  0%|          | 2/13317 [00:02<3:25:03,  1.08it/s]
  0%|          | 3/13317 [00:02<2:53:23,  1.28it/s]
                                                   
{'loss': 0.8431, 'grad_norm': 2.388704299926758, 'learning_rate': 2.2522522522522524e-08, 'epoch': 0.0}

  0%|          | 3/13317 [00:02<2:53:23,  1.28it/s]
  0%|          | 4/13317 [00:03<2:38:43,  1.40it/s]
                                                   
{'loss': 0.8816, 'grad_norm': 2.6599130630493164, 'learning_rate': 3.003003003003003e-08, 'epoch': 0.0}

  0%|          | 4/13317 [00:03<2:38:43,  1.40it/s]
  0%|          | 5/13317 [00:03<2:30:46,  1.47it/s]
                                                   
{'loss': 1.0193, 'grad_norm': 2.8670766353607178, 'learning_rate': 3.753753753753754e-08, 'epoch': 0.0}

  0%|          | 5/13317 [00:03<2:30:46,  1.47it/s]
  0%|          | 6/13317 [00:04<2:25:59,  1.52it/s]
                                                   
{'loss': 0.8743, 'grad_norm': 2.447897434234619, 'learning_rate': 4.504504504504505e-08, 'epoch': 0.0}

  0%|          | 6/13317 [00:04<2:25:59,  1.52it/s]
  0%|          | 7/13317 [00:05<2:22:48,  1.55it/s]
                                                   
{'loss': 1.0061, 'grad_norm': 2.9875383377075195, 'learning_rate': 5.255255255255256e-08, 'epoch': 0.0}

  0%|          | 7/13317 [00:05<2:22:48,  1.55it/s]
  0%|          | 8/13317 [00:05<2:20:38,  1.58it/s]
                                                   
{'loss': 0.974, 'grad_norm': 2.3990159034729004, 'learning_rate': 6.006006006006006e-08, 'epoch': 0.0}

  0%|          | 8/13317 [00:05<2:20:38,  1.58it/s]
  0%|          | 9/13317 [00:06<2:19:14,  1.59it/s]
                                                   
{'loss': 1.2645, 'grad_norm': 2.6647093296051025, 'learning_rate': 6.756756756756757e-08, 'epoch': 0.0}

  0%|          | 9/13317 [00:06<2:19:14,  1.59it/s]
  0%|          | 10/13317 [00:06<2:18:46,  1.60it/s]
                                                    
{'loss': 0.9874, 'grad_norm': 2.5233733654022217, 'learning_rate': 7.507507507507508e-08, 'epoch': 0.0}

  0%|          | 10/13317 [00:07<2:18:46,  1.60it/s]
  0%|          | 11/13317 [00:07<2:17:58,  1.61it/s]
                                                    
{'loss': 0.899, 'grad_norm': 2.0676119327545166, 'learning_rate': 8.258258258258259e-08, 'epoch': 0.0}

  0%|          | 11/13317 [00:07<2:17:58,  1.61it/s]
  0%|          | 12/13317 [00:08<2:17:30,  1.61it/s]
                                                    
{'loss': 0.9271, 'grad_norm': 2.888087034225464, 'learning_rate': 9.00900900900901e-08, 'epoch': 0.0}

  0%|          | 12/13317 [00:08<2:17:30,  1.61it/s]
  0%|          | 13/13317 [00:08<2:17:09,  1.62it/s]
                                                    
{'loss': 1.0218, 'grad_norm': 3.219238042831421, 'learning_rate': 9.75975975975976e-08, 'epoch': 0.0}

  0%|          | 13/13317 [00:08<2:17:09,  1.62it/s]
  0%|          | 14/13317 [00:09<2:16:52,  1.62it/s]
                                                    
{'loss': 1.1003, 'grad_norm': 3.288654327392578, 'learning_rate': 1.0510510510510513e-07, 'epoch': 0.0}

  0%|          | 14/13317 [00:09<2:16:52,  1.62it/s]
  0%|          | 15/13317 [00:10<2:16:57,  1.62it/s]
                                                    
{'loss': 1.1175, 'grad_norm': 3.4702346324920654, 'learning_rate': 1.1261261261261262e-07, 'epoch': 0.0}

  0%|          | 15/13317 [00:10<2:16:57,  1.62it/s]
  0%|          | 16/13317 [00:10<2:16:43,  1.62it/s]
                                                    
{'loss': 0.7981, 'grad_norm': 2.412877321243286, 'learning_rate': 1.2012012012012013e-07, 'epoch': 0.0}

  0%|          | 16/13317 [00:10<2:16:43,  1.62it/s]
  0%|          | 17/13317 [00:11<2:16:36,  1.62it/s]
                                                    
{'loss': 0.9397, 'grad_norm': 2.726624011993408, 'learning_rate': 1.2762762762762765e-07, 'epoch': 0.0}

  0%|          | 17/13317 [00:11<2:16:36,  1.62it/s]
  0%|          | 18/13317 [00:11<2:16:27,  1.62it/s]
                                                    
{'loss': 0.9479, 'grad_norm': 2.877187967300415, 'learning_rate': 1.3513513513513515e-07, 'epoch': 0.0}

  0%|          | 18/13317 [00:11<2:16:27,  1.62it/s]
  0%|          | 19/13317 [00:12<2:16:26,  1.62it/s]
                                                    
{'loss': 0.9864, 'grad_norm': 3.8969573974609375, 'learning_rate': 1.4264264264264264e-07, 'epoch': 0.0}

  0%|          | 19/13317 [00:12<2:16:26,  1.62it/s]
  0%|          | 20/13317 [00:13<2:16:41,  1.62it/s]
                                                    
{'loss': 0.9975, 'grad_norm': 3.142580032348633, 'learning_rate': 1.5015015015015016e-07, 'epoch': 0.0}

  0%|          | 20/13317 [00:13<2:16:41,  1.62it/s]
  0%|          | 21/13317 [00:13<2:16:48,  1.62it/s]
                                                    
{'loss': 0.8595, 'grad_norm': 2.11672043800354, 'learning_rate': 1.5765765765765766e-07, 'epoch': 0.0}

  0%|          | 21/13317 [00:13<2:16:48,  1.62it/s]
  0%|          | 22/13317 [00:14<2:16:35,  1.62it/s]
                                                    
{'loss': 0.9348, 'grad_norm': 2.5887250900268555, 'learning_rate': 1.6516516516516518e-07, 'epoch': 0.0}

  0%|          | 22/13317 [00:14<2:16:35,  1.62it/s]
  0%|          | 23/13317 [00:14<2:16:28,  1.62it/s]
                                                    
{'loss': 1.0693, 'grad_norm': 3.3582563400268555, 'learning_rate': 1.7267267267267267e-07, 'epoch': 0.01}

  0%|          | 23/13317 [00:15<2:16:28,  1.62it/s]
  0%|          | 24/13317 [00:15<2:16:23,  1.62it/s]
                                                    
{'loss': 0.9229, 'grad_norm': 2.886667490005493, 'learning_rate': 1.801801801801802e-07, 'epoch': 0.01}

  0%|          | 24/13317 [00:15<2:16:23,  1.62it/s]
  0%|          | 25/13317 [00:16<2:16:33,  1.62it/s]
                                                    
{'loss': 0.8707, 'grad_norm': 2.1457412242889404, 'learning_rate': 1.8768768768768772e-07, 'epoch': 0.01}

  0%|          | 25/13317 [00:16<2:16:33,  1.62it/s]
  0%|          | 26/13317 [00:16<2:16:22,  1.62it/s]
                                                    
{'loss': 0.9686, 'grad_norm': 2.731424331665039, 'learning_rate': 1.951951951951952e-07, 'epoch': 0.01}

  0%|          | 26/13317 [00:16<2:16:22,  1.62it/s]
  0%|          | 27/13317 [00:17<2:16:12,  1.63it/s]
                                                    
{'loss': 0.9873, 'grad_norm': 2.6451382637023926, 'learning_rate': 2.0270270270270273e-07, 'epoch': 0.01}

  0%|          | 27/13317 [00:17<2:16:12,  1.63it/s]
  0%|          | 28/13317 [00:18<2:16:08,  1.63it/s]
                                                    
{'loss': 0.9351, 'grad_norm': 2.7847933769226074, 'learning_rate': 2.1021021021021025e-07, 'epoch': 0.01}

  0%|          | 28/13317 [00:18<2:16:08,  1.63it/s]
  0%|          | 29/13317 [00:18<2:16:04,  1.63it/s]
                                                    
{'loss': 0.9784, 'grad_norm': 2.752998113632202, 'learning_rate': 2.1771771771771772e-07, 'epoch': 0.01}

  0%|          | 29/13317 [00:18<2:16:04,  1.63it/s]
  0%|          | 30/13317 [00:19<2:16:13,  1.63it/s]
                                                    
{'loss': 1.0221, 'grad_norm': 3.005768299102783, 'learning_rate': 2.2522522522522524e-07, 'epoch': 0.01}

  0%|          | 30/13317 [00:19<2:16:13,  1.63it/s]
  0%|          | 31/13317 [00:19<2:16:04,  1.63it/s]
                                                    
{'loss': 0.9297, 'grad_norm': 2.387977123260498, 'learning_rate': 2.3273273273273274e-07, 'epoch': 0.01}

  0%|          | 31/13317 [00:19<2:16:04,  1.63it/s]
  0%|          | 32/13317 [00:20<2:16:02,  1.63it/s]
                                                    
{'loss': 0.9542, 'grad_norm': 2.441345691680908, 'learning_rate': 2.4024024024024026e-07, 'epoch': 0.01}

  0%|          | 32/13317 [00:20<2:16:02,  1.63it/s]
  0%|          | 33/13317 [00:21<2:15:57,  1.63it/s]
                                                    
{'loss': 0.9051, 'grad_norm': 3.5449955463409424, 'learning_rate': 2.477477477477478e-07, 'epoch': 0.01}

  0%|          | 33/13317 [00:21<2:15:57,  1.63it/s]
  0%|          | 34/13317 [00:21<2:15:52,  1.63it/s]
                                                    
{'loss': 0.8718, 'grad_norm': 2.3236653804779053, 'learning_rate': 2.552552552552553e-07, 'epoch': 0.01}

  0%|          | 34/13317 [00:21<2:15:52,  1.63it/s]
  0%|          | 35/13317 [00:22<2:15:55,  1.63it/s]
                                                    
{'loss': 0.9744, 'grad_norm': 2.996504068374634, 'learning_rate': 2.6276276276276277e-07, 'epoch': 0.01}

  0%|          | 35/13317 [00:22<2:15:55,  1.63it/s]
  0%|          | 36/13317 [00:22<2:15:54,  1.63it/s]
                                                    
{'loss': 1.0229, 'grad_norm': 3.263207197189331, 'learning_rate': 2.702702702702703e-07, 'epoch': 0.01}

  0%|          | 36/13317 [00:23<2:15:54,  1.63it/s]
  0%|          | 37/13317 [00:23<2:15:52,  1.63it/s]
                                                    
{'loss': 0.8876, 'grad_norm': 2.4195640087127686, 'learning_rate': 2.7777777777777776e-07, 'epoch': 0.01}

  0%|          | 37/13317 [00:23<2:15:52,  1.63it/s]
  0%|          | 38/13317 [00:24<2:15:47,  1.63it/s]
                                                    
{'loss': 0.9029, 'grad_norm': 3.4358081817626953, 'learning_rate': 2.852852852852853e-07, 'epoch': 0.01}

  0%|          | 38/13317 [00:24<2:15:47,  1.63it/s]
  0%|          | 39/13317 [00:24<2:15:44,  1.63it/s]
                                                    
{'loss': 0.9481, 'grad_norm': 2.456697940826416, 'learning_rate': 2.927927927927928e-07, 'epoch': 0.01}

  0%|          | 39/13317 [00:24<2:15:44,  1.63it/s]
  0%|          | 40/13317 [00:25<2:15:51,  1.63it/s]
                                                    
{'loss': 0.8763, 'grad_norm': 2.5605924129486084, 'learning_rate': 3.003003003003003e-07, 'epoch': 0.01}

  0%|          | 40/13317 [00:25<2:15:51,  1.63it/s]
  0%|          | 41/13317 [00:25<2:15:44,  1.63it/s]
                                                    
{'loss': 0.8957, 'grad_norm': 2.5799906253814697, 'learning_rate': 3.0780780780780784e-07, 'epoch': 0.01}

  0%|          | 41/13317 [00:26<2:15:44,  1.63it/s]
  0%|          | 42/13317 [00:26<2:15:41,  1.63it/s]
                                                    
{'loss': 0.8849, 'grad_norm': 2.680360794067383, 'learning_rate': 3.153153153153153e-07, 'epoch': 0.01}

  0%|          | 42/13317 [00:26<2:15:41,  1.63it/s]
  0%|          | 43/13317 [00:27<2:15:51,  1.63it/s]
                                                    
{'loss': 0.9718, 'grad_norm': 2.6914453506469727, 'learning_rate': 3.2282282282282283e-07, 'epoch': 0.01}

  0%|          | 43/13317 [00:27<2:15:51,  1.63it/s]
  0%|          | 44/13317 [00:27<2:15:50,  1.63it/s]
                                                    
{'loss': 0.9262, 'grad_norm': 2.440049171447754, 'learning_rate': 3.3033033033033036e-07, 'epoch': 0.01}

  0%|          | 44/13317 [00:27<2:15:50,  1.63it/s]
  0%|          | 45/13317 [00:28<2:15:51,  1.63it/s]
                                                    
{'loss': 0.8985, 'grad_norm': 2.3033666610717773, 'learning_rate': 3.378378378378379e-07, 'epoch': 0.01}

  0%|          | 45/13317 [00:28<2:15:51,  1.63it/s]
  0%|          | 46/13317 [00:29<2:15:49,  1.63it/s]
                                                    
{'loss': 1.048, 'grad_norm': 2.7851650714874268, 'learning_rate': 3.4534534534534535e-07, 'epoch': 0.01}

  0%|          | 46/13317 [00:29<2:15:49,  1.63it/s]
  0%|          | 47/13317 [00:29<2:15:43,  1.63it/s]
                                                    
{'loss': 0.8574, 'grad_norm': 2.1973562240600586, 'learning_rate': 3.528528528528529e-07, 'epoch': 0.01}

  0%|          | 47/13317 [00:29<2:15:43,  1.63it/s]
  0%|          | 48/13317 [00:30<2:15:40,  1.63it/s]
                                                    
{'loss': 0.8476, 'grad_norm': 2.209846019744873, 'learning_rate': 3.603603603603604e-07, 'epoch': 0.01}

  0%|          | 48/13317 [00:30<2:15:40,  1.63it/s]
  0%|          | 49/13317 [00:30<2:15:37,  1.63it/s]
                                                    
{'loss': 0.7862, 'grad_norm': 2.46240234375, 'learning_rate': 3.6786786786786786e-07, 'epoch': 0.01}

  0%|          | 49/13317 [00:31<2:15:37,  1.63it/s]
  0%|          | 50/13317 [00:31<2:15:36,  1.63it/s]
                                                    
{'loss': 1.0251, 'grad_norm': 3.1315271854400635, 'learning_rate': 3.7537537537537543e-07, 'epoch': 0.01}

  0%|          | 50/13317 [00:31<2:15:36,  1.63it/s]
  0%|          | 51/13317 [00:32<2:15:33,  1.63it/s]
                                                    
{'loss': 0.9626, 'grad_norm': 2.0481953620910645, 'learning_rate': 3.828828828828829e-07, 'epoch': 0.01}

  0%|          | 51/13317 [00:32<2:15:33,  1.63it/s]
  0%|          | 52/13317 [00:32<2:15:37,  1.63it/s]
                                                    
{'loss': 0.9023, 'grad_norm': 3.5687384605407715, 'learning_rate': 3.903903903903904e-07, 'epoch': 0.01}

  0%|          | 52/13317 [00:32<2:15:37,  1.63it/s]
  0%|          | 53/13317 [00:33<2:15:58,  1.63it/s]
                                                    
{'loss': 0.8777, 'grad_norm': 2.469125747680664, 'learning_rate': 3.978978978978979e-07, 'epoch': 0.01}

  0%|          | 53/13317 [00:33<2:15:58,  1.63it/s]
  0%|          | 54/13317 [00:33<2:15:44,  1.63it/s]
                                                    
{'loss': 0.9081, 'grad_norm': 2.282897472381592, 'learning_rate': 4.0540540540540546e-07, 'epoch': 0.01}

  0%|          | 54/13317 [00:34<2:15:44,  1.63it/s]
  0%|          | 55/13317 [00:34<2:15:40,  1.63it/s]
                                                    
{'loss': 0.8049, 'grad_norm': 2.256535291671753, 'learning_rate': 4.1291291291291293e-07, 'epoch': 0.01}

  0%|          | 55/13317 [00:34<2:15:40,  1.63it/s]
  0%|          | 56/13317 [00:35<2:15:42,  1.63it/s]
                                                    
{'loss': 0.8434, 'grad_norm': 1.9026448726654053, 'learning_rate': 4.204204204204205e-07, 'epoch': 0.01}

  0%|          | 56/13317 [00:35<2:15:42,  1.63it/s]
  0%|          | 57/13317 [00:35<2:15:45,  1.63it/s]
                                                    
{'loss': 0.992, 'grad_norm': 2.9717507362365723, 'learning_rate': 4.27927927927928e-07, 'epoch': 0.01}

  0%|          | 57/13317 [00:35<2:15:45,  1.63it/s]
  0%|          | 58/13317 [00:36<2:15:43,  1.63it/s]
                                                    
{'loss': 0.8866, 'grad_norm': 2.3037383556365967, 'learning_rate': 4.3543543543543544e-07, 'epoch': 0.01}

  0%|          | 58/13317 [00:36<2:15:43,  1.63it/s]
  0%|          | 59/13317 [00:37<2:15:43,  1.63it/s]
                                                    
{'loss': 0.9086, 'grad_norm': 3.0257582664489746, 'learning_rate': 4.4294294294294296e-07, 'epoch': 0.01}

  0%|          | 59/13317 [00:37<2:15:43,  1.63it/s]
  0%|          | 60/13317 [00:37<2:15:48,  1.63it/s]
                                                    
{'loss': 0.8835, 'grad_norm': 2.610701084136963, 'learning_rate': 4.504504504504505e-07, 'epoch': 0.01}

  0%|          | 60/13317 [00:37<2:15:48,  1.63it/s]
  0%|          | 61/13317 [00:38<2:15:42,  1.63it/s]
                                                    
{'loss': 1.0007, 'grad_norm': 3.014319896697998, 'learning_rate': 4.57957957957958e-07, 'epoch': 0.01}

  0%|          | 61/13317 [00:38<2:15:42,  1.63it/s]
  0%|          | 62/13317 [00:38<2:15:41,  1.63it/s]
                                                    
{'loss': 1.0755, 'grad_norm': 2.6657915115356445, 'learning_rate': 4.654654654654655e-07, 'epoch': 0.01}

  0%|          | 62/13317 [00:39<2:15:41,  1.63it/s]
  0%|          | 63/13317 [00:39<2:15:35,  1.63it/s]
                                                    
{'loss': 0.8531, 'grad_norm': 3.8196001052856445, 'learning_rate': 4.7297297297297305e-07, 'epoch': 0.01}

  0%|          | 63/13317 [00:39<2:15:35,  1.63it/s]
  0%|          | 64/13317 [00:40<2:15:32,  1.63it/s]
                                                    
{'loss': 0.9078, 'grad_norm': 2.7949588298797607, 'learning_rate': 4.804804804804805e-07, 'epoch': 0.01}

  0%|          | 64/13317 [00:40<2:15:32,  1.63it/s]
  0%|          | 65/13317 [00:40<2:15:27,  1.63it/s]
                                                    
{'loss': 0.9631, 'grad_norm': 2.4826414585113525, 'learning_rate': 4.87987987987988e-07, 'epoch': 0.01}

  0%|          | 65/13317 [00:40<2:15:27,  1.63it/s]
  0%|          | 66/13317 [00:41<2:15:32,  1.63it/s]
                                                    
{'loss': 1.0726, 'grad_norm': 2.6283771991729736, 'learning_rate': 4.954954954954956e-07, 'epoch': 0.01}

  0%|          | 66/13317 [00:41<2:15:32,  1.63it/s]
  1%|          | 67/13317 [00:41<2:15:29,  1.63it/s]
                                                    
{'loss': 1.0398, 'grad_norm': 2.652327537536621, 'learning_rate': 5.03003003003003e-07, 'epoch': 0.02}

  1%|          | 67/13317 [00:42<2:15:29,  1.63it/s]
  1%|          | 68/13317 [00:42<2:15:31,  1.63it/s]
                                                    
{'loss': 0.9084, 'grad_norm': 2.3473963737487793, 'learning_rate': 5.105105105105106e-07, 'epoch': 0.02}

  1%|          | 68/13317 [00:42<2:15:31,  1.63it/s]
  1%|          | 69/13317 [00:43<2:15:31,  1.63it/s]
                                                    
{'loss': 0.8491, 'grad_norm': 1.987104892730713, 'learning_rate': 5.180180180180181e-07, 'epoch': 0.02}

  1%|          | 69/13317 [00:43<2:15:31,  1.63it/s]
  1%|          | 70/13317 [00:43<2:15:27,  1.63it/s]
                                                    
{'loss': 0.81, 'grad_norm': 1.9609936475753784, 'learning_rate': 5.255255255255255e-07, 'epoch': 0.02}

  1%|          | 70/13317 [00:43<2:15:27,  1.63it/s]
  1%|          | 71/13317 [00:44<2:15:31,  1.63it/s]
                                                    
{'loss': 0.8824, 'grad_norm': 2.3594648838043213, 'learning_rate': 5.330330330330331e-07, 'epoch': 0.02}

  1%|          | 71/13317 [00:44<2:15:31,  1.63it/s]
  1%|          | 72/13317 [00:45<2:15:25,  1.63it/s]
                                                    
{'loss': 0.9915, 'grad_norm': 2.994713306427002, 'learning_rate': 5.405405405405406e-07, 'epoch': 0.02}

  1%|          | 72/13317 [00:45<2:15:25,  1.63it/s]
  1%|          | 73/13317 [00:45<2:15:22,  1.63it/s]
                                                    
{'loss': 0.9885, 'grad_norm': 2.5328166484832764, 'learning_rate': 5.48048048048048e-07, 'epoch': 0.02}

  1%|          | 73/13317 [00:45<2:15:22,  1.63it/s]
  1%|          | 74/13317 [00:46<2:15:27,  1.63it/s]
                                                    
{'loss': 1.043, 'grad_norm': 2.9895694255828857, 'learning_rate': 5.555555555555555e-07, 'epoch': 0.02}

  1%|          | 74/13317 [00:46<2:15:27,  1.63it/s]
  1%|          | 75/13317 [00:46<2:15:25,  1.63it/s]
                                                    
{'loss': 0.8529, 'grad_norm': 2.164886951446533, 'learning_rate': 5.630630630630631e-07, 'epoch': 0.02}

  1%|          | 75/13317 [00:46<2:15:25,  1.63it/s]
  1%|          | 76/13317 [00:47<2:15:23,  1.63it/s]
                                                    
{'loss': 0.9133, 'grad_norm': 2.141491651535034, 'learning_rate': 5.705705705705706e-07, 'epoch': 0.02}

  1%|          | 76/13317 [00:47<2:15:23,  1.63it/s]
  1%|          | 77/13317 [00:48<2:15:21,  1.63it/s]
                                                    
{'loss': 0.9168, 'grad_norm': 2.2878506183624268, 'learning_rate': 5.780780780780781e-07, 'epoch': 0.02}

  1%|          | 77/13317 [00:48<2:15:21,  1.63it/s]
  1%|          | 78/13317 [00:48<2:15:19,  1.63it/s]
                                                    
{'loss': 0.8776, 'grad_norm': 2.0337722301483154, 'learning_rate': 5.855855855855856e-07, 'epoch': 0.02}

  1%|          | 78/13317 [00:48<2:15:19,  1.63it/s]
  1%|          | 79/13317 [00:49<2:15:19,  1.63it/s]
                                                    
{'loss': 0.7832, 'grad_norm': 2.0297250747680664, 'learning_rate': 5.930930930930932e-07, 'epoch': 0.02}

  1%|          | 79/13317 [00:49<2:15:19,  1.63it/s]
  1%|          | 80/13317 [00:49<2:15:18,  1.63it/s]
                                                    
{'loss': 0.98, 'grad_norm': 2.2675063610076904, 'learning_rate': 6.006006006006006e-07, 'epoch': 0.02}

  1%|          | 80/13317 [00:50<2:15:18,  1.63it/s]
  1%|          | 81/13317 [00:50<2:15:22,  1.63it/s]
                                                    
{'loss': 0.8088, 'grad_norm': 2.5821239948272705, 'learning_rate': 6.081081081081082e-07, 'epoch': 0.02}

  1%|          | 81/13317 [00:50<2:15:22,  1.63it/s]
  1%|          | 82/13317 [00:51<2:15:21,  1.63it/s]
                                                    
{'loss': 0.8705, 'grad_norm': 2.456202507019043, 'learning_rate': 6.156156156156157e-07, 'epoch': 0.02}

  1%|          | 82/13317 [00:51<2:15:21,  1.63it/s]
  1%|          | 83/13317 [00:51<2:15:20,  1.63it/s]
                                                    
{'loss': 0.7088, 'grad_norm': 1.882301688194275, 'learning_rate': 6.231231231231232e-07, 'epoch': 0.02}

  1%|          | 83/13317 [00:51<2:15:20,  1.63it/s]
  1%|          | 84/13317 [00:52<2:15:17,  1.63it/s]
                                                    
{'loss': 0.7914, 'grad_norm': 1.9859850406646729, 'learning_rate': 6.306306306306306e-07, 'epoch': 0.02}

  1%|          | 84/13317 [00:52<2:15:17,  1.63it/s]
  1%|          | 85/13317 [00:52<2:15:14,  1.63it/s]
                                                    
{'loss': 1.0035, 'grad_norm': 2.474757671356201, 'learning_rate': 6.381381381381382e-07, 'epoch': 0.02}

  1%|          | 85/13317 [00:53<2:15:14,  1.63it/s]
  1%|          | 86/13317 [00:53<2:15:13,  1.63it/s]
                                                    
{'loss': 0.8838, 'grad_norm': 2.3561854362487793, 'learning_rate': 6.456456456456457e-07, 'epoch': 0.02}

  1%|          | 86/13317 [00:53<2:15:13,  1.63it/s]
  1%|          | 87/13317 [00:54<2:15:14,  1.63it/s]
                                                    
{'loss': 0.9739, 'grad_norm': 2.212247371673584, 'learning_rate': 6.531531531531532e-07, 'epoch': 0.02}

  1%|          | 87/13317 [00:54<2:15:14,  1.63it/s]
  1%|          | 88/13317 [00:54<2:15:07,  1.63it/s]
                                                    
{'loss': 0.7848, 'grad_norm': 2.2994441986083984, 'learning_rate': 6.606606606606607e-07, 'epoch': 0.02}

  1%|          | 88/13317 [00:54<2:15:07,  1.63it/s]
  1%|          | 89/13317 [00:55<2:15:10,  1.63it/s]
                                                    
{'loss': 0.7899, 'grad_norm': 2.832188606262207, 'learning_rate': 6.681681681681683e-07, 'epoch': 0.02}

  1%|          | 89/13317 [00:55<2:15:10,  1.63it/s]
  1%|          | 90/13317 [00:56<2:15:12,  1.63it/s]
                                                    
{'loss': 0.8165, 'grad_norm': 1.9242957830429077, 'learning_rate': 6.756756756756758e-07, 'epoch': 0.02}

  1%|          | 90/13317 [00:56<2:15:12,  1.63it/s]
  1%|          | 91/13317 [00:56<2:15:08,  1.63it/s]
                                                    
{'loss': 0.9076, 'grad_norm': 2.4717724323272705, 'learning_rate': 6.831831831831832e-07, 'epoch': 0.02}

  1%|          | 91/13317 [00:56<2:15:08,  1.63it/s]
  1%|          | 92/13317 [00:57<2:15:10,  1.63it/s]
                                                    
{'loss': 0.9452, 'grad_norm': 2.2474710941314697, 'learning_rate': 6.906906906906907e-07, 'epoch': 0.02}

  1%|          | 92/13317 [00:57<2:15:10,  1.63it/s]
  1%|          | 93/13317 [00:57<2:15:10,  1.63it/s]
                                                    
{'loss': 0.852, 'grad_norm': 2.179138660430908, 'learning_rate': 6.981981981981982e-07, 'epoch': 0.02}

  1%|          | 93/13317 [00:58<2:15:10,  1.63it/s]
  1%|          | 94/13317 [00:58<2:15:10,  1.63it/s]
                                                    
{'loss': 0.8762, 'grad_norm': 2.266545295715332, 'learning_rate': 7.057057057057058e-07, 'epoch': 0.02}

  1%|          | 94/13317 [00:58<2:15:10,  1.63it/s]
  1%|          | 95/13317 [00:59<2:15:08,  1.63it/s]
                                                    
{'loss': 0.8161, 'grad_norm': 2.158430337905884, 'learning_rate': 7.132132132132133e-07, 'epoch': 0.02}

  1%|          | 95/13317 [00:59<2:15:08,  1.63it/s]
  1%|          | 96/13317 [00:59<2:15:10,  1.63it/s]
                                                    
{'loss': 0.7645, 'grad_norm': 2.0708978176116943, 'learning_rate': 7.207207207207208e-07, 'epoch': 0.02}

  1%|          | 96/13317 [00:59<2:15:10,  1.63it/s]
  1%|          | 97/13317 [01:00<2:15:06,  1.63it/s]
                                                    
{'loss': 0.7951, 'grad_norm': 2.2121434211730957, 'learning_rate': 7.282282282282282e-07, 'epoch': 0.02}

  1%|          | 97/13317 [01:00<2:15:06,  1.63it/s]
  1%|          | 98/13317 [01:00<2:15:04,  1.63it/s]
                                                    
{'loss': 0.8189, 'grad_norm': 2.240325927734375, 'learning_rate': 7.357357357357357e-07, 'epoch': 0.02}

  1%|          | 98/13317 [01:01<2:15:04,  1.63it/s]
  1%|          | 99/13317 [01:01<2:15:09,  1.63it/s]
                                                    
{'loss': 0.8679, 'grad_norm': 1.895703911781311, 'learning_rate': 7.432432432432434e-07, 'epoch': 0.02}

  1%|          | 99/13317 [01:01<2:15:09,  1.63it/s]
  1%|          | 100/13317 [01:02<2:15:13,  1.63it/s]
                                                     
{'loss': 0.8382, 'grad_norm': 1.9695020914077759, 'learning_rate': 7.507507507507509e-07, 'epoch': 0.02}

  1%|          | 100/13317 [01:02<2:15:13,  1.63it/s]
  1%|          | 101/13317 [01:02<2:15:17,  1.63it/s]
                                                     
{'loss': 0.7639, 'grad_norm': 1.8849538564682007, 'learning_rate': 7.582582582582583e-07, 'epoch': 0.02}

  1%|          | 101/13317 [01:02<2:15:17,  1.63it/s]
  1%|          | 102/13317 [01:03<2:15:17,  1.63it/s]
                                                     
{'loss': 0.8949, 'grad_norm': 1.8158146142959595, 'learning_rate': 7.657657657657658e-07, 'epoch': 0.02}

  1%|          | 102/13317 [01:03<2:15:17,  1.63it/s]
  1%|          | 103/13317 [01:04<2:15:17,  1.63it/s]
                                                     
{'loss': 0.7631, 'grad_norm': 1.7382808923721313, 'learning_rate': 7.732732732732734e-07, 'epoch': 0.02}

  1%|          | 103/13317 [01:04<2:15:17,  1.63it/s]
  1%|          | 104/13317 [01:04<2:15:17,  1.63it/s]
                                                     
{'loss': 0.7894, 'grad_norm': 1.8516268730163574, 'learning_rate': 7.807807807807808e-07, 'epoch': 0.02}

  1%|          | 104/13317 [01:04<2:15:17,  1.63it/s]
  1%|          | 105/13317 [01:05<2:15:20,  1.63it/s]
                                                     
{'loss': 0.8441, 'grad_norm': 1.9690122604370117, 'learning_rate': 7.882882882882883e-07, 'epoch': 0.02}

  1%|          | 105/13317 [01:05<2:15:20,  1.63it/s]
  1%|          | 106/13317 [01:05<2:15:14,  1.63it/s]
                                                     
{'loss': 0.838, 'grad_norm': 1.9234217405319214, 'learning_rate': 7.957957957957958e-07, 'epoch': 0.02}

  1%|          | 106/13317 [01:06<2:15:14,  1.63it/s]
  1%|          | 107/13317 [01:06<2:15:11,  1.63it/s]
                                                     
{'loss': 0.7771, 'grad_norm': 1.6662170886993408, 'learning_rate': 8.033033033033034e-07, 'epoch': 0.02}

  1%|          | 107/13317 [01:06<2:15:11,  1.63it/s]
  1%|          | 108/13317 [01:07<2:15:11,  1.63it/s]
                                                     
{'loss': 0.7269, 'grad_norm': 2.324514865875244, 'learning_rate': 8.108108108108109e-07, 'epoch': 0.02}

  1%|          | 108/13317 [01:07<2:15:11,  1.63it/s]
  1%|          | 109/13317 [01:07<2:15:07,  1.63it/s]
                                                     
{'loss': 0.9361, 'grad_norm': 2.0554516315460205, 'learning_rate': 8.183183183183184e-07, 'epoch': 0.02}

  1%|          | 109/13317 [01:07<2:15:07,  1.63it/s]
  1%|          | 110/13317 [01:08<2:15:12,  1.63it/s]
                                                     
{'loss': 0.8198, 'grad_norm': 1.7538983821868896, 'learning_rate': 8.258258258258259e-07, 'epoch': 0.02}

  1%|          | 110/13317 [01:08<2:15:12,  1.63it/s]
  1%|          | 111/13317 [01:08<2:15:09,  1.63it/s]
                                                     
{'loss': 0.7424, 'grad_norm': 1.5495929718017578, 'learning_rate': 8.333333333333333e-07, 'epoch': 0.03}

  1%|          | 111/13317 [01:09<2:15:09,  1.63it/s]
  1%|          | 112/13317 [01:09<2:15:01,  1.63it/s]
                                                     
{'loss': 0.8447, 'grad_norm': 1.9926998615264893, 'learning_rate': 8.40840840840841e-07, 'epoch': 0.03}

  1%|          | 112/13317 [01:09<2:15:01,  1.63it/s]
  1%|          | 113/13317 [01:10<2:14:58,  1.63it/s]
                                                     
{'loss': 0.7471, 'grad_norm': 1.607262372970581, 'learning_rate': 8.483483483483485e-07, 'epoch': 0.03}

  1%|          | 113/13317 [01:10<2:14:58,  1.63it/s]
  1%|          | 114/13317 [01:10<2:15:00,  1.63it/s]
                                                     
{'loss': 0.7899, 'grad_norm': 2.0737133026123047, 'learning_rate': 8.55855855855856e-07, 'epoch': 0.03}

  1%|          | 114/13317 [01:10<2:15:00,  1.63it/s]
  1%|          | 115/13317 [01:11<2:15:02,  1.63it/s]
                                                     
{'loss': 0.9086, 'grad_norm': 1.8293637037277222, 'learning_rate': 8.633633633633634e-07, 'epoch': 0.03}

  1%|          | 115/13317 [01:11<2:15:02,  1.63it/s]
  1%|          | 116/13317 [01:12<2:14:59,  1.63it/s]
                                                     
{'loss': 0.7126, 'grad_norm': 1.5163137912750244, 'learning_rate': 8.708708708708709e-07, 'epoch': 0.03}

  1%|          | 116/13317 [01:12<2:14:59,  1.63it/s]
  1%|          | 117/13317 [01:12<2:14:58,  1.63it/s]
                                                     
{'loss': 0.7203, 'grad_norm': 1.5233790874481201, 'learning_rate': 8.783783783783785e-07, 'epoch': 0.03}

  1%|          | 117/13317 [01:12<2:14:58,  1.63it/s]
  1%|          | 118/13317 [01:13<2:14:58,  1.63it/s]
                                                     
{'loss': 0.7637, 'grad_norm': 1.7197405099868774, 'learning_rate': 8.858858858858859e-07, 'epoch': 0.03}

  1%|          | 118/13317 [01:13<2:14:58,  1.63it/s]
  1%|          | 119/13317 [01:13<2:14:52,  1.63it/s]
                                                     
{'loss': 0.8108, 'grad_norm': 1.9985202550888062, 'learning_rate': 8.933933933933935e-07, 'epoch': 0.03}

  1%|          | 119/13317 [01:13<2:14:52,  1.63it/s]
  1%|          | 120/13317 [01:14<2:15:01,  1.63it/s]
                                                     
{'loss': 0.6914, 'grad_norm': 1.7084733247756958, 'learning_rate': 9.00900900900901e-07, 'epoch': 0.03}

  1%|          | 120/13317 [01:14<2:15:01,  1.63it/s]
  1%|          | 121/13317 [01:15<2:14:58,  1.63it/s]
                                                     
{'loss': 0.7111, 'grad_norm': 1.5084526538848877, 'learning_rate': 9.084084084084085e-07, 'epoch': 0.03}

  1%|          | 121/13317 [01:15<2:14:58,  1.63it/s]
  1%|          | 122/13317 [01:15<2:14:54,  1.63it/s]
                                                     
{'loss': 0.7237, 'grad_norm': 1.5757193565368652, 'learning_rate': 9.15915915915916e-07, 'epoch': 0.03}

  1%|          | 122/13317 [01:15<2:14:54,  1.63it/s]
  1%|          | 123/13317 [01:16<2:14:56,  1.63it/s]
                                                     
{'loss': 0.7567, 'grad_norm': 1.6613407135009766, 'learning_rate': 9.234234234234235e-07, 'epoch': 0.03}

  1%|          | 123/13317 [01:16<2:14:56,  1.63it/s]
  1%|          | 124/13317 [01:16<2:14:55,  1.63it/s]
                                                     
{'loss': 0.8299, 'grad_norm': 1.7441282272338867, 'learning_rate': 9.30930930930931e-07, 'epoch': 0.03}

  1%|          | 124/13317 [01:17<2:14:55,  1.63it/s]
  1%|          | 125/13317 [01:17<2:14:58,  1.63it/s]
                                                     
{'loss': 0.784, 'grad_norm': 1.7721879482269287, 'learning_rate': 9.384384384384384e-07, 'epoch': 0.03}

  1%|          | 125/13317 [01:17<2:14:58,  1.63it/s]
  1%|          | 126/13317 [01:18<2:15:33,  1.62it/s]
                                                     
{'loss': 0.6477, 'grad_norm': 1.5495719909667969, 'learning_rate': 9.459459459459461e-07, 'epoch': 0.03}

  1%|          | 126/13317 [01:18<2:15:33,  1.62it/s]
  1%|          | 127/13317 [01:18<2:15:20,  1.62it/s]
                                                     
{'loss': 0.753, 'grad_norm': 1.6460978984832764, 'learning_rate': 9.534534534534536e-07, 'epoch': 0.03}

  1%|          | 127/13317 [01:18<2:15:20,  1.62it/s]
  1%|          | 128/13317 [01:19<2:15:07,  1.63it/s]
                                                     
{'loss': 0.6963, 'grad_norm': 1.657727837562561, 'learning_rate': 9.60960960960961e-07, 'epoch': 0.03}

  1%|          | 128/13317 [01:19<2:15:07,  1.63it/s]
  1%|          | 129/13317 [01:20<2:15:05,  1.63it/s]
                                                     
{'loss': 0.7287, 'grad_norm': 1.6593786478042603, 'learning_rate': 9.684684684684686e-07, 'epoch': 0.03}

  1%|          | 129/13317 [01:20<2:15:05,  1.63it/s]
  1%|          | 130/13317 [01:20<2:15:07,  1.63it/s]
                                                     
{'loss': 0.7579, 'grad_norm': 1.6817530393600464, 'learning_rate': 9.75975975975976e-07, 'epoch': 0.03}

  1%|          | 130/13317 [01:20<2:15:07,  1.63it/s]
  1%|          | 131/13317 [01:21<2:14:56,  1.63it/s]
                                                     
{'loss': 0.6719, 'grad_norm': 1.4413070678710938, 'learning_rate': 9.834834834834835e-07, 'epoch': 0.03}

  1%|          | 131/13317 [01:21<2:14:56,  1.63it/s]
  1%|          | 132/13317 [01:21<2:14:53,  1.63it/s]
                                                     
{'loss': 0.7026, 'grad_norm': 2.2124626636505127, 'learning_rate': 9.909909909909911e-07, 'epoch': 0.03}

  1%|          | 132/13317 [01:21<2:14:53,  1.63it/s]
  1%|          | 133/13317 [01:22<2:14:51,  1.63it/s]
                                                     
{'loss': 0.5593, 'grad_norm': 1.3846427202224731, 'learning_rate': 9.984984984984985e-07, 'epoch': 0.03}

  1%|          | 133/13317 [01:22<2:14:51,  1.63it/s]
  1%|          | 134/13317 [01:23<2:14:54,  1.63it/s]
                                                     
{'loss': 0.647, 'grad_norm': 1.4754674434661865, 'learning_rate': 1.006006006006006e-06, 'epoch': 0.03}

  1%|          | 134/13317 [01:23<2:14:54,  1.63it/s]
  1%|          | 135/13317 [01:23<2:15:02,  1.63it/s]
                                                     
{'loss': 0.6921, 'grad_norm': 1.5349516868591309, 'learning_rate': 1.0135135135135136e-06, 'epoch': 0.03}

  1%|          | 135/13317 [01:23<2:15:02,  1.63it/s]
  1%|          | 136/13317 [01:24<2:14:53,  1.63it/s]
                                                     
{'loss': 0.7064, 'grad_norm': 1.4412708282470703, 'learning_rate': 1.0210210210210212e-06, 'epoch': 0.03}

  1%|          | 136/13317 [01:24<2:14:53,  1.63it/s]
  1%|          | 137/13317 [01:24<2:14:53,  1.63it/s]
                                                     
{'loss': 0.706, 'grad_norm': 1.5820810794830322, 'learning_rate': 1.0285285285285286e-06, 'epoch': 0.03}

  1%|          | 137/13317 [01:25<2:14:53,  1.63it/s]
  1%|          | 138/13317 [01:25<2:14:46,  1.63it/s]
                                                     
{'loss': 0.7606, 'grad_norm': 1.5930715799331665, 'learning_rate': 1.0360360360360361e-06, 'epoch': 0.03}

  1%|          | 138/13317 [01:25<2:14:46,  1.63it/s]
  1%|          | 139/13317 [01:26<2:14:46,  1.63it/s]
                                                     
{'loss': 0.6457, 'grad_norm': 1.452947735786438, 'learning_rate': 1.0435435435435435e-06, 'epoch': 0.03}

  1%|          | 139/13317 [01:26<2:14:46,  1.63it/s]
  1%|          | 140/13317 [01:26<2:15:02,  1.63it/s]
                                                     
{'loss': 0.5897, 'grad_norm': 1.244608998298645, 'learning_rate': 1.051051051051051e-06, 'epoch': 0.03}

  1%|          | 140/13317 [01:26<2:15:02,  1.63it/s]
  1%|          | 141/13317 [01:27<2:14:55,  1.63it/s]
                                                     
{'loss': 0.7207, 'grad_norm': 1.452004075050354, 'learning_rate': 1.0585585585585587e-06, 'epoch': 0.03}

  1%|          | 141/13317 [01:27<2:14:55,  1.63it/s]
  1%|          | 142/13317 [01:27<2:14:50,  1.63it/s]
                                                     
{'loss': 0.6814, 'grad_norm': 1.611354112625122, 'learning_rate': 1.0660660660660662e-06, 'epoch': 0.03}

  1%|          | 142/13317 [01:28<2:14:50,  1.63it/s]
  1%|          | 143/13317 [01:28<2:14:50,  1.63it/s]
                                                     
{'loss': 0.7198, 'grad_norm': 1.4763097763061523, 'learning_rate': 1.0735735735735736e-06, 'epoch': 0.03}

  1%|          | 143/13317 [01:28<2:14:50,  1.63it/s]
  1%|          | 144/13317 [01:29<2:14:46,  1.63it/s]
                                                     
{'loss': 0.582, 'grad_norm': 1.4349634647369385, 'learning_rate': 1.0810810810810812e-06, 'epoch': 0.03}

  1%|          | 144/13317 [01:29<2:14:46,  1.63it/s]
  1%|          | 145/13317 [01:29<2:14:57,  1.63it/s]
                                                     
{'loss': 0.6567, 'grad_norm': 1.3705693483352661, 'learning_rate': 1.0885885885885887e-06, 'epoch': 0.03}

  1%|          | 145/13317 [01:29<2:14:57,  1.63it/s]
  1%|          | 146/13317 [01:30<2:14:50,  1.63it/s]
                                                     
{'loss': 0.636, 'grad_norm': 1.373970627784729, 'learning_rate': 1.096096096096096e-06, 'epoch': 0.03}

  1%|          | 146/13317 [01:30<2:14:50,  1.63it/s]
  1%|          | 147/13317 [01:31<2:14:45,  1.63it/s]
                                                     
{'loss': 0.6333, 'grad_norm': 1.4731109142303467, 'learning_rate': 1.1036036036036037e-06, 'epoch': 0.03}

  1%|          | 147/13317 [01:31<2:14:45,  1.63it/s]
  1%|          | 148/13317 [01:31<2:14:40,  1.63it/s]
                                                     
{'loss': 0.7008, 'grad_norm': 1.4753167629241943, 'learning_rate': 1.111111111111111e-06, 'epoch': 0.03}

  1%|          | 148/13317 [01:31<2:14:40,  1.63it/s]
  1%|          | 149/13317 [01:32<2:14:42,  1.63it/s]
                                                     
{'loss': 0.7357, 'grad_norm': 1.5240771770477295, 'learning_rate': 1.1186186186186188e-06, 'epoch': 0.03}

  1%|          | 149/13317 [01:32<2:14:42,  1.63it/s]
  1%|          | 150/13317 [01:32<2:14:45,  1.63it/s]
                                                     
{'loss': 0.6652, 'grad_norm': 1.3055368661880493, 'learning_rate': 1.1261261261261262e-06, 'epoch': 0.03}

  1%|          | 150/13317 [01:33<2:14:45,  1.63it/s]
  1%|          | 151/13317 [01:33<2:14:39,  1.63it/s]
                                                     
{'loss': 0.5688, 'grad_norm': 1.1227576732635498, 'learning_rate': 1.1336336336336338e-06, 'epoch': 0.03}

  1%|          | 151/13317 [01:33<2:14:39,  1.63it/s]
  1%|          | 152/13317 [01:34<2:14:40,  1.63it/s]
                                                     
{'loss': 0.691, 'grad_norm': 1.3222991228103638, 'learning_rate': 1.1411411411411411e-06, 'epoch': 0.03}

  1%|          | 152/13317 [01:34<2:14:40,  1.63it/s]
  1%|          | 153/13317 [01:34<2:14:33,  1.63it/s]
                                                     
{'loss': 0.6524, 'grad_norm': 1.4769599437713623, 'learning_rate': 1.148648648648649e-06, 'epoch': 0.03}

  1%|          | 153/13317 [01:34<2:14:33,  1.63it/s]
  1%|          | 154/13317 [01:35<2:14:33,  1.63it/s]
                                                     
{'loss': 0.6384, 'grad_norm': 1.217717170715332, 'learning_rate': 1.1561561561561563e-06, 'epoch': 0.03}

  1%|          | 154/13317 [01:35<2:14:33,  1.63it/s]
  1%|          | 155/13317 [01:35<2:14:42,  1.63it/s]
                                                     
{'loss': 0.7631, 'grad_norm': 1.3133342266082764, 'learning_rate': 1.1636636636636638e-06, 'epoch': 0.03}

  1%|          | 155/13317 [01:36<2:14:42,  1.63it/s]
  1%|          | 156/13317 [01:36<2:14:39,  1.63it/s]
                                                     
{'loss': 0.6493, 'grad_norm': 1.3204808235168457, 'learning_rate': 1.1711711711711712e-06, 'epoch': 0.04}

  1%|          | 156/13317 [01:36<2:14:39,  1.63it/s]
  1%|          | 157/13317 [01:37<2:14:37,  1.63it/s]
                                                     
{'loss': 0.4938, 'grad_norm': 1.0120199918746948, 'learning_rate': 1.1786786786786788e-06, 'epoch': 0.04}

  1%|          | 157/13317 [01:37<2:14:37,  1.63it/s]
  1%|          | 158/13317 [01:37<2:14:39,  1.63it/s]
                                                     
{'loss': 0.4971, 'grad_norm': 0.848904013633728, 'learning_rate': 1.1861861861861864e-06, 'epoch': 0.04}

  1%|          | 158/13317 [01:37<2:14:39,  1.63it/s]
  1%|          | 159/13317 [01:38<2:14:34,  1.63it/s]
                                                     
{'loss': 0.5508, 'grad_norm': 1.0298947095870972, 'learning_rate': 1.1936936936936937e-06, 'epoch': 0.04}

  1%|          | 159/13317 [01:38<2:14:34,  1.63it/s]
  1%|          | 160/13317 [01:39<2:14:41,  1.63it/s]
                                                     
{'loss': 0.5656, 'grad_norm': 1.0605055093765259, 'learning_rate': 1.2012012012012013e-06, 'epoch': 0.04}

  1%|          | 160/13317 [01:39<2:14:41,  1.63it/s]
  1%|          | 161/13317 [01:39<2:14:43,  1.63it/s]
                                                     
{'loss': 0.5773, 'grad_norm': 1.1684128046035767, 'learning_rate': 1.2087087087087089e-06, 'epoch': 0.04}

  1%|          | 161/13317 [01:39<2:14:43,  1.63it/s]
  1%|          | 162/13317 [01:40<2:14:42,  1.63it/s]
                                                     
{'loss': 0.6469, 'grad_norm': 1.3407416343688965, 'learning_rate': 1.2162162162162164e-06, 'epoch': 0.04}

  1%|          | 162/13317 [01:40<2:14:42,  1.63it/s]
  1%|          | 163/13317 [01:40<2:14:38,  1.63it/s]
                                                     
{'loss': 0.5085, 'grad_norm': 0.8725384473800659, 'learning_rate': 1.2237237237237238e-06, 'epoch': 0.04}

  1%|          | 163/13317 [01:41<2:14:38,  1.63it/s]
  1%|          | 164/13317 [01:41<2:14:33,  1.63it/s]
                                                     
{'loss': 0.6691, 'grad_norm': 1.4522643089294434, 'learning_rate': 1.2312312312312314e-06, 'epoch': 0.04}

  1%|          | 164/13317 [01:41<2:14:33,  1.63it/s]
  1%|          | 165/13317 [01:42<2:14:54,  1.62it/s]
                                                     
{'loss': 0.5529, 'grad_norm': 1.0491087436676025, 'learning_rate': 1.2387387387387387e-06, 'epoch': 0.04}

  1%|          | 165/13317 [01:42<2:14:54,  1.62it/s]
  1%|          | 166/13317 [01:42<2:14:42,  1.63it/s]
                                                     
{'loss': 0.553, 'grad_norm': 1.1312469244003296, 'learning_rate': 1.2462462462462463e-06, 'epoch': 0.04}

  1%|          | 166/13317 [01:42<2:14:42,  1.63it/s]
  1%|▏         | 167/13317 [01:43<2:14:36,  1.63it/s]
                                                     
{'loss': 0.5845, 'grad_norm': 1.0458314418792725, 'learning_rate': 1.2537537537537539e-06, 'epoch': 0.04}

  1%|▏         | 167/13317 [01:43<2:14:36,  1.63it/s]
  1%|▏         | 168/13317 [01:43<2:14:32,  1.63it/s]
                                                     
{'loss': 0.621, 'grad_norm': 1.118099331855774, 'learning_rate': 1.2612612612612613e-06, 'epoch': 0.04}

  1%|▏         | 168/13317 [01:44<2:14:32,  1.63it/s]
  1%|▏         | 169/13317 [01:44<2:14:25,  1.63it/s]
                                                     
{'loss': 0.6034, 'grad_norm': 1.1403764486312866, 'learning_rate': 1.2687687687687688e-06, 'epoch': 0.04}

  1%|▏         | 169/13317 [01:44<2:14:25,  1.63it/s]
  1%|▏         | 170/13317 [01:45<2:15:06,  1.62it/s]
                                                     
{'loss': 0.5307, 'grad_norm': 0.9272016286849976, 'learning_rate': 1.2762762762762764e-06, 'epoch': 0.04}

  1%|▏         | 170/13317 [01:45<2:15:06,  1.62it/s]
  1%|▏         | 171/13317 [01:45<2:14:52,  1.62it/s]
                                                     
{'loss': 0.5721, 'grad_norm': 1.065805435180664, 'learning_rate': 1.2837837837837838e-06, 'epoch': 0.04}

  1%|▏         | 171/13317 [01:45<2:14:52,  1.62it/s]
  1%|▏         | 172/13317 [01:46<2:14:42,  1.63it/s]
                                                     
{'loss': 0.5909, 'grad_norm': 1.014577865600586, 'learning_rate': 1.2912912912912913e-06, 'epoch': 0.04}

  1%|▏         | 172/13317 [01:46<2:14:42,  1.63it/s]
  1%|▏         | 173/13317 [01:47<2:14:35,  1.63it/s]
                                                     
{'loss': 0.573, 'grad_norm': 1.0064231157302856, 'learning_rate': 1.2987987987987987e-06, 'epoch': 0.04}

  1%|▏         | 173/13317 [01:47<2:14:35,  1.63it/s]
  1%|▏         | 174/13317 [01:47<2:14:34,  1.63it/s]
                                                     
{'loss': 0.5824, 'grad_norm': 1.1200244426727295, 'learning_rate': 1.3063063063063065e-06, 'epoch': 0.04}

  1%|▏         | 174/13317 [01:47<2:14:34,  1.63it/s]
  1%|▏         | 175/13317 [01:48<2:14:40,  1.63it/s]
                                                     
{'loss': 0.4886, 'grad_norm': 0.9291266798973083, 'learning_rate': 1.313813813813814e-06, 'epoch': 0.04}

  1%|▏         | 175/13317 [01:48<2:14:40,  1.63it/s]
  1%|▏         | 176/13317 [01:48<2:14:32,  1.63it/s]
                                                     
{'loss': 0.4236, 'grad_norm': 0.7707609534263611, 'learning_rate': 1.3213213213213214e-06, 'epoch': 0.04}

  1%|▏         | 176/13317 [01:49<2:14:32,  1.63it/s]
  1%|▏         | 177/13317 [01:49<2:14:27,  1.63it/s]
                                                     
{'loss': 0.5926, 'grad_norm': 1.1367814540863037, 'learning_rate': 1.328828828828829e-06, 'epoch': 0.04}

  1%|▏         | 177/13317 [01:49<2:14:27,  1.63it/s]
  1%|▏         | 178/13317 [01:50<2:14:27,  1.63it/s]
                                                     
{'loss': 0.6172, 'grad_norm': 1.0455306768417358, 'learning_rate': 1.3363363363363366e-06, 'epoch': 0.04}

  1%|▏         | 178/13317 [01:50<2:14:27,  1.63it/s]
  1%|▏         | 179/13317 [01:50<2:14:25,  1.63it/s]
                                                     
{'loss': 0.5864, 'grad_norm': 1.0398110151290894, 'learning_rate': 1.343843843843844e-06, 'epoch': 0.04}

  1%|▏         | 179/13317 [01:50<2:14:25,  1.63it/s]
  1%|▏         | 180/13317 [01:51<2:14:28,  1.63it/s]
                                                     
{'loss': 0.5299, 'grad_norm': 1.0818138122558594, 'learning_rate': 1.3513513513513515e-06, 'epoch': 0.04}

  1%|▏         | 180/13317 [01:51<2:14:28,  1.63it/s]
  1%|▏         | 181/13317 [01:51<2:14:23,  1.63it/s]
                                                     
{'loss': 0.6409, 'grad_norm': 1.0900903940200806, 'learning_rate': 1.3588588588588589e-06, 'epoch': 0.04}

  1%|▏         | 181/13317 [01:52<2:14:23,  1.63it/s]
  1%|▏         | 182/13317 [01:52<2:14:24,  1.63it/s]
                                                     
{'loss': 0.4352, 'grad_norm': 0.8942669034004211, 'learning_rate': 1.3663663663663664e-06, 'epoch': 0.04}

  1%|▏         | 182/13317 [01:52<2:14:24,  1.63it/s]
  1%|▏         | 183/13317 [01:53<2:14:24,  1.63it/s]
                                                     
{'loss': 0.6471, 'grad_norm': 1.1304272413253784, 'learning_rate': 1.373873873873874e-06, 'epoch': 0.04}

  1%|▏         | 183/13317 [01:53<2:14:24,  1.63it/s]
  1%|▏         | 184/13317 [01:53<2:14:18,  1.63it/s]
                                                     
{'loss': 0.4891, 'grad_norm': 0.8067008256912231, 'learning_rate': 1.3813813813813814e-06, 'epoch': 0.04}

  1%|▏         | 184/13317 [01:53<2:14:18,  1.63it/s]
  1%|▏         | 185/13317 [01:54<2:14:25,  1.63it/s]
                                                     
{'loss': 0.5565, 'grad_norm': 0.9788805246353149, 'learning_rate': 1.3888888888888892e-06, 'epoch': 0.04}

  1%|▏         | 185/13317 [01:54<2:14:25,  1.63it/s]
  1%|▏         | 186/13317 [01:55<2:14:24,  1.63it/s]
                                                     
{'loss': 0.5082, 'grad_norm': 1.0063388347625732, 'learning_rate': 1.3963963963963963e-06, 'epoch': 0.04}

  1%|▏         | 186/13317 [01:55<2:14:24,  1.63it/s]
  1%|▏         | 187/13317 [01:55<2:14:21,  1.63it/s]
                                                     
{'loss': 0.6214, 'grad_norm': 1.0533509254455566, 'learning_rate': 1.403903903903904e-06, 'epoch': 0.04}

  1%|▏         | 187/13317 [01:55<2:14:21,  1.63it/s]
  1%|▏         | 188/13317 [01:56<2:14:19,  1.63it/s]
                                                     
{'loss': 0.5873, 'grad_norm': 1.0716609954833984, 'learning_rate': 1.4114114114114117e-06, 'epoch': 0.04}

  1%|▏         | 188/13317 [01:56<2:14:19,  1.63it/s]
  1%|▏         | 189/13317 [01:56<2:14:20,  1.63it/s]
                                                     
{'loss': 0.6015, 'grad_norm': 0.918181836605072, 'learning_rate': 1.418918918918919e-06, 'epoch': 0.04}

  1%|▏         | 189/13317 [01:56<2:14:20,  1.63it/s]
  1%|▏         | 190/13317 [01:57<2:14:22,  1.63it/s]
                                                     
{'loss': 0.4706, 'grad_norm': 0.8621892333030701, 'learning_rate': 1.4264264264264266e-06, 'epoch': 0.04}

  1%|▏         | 190/13317 [01:57<2:14:22,  1.63it/s]
  1%|▏         | 191/13317 [01:58<2:14:20,  1.63it/s]
                                                     
{'loss': 0.6133, 'grad_norm': 1.162561297416687, 'learning_rate': 1.433933933933934e-06, 'epoch': 0.04}

  1%|▏         | 191/13317 [01:58<2:14:20,  1.63it/s]
  1%|▏         | 192/13317 [01:58<2:14:15,  1.63it/s]
                                                     
{'loss': 0.6365, 'grad_norm': 1.1295325756072998, 'learning_rate': 1.4414414414414416e-06, 'epoch': 0.04}

  1%|▏         | 192/13317 [01:58<2:14:15,  1.63it/s]
  1%|▏         | 193/13317 [01:59<2:14:14,  1.63it/s]
                                                     
{'loss': 0.5639, 'grad_norm': 1.0269994735717773, 'learning_rate': 1.4489489489489491e-06, 'epoch': 0.04}

  1%|▏         | 193/13317 [01:59<2:14:14,  1.63it/s]
  1%|▏         | 194/13317 [01:59<2:14:14,  1.63it/s]
                                                     
{'loss': 0.5096, 'grad_norm': 1.001244068145752, 'learning_rate': 1.4564564564564565e-06, 'epoch': 0.04}

  1%|▏         | 194/13317 [02:00<2:14:14,  1.63it/s]
  1%|▏         | 195/13317 [02:00<2:14:18,  1.63it/s]
                                                     
{'loss': 0.5165, 'grad_norm': 1.137724757194519, 'learning_rate': 1.463963963963964e-06, 'epoch': 0.04}

  1%|▏         | 195/13317 [02:00<2:14:18,  1.63it/s]
  1%|▏         | 196/13317 [02:01<2:14:16,  1.63it/s]
                                                     
{'loss': 0.5814, 'grad_norm': 0.8230946660041809, 'learning_rate': 1.4714714714714714e-06, 'epoch': 0.04}

  1%|▏         | 196/13317 [02:01<2:14:16,  1.63it/s]
  1%|▏         | 197/13317 [02:01<2:14:11,  1.63it/s]
                                                     
{'loss': 0.5278, 'grad_norm': 0.7805623412132263, 'learning_rate': 1.478978978978979e-06, 'epoch': 0.04}

  1%|▏         | 197/13317 [02:01<2:14:11,  1.63it/s]
  1%|▏         | 198/13317 [02:02<2:14:09,  1.63it/s]
                                                     
{'loss': 0.5008, 'grad_norm': 0.9030367732048035, 'learning_rate': 1.4864864864864868e-06, 'epoch': 0.04}

  1%|▏         | 198/13317 [02:02<2:14:09,  1.63it/s]
  1%|▏         | 199/13317 [02:02<2:14:16,  1.63it/s]
                                                     
{'loss': 0.5205, 'grad_norm': 0.9333950281143188, 'learning_rate': 1.493993993993994e-06, 'epoch': 0.04}

  1%|▏         | 199/13317 [02:03<2:14:16,  1.63it/s]
  2%|▏         | 200/13317 [02:03<2:14:22,  1.63it/s]
                                                     
{'loss': 0.5589, 'grad_norm': 0.9850483536720276, 'learning_rate': 1.5015015015015017e-06, 'epoch': 0.05}

  2%|▏         | 200/13317 [02:03<2:14:22,  1.63it/s]
  2%|▏         | 201/13317 [02:04<2:14:22,  1.63it/s]
                                                     
{'loss': 0.4891, 'grad_norm': 0.9021425247192383, 'learning_rate': 1.5090090090090093e-06, 'epoch': 0.05}

  2%|▏         | 201/13317 [02:04<2:14:22,  1.63it/s]
  2%|▏         | 202/13317 [02:04<2:14:36,  1.62it/s]
                                                     
{'loss': 0.534, 'grad_norm': 1.1325154304504395, 'learning_rate': 1.5165165165165167e-06, 'epoch': 0.05}

  2%|▏         | 202/13317 [02:04<2:14:36,  1.62it/s]
  2%|▏         | 203/13317 [02:05<2:14:18,  1.63it/s]
                                                     
{'loss': 0.6164, 'grad_norm': 1.1065471172332764, 'learning_rate': 1.5240240240240242e-06, 'epoch': 0.05}

  2%|▏         | 203/13317 [02:05<2:14:18,  1.63it/s]
  2%|▏         | 204/13317 [02:06<2:14:32,  1.62it/s]
                                                     
{'loss': 0.5049, 'grad_norm': 1.1567038297653198, 'learning_rate': 1.5315315315315316e-06, 'epoch': 0.05}

  2%|▏         | 204/13317 [02:06<2:14:32,  1.62it/s]
  2%|▏         | 205/13317 [02:06<2:14:35,  1.62it/s]
                                                     
{'loss': 0.6127, 'grad_norm': 0.8912550806999207, 'learning_rate': 1.5390390390390392e-06, 'epoch': 0.05}

  2%|▏         | 205/13317 [02:06<2:14:35,  1.62it/s]
  2%|▏         | 206/13317 [02:07<2:14:27,  1.63it/s]
                                                     
{'loss': 0.5048, 'grad_norm': 0.9775712490081787, 'learning_rate': 1.5465465465465467e-06, 'epoch': 0.05}

  2%|▏         | 206/13317 [02:07<2:14:27,  1.63it/s]
  2%|▏         | 207/13317 [02:07<2:14:22,  1.63it/s]
                                                     
{'loss': 0.6139, 'grad_norm': 1.0000442266464233, 'learning_rate': 1.5540540540540541e-06, 'epoch': 0.05}

  2%|▏         | 207/13317 [02:08<2:14:22,  1.63it/s]
  2%|▏         | 208/13317 [02:08<2:14:15,  1.63it/s]
                                                     
{'loss': 0.5767, 'grad_norm': 0.9816790223121643, 'learning_rate': 1.5615615615615617e-06, 'epoch': 0.05}

  2%|▏         | 208/13317 [02:08<2:14:15,  1.63it/s]
  2%|▏         | 209/13317 [02:09<2:14:06,  1.63it/s]
                                                     
{'loss': 0.5635, 'grad_norm': 0.9469976425170898, 'learning_rate': 1.569069069069069e-06, 'epoch': 0.05}

  2%|▏         | 209/13317 [02:09<2:14:06,  1.63it/s]
  2%|▏         | 210/13317 [02:09<2:14:15,  1.63it/s]
                                                     
{'loss': 0.5211, 'grad_norm': 1.0825903415679932, 'learning_rate': 1.5765765765765766e-06, 'epoch': 0.05}

  2%|▏         | 210/13317 [02:09<2:14:15,  1.63it/s]
  2%|▏         | 211/13317 [02:10<2:14:16,  1.63it/s]
                                                     
{'loss': 0.5048, 'grad_norm': 0.8686360716819763, 'learning_rate': 1.5840840840840844e-06, 'epoch': 0.05}

  2%|▏         | 211/13317 [02:10<2:14:16,  1.63it/s]
  2%|▏         | 212/13317 [02:10<2:14:10,  1.63it/s]
                                                     
{'loss': 0.6054, 'grad_norm': 1.098149061203003, 'learning_rate': 1.5915915915915916e-06, 'epoch': 0.05}

  2%|▏         | 212/13317 [02:11<2:14:10,  1.63it/s]
  2%|▏         | 213/13317 [02:11<2:14:05,  1.63it/s]
                                                     
{'loss': 0.5983, 'grad_norm': 0.9329115152359009, 'learning_rate': 1.5990990990990993e-06, 'epoch': 0.05}

  2%|▏         | 213/13317 [02:11<2:14:05,  1.63it/s]
  2%|▏         | 214/13317 [02:12<2:14:07,  1.63it/s]
                                                     
{'loss': 0.5157, 'grad_norm': 0.8366087079048157, 'learning_rate': 1.6066066066066067e-06, 'epoch': 0.05}

  2%|▏         | 214/13317 [02:12<2:14:07,  1.63it/s]
  2%|▏         | 215/13317 [02:12<2:14:04,  1.63it/s]
                                                     
{'loss': 0.455, 'grad_norm': 0.8130690455436707, 'learning_rate': 1.6141141141141143e-06, 'epoch': 0.05}

  2%|▏         | 215/13317 [02:12<2:14:04,  1.63it/s]
  2%|▏         | 216/13317 [02:13<2:14:01,  1.63it/s]
                                                     
{'loss': 0.5575, 'grad_norm': 0.945459246635437, 'learning_rate': 1.6216216216216219e-06, 'epoch': 0.05}

  2%|▏         | 216/13317 [02:13<2:14:01,  1.63it/s]
  2%|▏         | 217/13317 [02:14<2:14:01,  1.63it/s]
                                                     
{'loss': 0.5237, 'grad_norm': 0.9335184097290039, 'learning_rate': 1.6291291291291292e-06, 'epoch': 0.05}

  2%|▏         | 217/13317 [02:14<2:14:01,  1.63it/s]
  2%|▏         | 218/13317 [02:14<2:13:56,  1.63it/s]
                                                     
{'loss': 0.542, 'grad_norm': 0.8258176445960999, 'learning_rate': 1.6366366366366368e-06, 'epoch': 0.05}

  2%|▏         | 218/13317 [02:14<2:13:56,  1.63it/s]
  2%|▏         | 219/13317 [02:15<2:13:54,  1.63it/s]
                                                     
{'loss': 0.6317, 'grad_norm': 1.1205437183380127, 'learning_rate': 1.6441441441441444e-06, 'epoch': 0.05}

  2%|▏         | 219/13317 [02:15<2:13:54,  1.63it/s]
  2%|▏         | 220/13317 [02:15<2:13:55,  1.63it/s]
                                                     
{'loss': 0.513, 'grad_norm': 1.0060783624649048, 'learning_rate': 1.6516516516516517e-06, 'epoch': 0.05}

  2%|▏         | 220/13317 [02:16<2:13:55,  1.63it/s]
  2%|▏         | 221/13317 [02:16<2:13:51,  1.63it/s]
                                                     
{'loss': 0.4756, 'grad_norm': 0.8755509257316589, 'learning_rate': 1.6591591591591593e-06, 'epoch': 0.05}

  2%|▏         | 221/13317 [02:16<2:13:51,  1.63it/s]
  2%|▏         | 222/13317 [02:17<2:13:51,  1.63it/s]
                                                     
{'loss': 0.4336, 'grad_norm': 0.7452938556671143, 'learning_rate': 1.6666666666666667e-06, 'epoch': 0.05}

  2%|▏         | 222/13317 [02:17<2:13:51,  1.63it/s]
  2%|▏         | 223/13317 [02:17<2:13:48,  1.63it/s]
                                                     
{'loss': 0.6685, 'grad_norm': 0.8382483720779419, 'learning_rate': 1.6741741741741742e-06, 'epoch': 0.05}

  2%|▏         | 223/13317 [02:17<2:13:48,  1.63it/s]
  2%|▏         | 224/13317 [02:18<2:13:48,  1.63it/s]
                                                     
{'loss': 0.4601, 'grad_norm': 0.8271287083625793, 'learning_rate': 1.681681681681682e-06, 'epoch': 0.05}

  2%|▏         | 224/13317 [02:18<2:13:48,  1.63it/s]
  2%|▏         | 225/13317 [02:18<2:13:51,  1.63it/s]
                                                     
{'loss': 0.5016, 'grad_norm': 0.8500949144363403, 'learning_rate': 1.6891891891891894e-06, 'epoch': 0.05}

  2%|▏         | 225/13317 [02:19<2:13:51,  1.63it/s]
  2%|▏         | 226/13317 [02:19<2:13:49,  1.63it/s]
                                                     
{'loss': 0.4385, 'grad_norm': 0.7964163422584534, 'learning_rate': 1.696696696696697e-06, 'epoch': 0.05}

  2%|▏         | 226/13317 [02:19<2:13:49,  1.63it/s]
  2%|▏         | 227/13317 [02:20<2:13:50,  1.63it/s]
                                                     
{'loss': 0.3599, 'grad_norm': 0.7702478766441345, 'learning_rate': 1.7042042042042043e-06, 'epoch': 0.05}

  2%|▏         | 227/13317 [02:20<2:13:50,  1.63it/s]
  2%|▏         | 228/13317 [02:20<2:13:50,  1.63it/s]
                                                     
{'loss': 0.4629, 'grad_norm': 0.7608579993247986, 'learning_rate': 1.711711711711712e-06, 'epoch': 0.05}

  2%|▏         | 228/13317 [02:20<2:13:50,  1.63it/s]
  2%|▏         | 229/13317 [02:21<2:13:54,  1.63it/s]
                                                     
{'loss': 0.5383, 'grad_norm': 0.899407684803009, 'learning_rate': 1.7192192192192195e-06, 'epoch': 0.05}

  2%|▏         | 229/13317 [02:21<2:13:54,  1.63it/s]
  2%|▏         | 230/13317 [02:22<2:13:48,  1.63it/s]
                                                     
{'loss': 0.5725, 'grad_norm': 0.8703835010528564, 'learning_rate': 1.7267267267267268e-06, 'epoch': 0.05}

  2%|▏         | 230/13317 [02:22<2:13:48,  1.63it/s]
  2%|▏         | 231/13317 [02:22<2:13:49,  1.63it/s]
                                                     
{'loss': 0.5911, 'grad_norm': 0.8942667245864868, 'learning_rate': 1.7342342342342344e-06, 'epoch': 0.05}

  2%|▏         | 231/13317 [02:22<2:13:49,  1.63it/s]
  2%|▏         | 232/13317 [02:23<2:13:50,  1.63it/s]
                                                     
{'loss': 0.468, 'grad_norm': 0.8283909559249878, 'learning_rate': 1.7417417417417418e-06, 'epoch': 0.05}

  2%|▏         | 232/13317 [02:23<2:13:50,  1.63it/s]
  2%|▏         | 233/13317 [02:23<2:13:47,  1.63it/s]
                                                     
{'loss': 0.4996, 'grad_norm': 0.8855001926422119, 'learning_rate': 1.7492492492492493e-06, 'epoch': 0.05}

  2%|▏         | 233/13317 [02:24<2:13:47,  1.63it/s]
  2%|▏         | 234/13317 [02:24<2:13:46,  1.63it/s]
                                                     
{'loss': 0.7024, 'grad_norm': 0.9389349222183228, 'learning_rate': 1.756756756756757e-06, 'epoch': 0.05}

  2%|▏         | 234/13317 [02:24<2:13:46,  1.63it/s]
  2%|▏         | 235/13317 [02:25<2:13:48,  1.63it/s]
                                                     
{'loss': 0.4885, 'grad_norm': 0.8391992449760437, 'learning_rate': 1.7642642642642643e-06, 'epoch': 0.05}

  2%|▏         | 235/13317 [02:25<2:13:48,  1.63it/s]
  2%|▏         | 236/13317 [02:25<2:13:45,  1.63it/s]
                                                     
{'loss': 0.4532, 'grad_norm': 0.9417365789413452, 'learning_rate': 1.7717717717717719e-06, 'epoch': 0.05}

  2%|▏         | 236/13317 [02:25<2:13:45,  1.63it/s]
  2%|▏         | 237/13317 [02:26<2:13:42,  1.63it/s]
                                                     
{'loss': 0.5012, 'grad_norm': 0.8652413487434387, 'learning_rate': 1.7792792792792792e-06, 'epoch': 0.05}

  2%|▏         | 237/13317 [02:26<2:13:42,  1.63it/s]
  2%|▏         | 238/13317 [02:26<2:13:40,  1.63it/s]
                                                     
{'loss': 0.5094, 'grad_norm': 0.9209291338920593, 'learning_rate': 1.786786786786787e-06, 'epoch': 0.05}

  2%|▏         | 238/13317 [02:27<2:13:40,  1.63it/s]
  2%|▏         | 239/13317 [02:27<2:13:41,  1.63it/s]
                                                     
{'loss': 0.3937, 'grad_norm': 0.5748335719108582, 'learning_rate': 1.7942942942942946e-06, 'epoch': 0.05}

  2%|▏         | 239/13317 [02:27<2:13:41,  1.63it/s]
  2%|▏         | 240/13317 [02:28<2:13:40,  1.63it/s]
                                                     
{'loss': 0.502, 'grad_norm': 0.9317153692245483, 'learning_rate': 1.801801801801802e-06, 'epoch': 0.05}

  2%|▏         | 240/13317 [02:28<2:13:40,  1.63it/s]
  2%|▏         | 241/13317 [02:28<2:13:37,  1.63it/s]
                                                     
{'loss': 0.4844, 'grad_norm': 0.9431211352348328, 'learning_rate': 1.8093093093093095e-06, 'epoch': 0.05}

  2%|▏         | 241/13317 [02:28<2:13:37,  1.63it/s]
  2%|▏         | 242/13317 [02:29<2:13:36,  1.63it/s]
                                                     
{'loss': 0.399, 'grad_norm': 0.7144924998283386, 'learning_rate': 1.816816816816817e-06, 'epoch': 0.05}

  2%|▏         | 242/13317 [02:29<2:13:36,  1.63it/s]
  2%|▏         | 243/13317 [02:30<2:13:38,  1.63it/s]
                                                     
{'loss': 0.5149, 'grad_norm': 0.8570777773857117, 'learning_rate': 1.8243243243243245e-06, 'epoch': 0.05}

  2%|▏         | 243/13317 [02:30<2:13:38,  1.63it/s]
  2%|▏         | 244/13317 [02:30<2:13:37,  1.63it/s]
                                                     
{'loss': 0.4967, 'grad_norm': 0.8890277743339539, 'learning_rate': 1.831831831831832e-06, 'epoch': 0.05}

  2%|▏         | 244/13317 [02:30<2:13:37,  1.63it/s]
  2%|▏         | 245/13317 [02:31<2:13:39,  1.63it/s]
                                                     
{'loss': 0.5036, 'grad_norm': 0.8584218621253967, 'learning_rate': 1.8393393393393394e-06, 'epoch': 0.06}

  2%|▏         | 245/13317 [02:31<2:13:39,  1.63it/s]
  2%|▏         | 246/13317 [02:31<2:13:38,  1.63it/s]
                                                     
{'loss': 0.584, 'grad_norm': 0.9183411002159119, 'learning_rate': 1.846846846846847e-06, 'epoch': 0.06}

  2%|▏         | 246/13317 [02:31<2:13:38,  1.63it/s]
  2%|▏         | 247/13317 [02:32<2:13:36,  1.63it/s]
                                                     
{'loss': 0.6233, 'grad_norm': 0.933334469795227, 'learning_rate': 1.8543543543543545e-06, 'epoch': 0.06}

  2%|▏         | 247/13317 [02:32<2:13:36,  1.63it/s]
  2%|▏         | 248/13317 [02:33<2:13:37,  1.63it/s]
                                                     
{'loss': 0.4422, 'grad_norm': 0.7942708134651184, 'learning_rate': 1.861861861861862e-06, 'epoch': 0.06}

  2%|▏         | 248/13317 [02:33<2:13:37,  1.63it/s]
  2%|▏         | 249/13317 [02:33<2:13:34,  1.63it/s]
                                                     
{'loss': 0.5316, 'grad_norm': 0.8726782202720642, 'learning_rate': 1.8693693693693697e-06, 'epoch': 0.06}

  2%|▏         | 249/13317 [02:33<2:13:34,  1.63it/s]
  2%|▏         | 250/13317 [02:34<2:13:36,  1.63it/s]
                                                     
{'loss': 0.575, 'grad_norm': 0.8334015011787415, 'learning_rate': 1.8768768768768768e-06, 'epoch': 0.06}

  2%|▏         | 250/13317 [02:34<2:13:36,  1.63it/s]
  2%|▏         | 251/13317 [02:34<2:13:34,  1.63it/s]
                                                     
{'loss': 0.5157, 'grad_norm': 0.8072992563247681, 'learning_rate': 1.8843843843843846e-06, 'epoch': 0.06}

  2%|▏         | 251/13317 [02:35<2:13:34,  1.63it/s]
  2%|▏         | 252/13317 [02:35<2:13:30,  1.63it/s]
                                                     
{'loss': 0.4754, 'grad_norm': 0.7553905248641968, 'learning_rate': 1.8918918918918922e-06, 'epoch': 0.06}

  2%|▏         | 252/13317 [02:35<2:13:30,  1.63it/s]
  2%|▏         | 253/13317 [02:36<2:13:30,  1.63it/s]
                                                     
{'loss': 0.5008, 'grad_norm': 0.795973002910614, 'learning_rate': 1.8993993993993996e-06, 'epoch': 0.06}

  2%|▏         | 253/13317 [02:36<2:13:30,  1.63it/s]
  2%|▏         | 254/13317 [02:36<2:13:31,  1.63it/s]
                                                     
{'loss': 0.5195, 'grad_norm': 0.8347752094268799, 'learning_rate': 1.9069069069069071e-06, 'epoch': 0.06}

  2%|▏         | 254/13317 [02:36<2:13:31,  1.63it/s]
  2%|▏         | 255/13317 [02:37<2:13:28,  1.63it/s]
                                                     
{'loss': 0.5405, 'grad_norm': 0.9817022085189819, 'learning_rate': 1.9144144144144145e-06, 'epoch': 0.06}

  2%|▏         | 255/13317 [02:37<2:13:28,  1.63it/s]
  2%|▏         | 256/13317 [02:37<2:13:27,  1.63it/s]
                                                     
{'loss': 0.4624, 'grad_norm': 0.8740941882133484, 'learning_rate': 1.921921921921922e-06, 'epoch': 0.06}

  2%|▏         | 256/13317 [02:38<2:13:27,  1.63it/s]
  2%|▏         | 257/13317 [02:38<2:13:30,  1.63it/s]
                                                     
{'loss': 0.5269, 'grad_norm': 0.7141740322113037, 'learning_rate': 1.9294294294294296e-06, 'epoch': 0.06}

  2%|▏         | 257/13317 [02:38<2:13:30,  1.63it/s]
  2%|▏         | 258/13317 [02:39<2:13:31,  1.63it/s]
                                                     
{'loss': 0.6435, 'grad_norm': 0.9771051406860352, 'learning_rate': 1.9369369369369372e-06, 'epoch': 0.06}

  2%|▏         | 258/13317 [02:39<2:13:31,  1.63it/s]
  2%|▏         | 259/13317 [02:39<2:13:31,  1.63it/s]
                                                     
{'loss': 0.5391, 'grad_norm': 0.8392442464828491, 'learning_rate': 1.944444444444445e-06, 'epoch': 0.06}

  2%|▏         | 259/13317 [02:39<2:13:31,  1.63it/s]
  2%|▏         | 260/13317 [02:40<2:13:31,  1.63it/s]
                                                     
{'loss': 0.4884, 'grad_norm': 0.7445134520530701, 'learning_rate': 1.951951951951952e-06, 'epoch': 0.06}

  2%|▏         | 260/13317 [02:40<2:13:31,  1.63it/s]
  2%|▏         | 261/13317 [02:41<2:13:33,  1.63it/s]
                                                     
{'loss': 0.5251, 'grad_norm': 0.943398118019104, 'learning_rate': 1.9594594594594595e-06, 'epoch': 0.06}

  2%|▏         | 261/13317 [02:41<2:13:33,  1.63it/s]
  2%|▏         | 262/13317 [02:41<2:13:29,  1.63it/s]
                                                     
{'loss': 0.5295, 'grad_norm': 0.944298267364502, 'learning_rate': 1.966966966966967e-06, 'epoch': 0.06}

  2%|▏         | 262/13317 [02:41<2:13:29,  1.63it/s]
  2%|▏         | 263/13317 [02:42<2:13:30,  1.63it/s]
                                                     
{'loss': 0.4428, 'grad_norm': 0.69712895154953, 'learning_rate': 1.9744744744744747e-06, 'epoch': 0.06}

  2%|▏         | 263/13317 [02:42<2:13:30,  1.63it/s]
  2%|▏         | 264/13317 [02:42<2:13:32,  1.63it/s]
                                                     
{'loss': 0.4701, 'grad_norm': 0.7624127268791199, 'learning_rate': 1.9819819819819822e-06, 'epoch': 0.06}

  2%|▏         | 264/13317 [02:43<2:13:32,  1.63it/s]
  2%|▏         | 265/13317 [02:43<2:13:30,  1.63it/s]
                                                     
{'loss': 0.4866, 'grad_norm': 0.7792107462882996, 'learning_rate': 1.98948948948949e-06, 'epoch': 0.06}

  2%|▏         | 265/13317 [02:43<2:13:30,  1.63it/s]
  2%|▏         | 266/13317 [02:44<2:13:27,  1.63it/s]
                                                     
{'loss': 0.5844, 'grad_norm': 0.8491565585136414, 'learning_rate': 1.996996996996997e-06, 'epoch': 0.06}

  2%|▏         | 266/13317 [02:44<2:13:27,  1.63it/s]
  2%|▏         | 267/13317 [02:44<2:13:26,  1.63it/s]
                                                     
{'loss': 0.5189, 'grad_norm': 0.8475441336631775, 'learning_rate': 2.0045045045045045e-06, 'epoch': 0.06}

  2%|▏         | 267/13317 [02:44<2:13:26,  1.63it/s]
  2%|▏         | 268/13317 [02:45<2:13:23,  1.63it/s]
                                                     
{'loss': 0.5341, 'grad_norm': 0.8206396102905273, 'learning_rate': 2.012012012012012e-06, 'epoch': 0.06}

  2%|▏         | 268/13317 [02:45<2:13:23,  1.63it/s]
  2%|▏         | 269/13317 [02:45<2:13:24,  1.63it/s]
                                                     
{'loss': 0.5189, 'grad_norm': 0.841033399105072, 'learning_rate': 2.0195195195195197e-06, 'epoch': 0.06}

  2%|▏         | 269/13317 [02:46<2:13:24,  1.63it/s]
  2%|▏         | 270/13317 [02:46<2:13:25,  1.63it/s]
                                                     
{'loss': 0.5185, 'grad_norm': 0.7907808423042297, 'learning_rate': 2.0270270270270273e-06, 'epoch': 0.06}

  2%|▏         | 270/13317 [02:46<2:13:25,  1.63it/s]
  2%|▏         | 271/13317 [02:47<2:13:23,  1.63it/s]
                                                     
{'loss': 0.5065, 'grad_norm': 0.9445785284042358, 'learning_rate': 2.034534534534535e-06, 'epoch': 0.06}

  2%|▏         | 271/13317 [02:47<2:13:23,  1.63it/s]
  2%|▏         | 272/13317 [02:47<2:13:28,  1.63it/s]
                                                     
{'loss': 0.5054, 'grad_norm': 0.8224158883094788, 'learning_rate': 2.0420420420420424e-06, 'epoch': 0.06}

  2%|▏         | 272/13317 [02:47<2:13:28,  1.63it/s]
  2%|▏         | 273/13317 [02:48<2:13:31,  1.63it/s]
                                                     
{'loss': 0.5465, 'grad_norm': 0.7618244290351868, 'learning_rate': 2.0495495495495496e-06, 'epoch': 0.06}

  2%|▏         | 273/13317 [02:48<2:13:31,  1.63it/s]
  2%|▏         | 274/13317 [02:49<2:13:33,  1.63it/s]
                                                     
{'loss': 0.5051, 'grad_norm': 0.8075991272926331, 'learning_rate': 2.057057057057057e-06, 'epoch': 0.06}

  2%|▏         | 274/13317 [02:49<2:13:33,  1.63it/s]
  2%|▏         | 275/13317 [02:49<2:13:31,  1.63it/s]
                                                     
{'loss': 0.5658, 'grad_norm': 0.8247211575508118, 'learning_rate': 2.0645645645645647e-06, 'epoch': 0.06}

  2%|▏         | 275/13317 [02:49<2:13:31,  1.63it/s]
  2%|▏         | 276/13317 [02:50<2:13:38,  1.63it/s]
                                                     
{'loss': 0.4855, 'grad_norm': 0.7994195818901062, 'learning_rate': 2.0720720720720723e-06, 'epoch': 0.06}

  2%|▏         | 276/13317 [02:50<2:13:38,  1.63it/s]
  2%|▏         | 277/13317 [02:50<2:13:55,  1.62it/s]
                                                     
{'loss': 0.532, 'grad_norm': 0.7584813833236694, 'learning_rate': 2.07957957957958e-06, 'epoch': 0.06}

  2%|▏         | 277/13317 [02:51<2:13:55,  1.62it/s]
  2%|▏         | 278/13317 [02:51<2:13:43,  1.63it/s]
                                                     
{'loss': 0.5082, 'grad_norm': 0.8021715879440308, 'learning_rate': 2.087087087087087e-06, 'epoch': 0.06}

  2%|▏         | 278/13317 [02:51<2:13:43,  1.63it/s]
  2%|▏         | 279/13317 [02:52<2:13:34,  1.63it/s]
                                                     
{'loss': 0.4618, 'grad_norm': 0.7866194248199463, 'learning_rate': 2.0945945945945946e-06, 'epoch': 0.06}

  2%|▏         | 279/13317 [02:52<2:13:34,  1.63it/s]
  2%|▏         | 280/13317 [02:52<2:14:15,  1.62it/s]
                                                     
{'loss': 0.4097, 'grad_norm': 0.6870883703231812, 'learning_rate': 2.102102102102102e-06, 'epoch': 0.06}

  2%|▏         | 280/13317 [02:52<2:14:15,  1.62it/s]
  2%|▏         | 281/13317 [02:53<2:14:01,  1.62it/s]
                                                     
{'loss': 0.4818, 'grad_norm': 0.9430485963821411, 'learning_rate': 2.1096096096096097e-06, 'epoch': 0.06}

  2%|▏         | 281/13317 [02:53<2:14:01,  1.62it/s]
  2%|▏         | 282/13317 [02:53<2:13:47,  1.62it/s]
                                                     
{'loss': 0.4226, 'grad_norm': 0.6739403605461121, 'learning_rate': 2.1171171171171173e-06, 'epoch': 0.06}

  2%|▏         | 282/13317 [02:54<2:13:47,  1.62it/s]
  2%|▏         | 283/13317 [02:54<2:13:37,  1.63it/s]
                                                     
{'loss': 0.3826, 'grad_norm': 0.7420628666877747, 'learning_rate': 2.124624624624625e-06, 'epoch': 0.06}

  2%|▏         | 283/13317 [02:54<2:13:37,  1.63it/s]
  2%|▏         | 284/13317 [02:55<2:13:29,  1.63it/s]
                                                     
{'loss': 0.4964, 'grad_norm': 0.8750054836273193, 'learning_rate': 2.1321321321321325e-06, 'epoch': 0.06}

  2%|▏         | 284/13317 [02:55<2:13:29,  1.63it/s]
  2%|▏         | 285/13317 [02:55<2:13:33,  1.63it/s]
                                                     
{'loss': 0.459, 'grad_norm': 0.774254322052002, 'learning_rate': 2.13963963963964e-06, 'epoch': 0.06}

  2%|▏         | 285/13317 [02:55<2:13:33,  1.63it/s]
  2%|▏         | 286/13317 [02:56<2:13:27,  1.63it/s]
                                                     
{'loss': 0.5433, 'grad_norm': 0.8343825936317444, 'learning_rate': 2.147147147147147e-06, 'epoch': 0.06}

  2%|▏         | 286/13317 [02:56<2:13:27,  1.63it/s]
  2%|▏         | 287/13317 [02:57<2:13:21,  1.63it/s]
                                                     
{'loss': 0.4295, 'grad_norm': 0.8560160398483276, 'learning_rate': 2.1546546546546548e-06, 'epoch': 0.06}

  2%|▏         | 287/13317 [02:57<2:13:21,  1.63it/s]
  2%|▏         | 288/13317 [02:57<2:13:16,  1.63it/s]
                                                     
{'loss': 0.4836, 'grad_norm': 0.8438105583190918, 'learning_rate': 2.1621621621621623e-06, 'epoch': 0.06}

  2%|▏         | 288/13317 [02:57<2:13:16,  1.63it/s]
  2%|▏         | 289/13317 [02:58<2:13:13,  1.63it/s]
                                                     
{'loss': 0.4323, 'grad_norm': 0.7245703339576721, 'learning_rate': 2.16966966966967e-06, 'epoch': 0.07}

  2%|▏         | 289/13317 [02:58<2:13:13,  1.63it/s]
  2%|▏         | 290/13317 [02:58<2:13:22,  1.63it/s]
                                                     
{'loss': 0.4935, 'grad_norm': 0.7448390126228333, 'learning_rate': 2.1771771771771775e-06, 'epoch': 0.07}

  2%|▏         | 290/13317 [02:58<2:13:22,  1.63it/s]
  2%|▏         | 291/13317 [02:59<2:13:17,  1.63it/s]
                                                     
{'loss': 0.487, 'grad_norm': 0.852615237236023, 'learning_rate': 2.1846846846846846e-06, 'epoch': 0.07}

  2%|▏         | 291/13317 [02:59<2:13:17,  1.63it/s]
  2%|▏         | 292/13317 [03:00<2:13:31,  1.63it/s]
                                                     
{'loss': 0.4669, 'grad_norm': 0.7476748824119568, 'learning_rate': 2.192192192192192e-06, 'epoch': 0.07}

  2%|▏         | 292/13317 [03:00<2:13:31,  1.63it/s]
  2%|▏         | 293/13317 [03:00<2:13:25,  1.63it/s]
                                                     
{'loss': 0.5296, 'grad_norm': 0.8050178289413452, 'learning_rate': 2.1996996996996998e-06, 'epoch': 0.07}

  2%|▏         | 293/13317 [03:00<2:13:25,  1.63it/s]
  2%|▏         | 294/13317 [03:01<2:13:21,  1.63it/s]
                                                     
{'loss': 0.4961, 'grad_norm': 0.7921327352523804, 'learning_rate': 2.2072072072072073e-06, 'epoch': 0.07}

  2%|▏         | 294/13317 [03:01<2:13:21,  1.63it/s]
  2%|▏         | 295/13317 [03:01<2:13:37,  1.62it/s]
                                                     
{'loss': 0.4574, 'grad_norm': 0.8301498293876648, 'learning_rate': 2.214714714714715e-06, 'epoch': 0.07}

  2%|▏         | 295/13317 [03:02<2:13:37,  1.62it/s]
  2%|▏         | 296/13317 [03:02<2:13:25,  1.63it/s]
                                                     
{'loss': 0.5313, 'grad_norm': 0.8647900819778442, 'learning_rate': 2.222222222222222e-06, 'epoch': 0.07}

  2%|▏         | 296/13317 [03:02<2:13:25,  1.63it/s]
  2%|▏         | 297/13317 [03:03<2:13:26,  1.63it/s]
                                                     
{'loss': 0.5055, 'grad_norm': 1.0204482078552246, 'learning_rate': 2.22972972972973e-06, 'epoch': 0.07}

  2%|▏         | 297/13317 [03:03<2:13:26,  1.63it/s]
  2%|▏         | 298/13317 [03:03<2:13:15,  1.63it/s]
                                                     
{'loss': 0.4443, 'grad_norm': 0.8799636363983154, 'learning_rate': 2.2372372372372376e-06, 'epoch': 0.07}

  2%|▏         | 298/13317 [03:03<2:13:15,  1.63it/s]
  2%|▏         | 299/13317 [03:04<2:13:11,  1.63it/s]
                                                     
{'loss': 0.5365, 'grad_norm': 0.8481038212776184, 'learning_rate': 2.244744744744745e-06, 'epoch': 0.07}

  2%|▏         | 299/13317 [03:04<2:13:11,  1.63it/s]
  2%|▏         | 300/13317 [03:05<2:13:21,  1.63it/s]
                                                     
{'loss': 0.377, 'grad_norm': 0.7336224913597107, 'learning_rate': 2.2522522522522524e-06, 'epoch': 0.07}

  2%|▏         | 300/13317 [03:05<2:13:21,  1.63it/s]
  2%|▏         | 301/13317 [03:05<2:13:19,  1.63it/s]
                                                     
{'loss': 0.5005, 'grad_norm': 0.7111504673957825, 'learning_rate': 2.25975975975976e-06, 'epoch': 0.07}

  2%|▏         | 301/13317 [03:05<2:13:19,  1.63it/s]
  2%|▏         | 302/13317 [03:06<2:13:11,  1.63it/s]
                                                     
{'loss': 0.4158, 'grad_norm': 0.7109354138374329, 'learning_rate': 2.2672672672672675e-06, 'epoch': 0.07}

  2%|▏         | 302/13317 [03:06<2:13:11,  1.63it/s]
  2%|▏         | 303/13317 [03:06<2:13:09,  1.63it/s]
                                                     
{'loss': 0.4644, 'grad_norm': 0.7365208268165588, 'learning_rate': 2.274774774774775e-06, 'epoch': 0.07}

  2%|▏         | 303/13317 [03:06<2:13:09,  1.63it/s]
  2%|▏         | 304/13317 [03:07<2:13:07,  1.63it/s]
                                                     
{'loss': 0.4992, 'grad_norm': 0.8309431672096252, 'learning_rate': 2.2822822822822822e-06, 'epoch': 0.07}

  2%|▏         | 304/13317 [03:07<2:13:07,  1.63it/s]
  2%|▏         | 305/13317 [03:08<2:13:17,  1.63it/s]
                                                     
{'loss': 0.5318, 'grad_norm': 0.852641761302948, 'learning_rate': 2.28978978978979e-06, 'epoch': 0.07}

  2%|▏         | 305/13317 [03:08<2:13:17,  1.63it/s]
  2%|▏         | 306/13317 [03:08<2:13:11,  1.63it/s]
                                                     
{'loss': 0.4677, 'grad_norm': 0.7747059464454651, 'learning_rate': 2.297297297297298e-06, 'epoch': 0.07}

  2%|▏         | 306/13317 [03:08<2:13:11,  1.63it/s]
  2%|▏         | 307/13317 [03:09<2:13:04,  1.63it/s]
                                                     
{'loss': 0.5994, 'grad_norm': 0.8614656329154968, 'learning_rate': 2.304804804804805e-06, 'epoch': 0.07}

  2%|▏         | 307/13317 [03:09<2:13:04,  1.63it/s]
  2%|▏         | 308/13317 [03:09<2:13:01,  1.63it/s]
                                                     
{'loss': 0.5473, 'grad_norm': 0.8134952187538147, 'learning_rate': 2.3123123123123125e-06, 'epoch': 0.07}

  2%|▏         | 308/13317 [03:10<2:13:01,  1.63it/s]
  2%|▏         | 309/13317 [03:10<2:13:02,  1.63it/s]
                                                     
{'loss': 0.65, 'grad_norm': 0.893435001373291, 'learning_rate': 2.31981981981982e-06, 'epoch': 0.07}

  2%|▏         | 309/13317 [03:10<2:13:02,  1.63it/s]
  2%|▏         | 310/13317 [03:11<2:13:15,  1.63it/s]
                                                     
{'loss': 0.4952, 'grad_norm': 0.8762337565422058, 'learning_rate': 2.3273273273273277e-06, 'epoch': 0.07}

  2%|▏         | 310/13317 [03:11<2:13:15,  1.63it/s]
  2%|▏         | 311/13317 [03:11<2:13:09,  1.63it/s]
                                                     
{'loss': 0.474, 'grad_norm': 0.7566710710525513, 'learning_rate': 2.3348348348348353e-06, 'epoch': 0.07}

  2%|▏         | 311/13317 [03:11<2:13:09,  1.63it/s]
  2%|▏         | 312/13317 [03:12<2:13:01,  1.63it/s]
                                                     
{'loss': 0.3861, 'grad_norm': 0.64943528175354, 'learning_rate': 2.3423423423423424e-06, 'epoch': 0.07}

  2%|▏         | 312/13317 [03:12<2:13:01,  1.63it/s]
  2%|▏         | 313/13317 [03:12<2:12:59,  1.63it/s]
                                                     
{'loss': 0.5086, 'grad_norm': 0.7462379932403564, 'learning_rate': 2.34984984984985e-06, 'epoch': 0.07}

  2%|▏         | 313/13317 [03:13<2:12:59,  1.63it/s]
  2%|▏         | 314/13317 [03:13<2:13:03,  1.63it/s]
                                                     
{'loss': 0.4086, 'grad_norm': 0.701115608215332, 'learning_rate': 2.3573573573573576e-06, 'epoch': 0.07}

  2%|▏         | 314/13317 [03:13<2:13:03,  1.63it/s]
  2%|▏         | 315/13317 [03:14<2:13:15,  1.63it/s]
                                                     
{'loss': 0.4928, 'grad_norm': 0.816949725151062, 'learning_rate': 2.364864864864865e-06, 'epoch': 0.07}

  2%|▏         | 315/13317 [03:14<2:13:15,  1.63it/s]
  2%|▏         | 316/13317 [03:14<2:13:11,  1.63it/s]
                                                     
{'loss': 0.4355, 'grad_norm': 0.8386324644088745, 'learning_rate': 2.3723723723723727e-06, 'epoch': 0.07}

  2%|▏         | 316/13317 [03:14<2:13:11,  1.63it/s]
  2%|▏         | 317/13317 [03:15<2:13:04,  1.63it/s]
                                                     
{'loss': 0.541, 'grad_norm': 0.7953378558158875, 'learning_rate': 2.37987987987988e-06, 'epoch': 0.07}

  2%|▏         | 317/13317 [03:15<2:13:04,  1.63it/s]
  2%|▏         | 318/13317 [03:16<2:13:01,  1.63it/s]
                                                     
{'loss': 0.4309, 'grad_norm': 0.7725192904472351, 'learning_rate': 2.3873873873873874e-06, 'epoch': 0.07}

  2%|▏         | 318/13317 [03:16<2:13:01,  1.63it/s]
  2%|▏         | 319/13317 [03:16<2:13:02,  1.63it/s]
                                                     
{'loss': 0.5655, 'grad_norm': 0.8880619406700134, 'learning_rate': 2.394894894894895e-06, 'epoch': 0.07}

  2%|▏         | 319/13317 [03:16<2:13:02,  1.63it/s]
  2%|▏         | 320/13317 [03:17<2:13:13,  1.63it/s]
                                                     
{'loss': 0.5018, 'grad_norm': 0.7653263211250305, 'learning_rate': 2.4024024024024026e-06, 'epoch': 0.07}

  2%|▏         | 320/13317 [03:17<2:13:13,  1.63it/s]
  2%|▏         | 321/13317 [03:17<2:13:06,  1.63it/s]
                                                     
{'loss': 0.535, 'grad_norm': 0.9908320903778076, 'learning_rate': 2.40990990990991e-06, 'epoch': 0.07}

  2%|▏         | 321/13317 [03:18<2:13:06,  1.63it/s]
  2%|▏         | 322/13317 [03:18<2:13:05,  1.63it/s]
                                                     
{'loss': 0.4911, 'grad_norm': 0.8123887181282043, 'learning_rate': 2.4174174174174177e-06, 'epoch': 0.07}

  2%|▏         | 322/13317 [03:18<2:13:05,  1.63it/s]
  2%|▏         | 323/13317 [03:19<2:13:00,  1.63it/s]
                                                     
{'loss': 0.4508, 'grad_norm': 0.7047619223594666, 'learning_rate': 2.4249249249249253e-06, 'epoch': 0.07}

  2%|▏         | 323/13317 [03:19<2:13:00,  1.63it/s]
  2%|▏         | 324/13317 [03:19<2:12:54,  1.63it/s]
                                                     
{'loss': 0.4737, 'grad_norm': 0.7331838607788086, 'learning_rate': 2.432432432432433e-06, 'epoch': 0.07}

  2%|▏         | 324/13317 [03:19<2:12:54,  1.63it/s]
  2%|▏         | 325/13317 [03:20<2:13:09,  1.63it/s]
                                                     
{'loss': 0.4795, 'grad_norm': 0.8097270131111145, 'learning_rate': 2.43993993993994e-06, 'epoch': 0.07}

  2%|▏         | 325/13317 [03:20<2:13:09,  1.63it/s]
  2%|▏         | 326/13317 [03:20<2:13:03,  1.63it/s]
                                                     
{'loss': 0.5617, 'grad_norm': 0.9416517019271851, 'learning_rate': 2.4474474474474476e-06, 'epoch': 0.07}

  2%|▏         | 326/13317 [03:21<2:13:03,  1.63it/s]
  2%|▏         | 327/13317 [03:21<2:12:55,  1.63it/s]
                                                     
{'loss': 0.4613, 'grad_norm': 0.7812457084655762, 'learning_rate': 2.454954954954955e-06, 'epoch': 0.07}

  2%|▏         | 327/13317 [03:21<2:12:55,  1.63it/s]
  2%|▏         | 328/13317 [03:22<2:12:50,  1.63it/s]
                                                     
{'loss': 0.4474, 'grad_norm': 0.7613444924354553, 'learning_rate': 2.4624624624624628e-06, 'epoch': 0.07}

  2%|▏         | 328/13317 [03:22<2:12:50,  1.63it/s]
  2%|▏         | 329/13317 [03:22<2:12:51,  1.63it/s]
                                                     
{'loss': 0.4647, 'grad_norm': 0.6596733927726746, 'learning_rate': 2.4699699699699703e-06, 'epoch': 0.07}

  2%|▏         | 329/13317 [03:22<2:12:51,  1.63it/s]
  2%|▏         | 330/13317 [03:23<2:12:59,  1.63it/s]
                                                     
{'loss': 0.492, 'grad_norm': 0.7416242957115173, 'learning_rate': 2.4774774774774775e-06, 'epoch': 0.07}

  2%|▏         | 330/13317 [03:23<2:12:59,  1.63it/s]
  2%|▏         | 331/13317 [03:24<2:12:55,  1.63it/s]
                                                     
{'loss': 0.4028, 'grad_norm': 0.6864984035491943, 'learning_rate': 2.484984984984985e-06, 'epoch': 0.07}

  2%|▏         | 331/13317 [03:24<2:12:55,  1.63it/s]
  2%|▏         | 332/13317 [03:24<2:19:49,  1.55it/s]
                                                     
{'loss': 0.5702, 'grad_norm': 0.8038687705993652, 'learning_rate': 2.4924924924924926e-06, 'epoch': 0.07}

  2%|▏         | 332/13317 [03:24<2:19:49,  1.55it/s]
  3%|▎         | 333/13317 [03:25<2:17:38,  1.57it/s]
                                                     
{'loss': 0.4988, 'grad_norm': 0.8439292311668396, 'learning_rate': 2.5e-06, 'epoch': 0.08}

  3%|▎         | 333/13317 [03:25<2:17:38,  1.57it/s]
  3%|▎         | 334/13317 [03:25<2:16:10,  1.59it/s]
                                                     
{'loss': 0.4337, 'grad_norm': 0.733538806438446, 'learning_rate': 2.5075075075075078e-06, 'epoch': 0.08}

  3%|▎         | 334/13317 [03:26<2:16:10,  1.59it/s]
  3%|▎         | 335/13317 [03:26<2:15:17,  1.60it/s]
                                                     
{'loss': 0.5706, 'grad_norm': 0.8551611304283142, 'learning_rate': 2.5150150150150154e-06, 'epoch': 0.08}

  3%|▎         | 335/13317 [03:26<2:15:17,  1.60it/s]
  3%|▎         | 336/13317 [03:27<2:14:28,  1.61it/s]
                                                     
{'loss': 0.5388, 'grad_norm': 0.9304776787757874, 'learning_rate': 2.5225225225225225e-06, 'epoch': 0.08}

  3%|▎         | 336/13317 [03:27<2:14:28,  1.61it/s]
  3%|▎         | 337/13317 [03:27<2:13:55,  1.62it/s]
                                                     
{'loss': 0.4888, 'grad_norm': 0.8561441898345947, 'learning_rate': 2.53003003003003e-06, 'epoch': 0.08}

  3%|▎         | 337/13317 [03:27<2:13:55,  1.62it/s]
  3%|▎         | 338/13317 [03:28<2:13:32,  1.62it/s]
                                                     
{'loss': 0.458, 'grad_norm': 0.7742632031440735, 'learning_rate': 2.5375375375375377e-06, 'epoch': 0.08}

  3%|▎         | 338/13317 [03:28<2:13:32,  1.62it/s]
  3%|▎         | 339/13317 [03:29<2:13:20,  1.62it/s]
                                                     
{'loss': 0.4706, 'grad_norm': 0.8861637115478516, 'learning_rate': 2.5450450450450452e-06, 'epoch': 0.08}

  3%|▎         | 339/13317 [03:29<2:13:20,  1.62it/s]
  3%|▎         | 340/13317 [03:29<2:13:05,  1.62it/s]
                                                     
{'loss': 0.4587, 'grad_norm': 0.9356284737586975, 'learning_rate': 2.552552552552553e-06, 'epoch': 0.08}

  3%|▎         | 340/13317 [03:29<2:13:05,  1.62it/s]
  3%|▎         | 341/13317 [03:30<2:13:01,  1.63it/s]
                                                     
{'loss': 0.4935, 'grad_norm': 0.7564237117767334, 'learning_rate': 2.56006006006006e-06, 'epoch': 0.08}

  3%|▎         | 341/13317 [03:30<2:13:01,  1.63it/s]
  3%|▎         | 342/13317 [03:30<2:12:53,  1.63it/s]
                                                     
{'loss': 0.6443, 'grad_norm': 0.7717772126197815, 'learning_rate': 2.5675675675675675e-06, 'epoch': 0.08}

  3%|▎         | 342/13317 [03:31<2:12:53,  1.63it/s]
  3%|▎         | 343/13317 [03:31<2:12:46,  1.63it/s]
                                                     
{'loss': 0.406, 'grad_norm': 0.6767715811729431, 'learning_rate': 2.575075075075075e-06, 'epoch': 0.08}

  3%|▎         | 343/13317 [03:31<2:12:46,  1.63it/s]
  3%|▎         | 344/13317 [03:32<2:13:07,  1.62it/s]
                                                     
{'loss': 0.5462, 'grad_norm': 0.7747556567192078, 'learning_rate': 2.5825825825825827e-06, 'epoch': 0.08}

  3%|▎         | 344/13317 [03:32<2:13:07,  1.62it/s]
  3%|▎         | 345/13317 [03:32<2:13:00,  1.63it/s]
                                                     
{'loss': 0.3798, 'grad_norm': 0.6473003625869751, 'learning_rate': 2.5900900900900907e-06, 'epoch': 0.08}

  3%|▎         | 345/13317 [03:32<2:13:00,  1.63it/s]
  3%|▎         | 346/13317 [03:33<2:12:58,  1.63it/s]
                                                     
{'loss': 0.5846, 'grad_norm': 0.8775733709335327, 'learning_rate': 2.5975975975975974e-06, 'epoch': 0.08}

  3%|▎         | 346/13317 [03:33<2:12:58,  1.63it/s]
  3%|▎         | 347/13317 [03:33<2:12:50,  1.63it/s]
                                                     
{'loss': 0.4178, 'grad_norm': 0.7576457262039185, 'learning_rate': 2.605105105105105e-06, 'epoch': 0.08}

  3%|▎         | 347/13317 [03:34<2:12:50,  1.63it/s]
  3%|▎         | 348/13317 [03:34<2:12:48,  1.63it/s]
                                                     
{'loss': 0.5195, 'grad_norm': 0.7039090394973755, 'learning_rate': 2.612612612612613e-06, 'epoch': 0.08}

  3%|▎         | 348/13317 [03:34<2:12:48,  1.63it/s]
  3%|▎         | 349/13317 [03:35<2:12:47,  1.63it/s]
                                                     
{'loss': 0.448, 'grad_norm': 0.7303341031074524, 'learning_rate': 2.6201201201201205e-06, 'epoch': 0.08}

  3%|▎         | 349/13317 [03:35<2:12:47,  1.63it/s]
  3%|▎         | 350/13317 [03:35<2:12:40,  1.63it/s]
                                                     
{'loss': 0.4704, 'grad_norm': 0.7013117074966431, 'learning_rate': 2.627627627627628e-06, 'epoch': 0.08}

  3%|▎         | 350/13317 [03:35<2:12:40,  1.63it/s]
  3%|▎         | 351/13317 [03:36<2:12:43,  1.63it/s]
                                                     
{'loss': 0.5373, 'grad_norm': 0.7986212968826294, 'learning_rate': 2.6351351351351353e-06, 'epoch': 0.08}

  3%|▎         | 351/13317 [03:36<2:12:43,  1.63it/s]
  3%|▎         | 352/13317 [03:37<2:12:40,  1.63it/s]
                                                     
{'loss': 0.5989, 'grad_norm': 0.8149718046188354, 'learning_rate': 2.642642642642643e-06, 'epoch': 0.08}

  3%|▎         | 352/13317 [03:37<2:12:40,  1.63it/s]
  3%|▎         | 353/13317 [03:37<2:12:48,  1.63it/s]
                                                     
{'loss': 0.5062, 'grad_norm': 0.7915011048316956, 'learning_rate': 2.6501501501501504e-06, 'epoch': 0.08}

  3%|▎         | 353/13317 [03:37<2:12:48,  1.63it/s]
  3%|▎         | 354/13317 [03:38<2:12:44,  1.63it/s]
                                                     
{'loss': 0.5307, 'grad_norm': 0.7167931795120239, 'learning_rate': 2.657657657657658e-06, 'epoch': 0.08}

  3%|▎         | 354/13317 [03:38<2:12:44,  1.63it/s]
  3%|▎         | 355/13317 [03:38<2:12:36,  1.63it/s]
                                                     
{'loss': 0.6471, 'grad_norm': 0.8861395716667175, 'learning_rate': 2.6651651651651656e-06, 'epoch': 0.08}

  3%|▎         | 355/13317 [03:39<2:12:36,  1.63it/s]
  3%|▎         | 356/13317 [03:39<2:12:36,  1.63it/s]
                                                     
{'loss': 0.4904, 'grad_norm': 0.767351508140564, 'learning_rate': 2.672672672672673e-06, 'epoch': 0.08}

  3%|▎         | 356/13317 [03:39<2:12:36,  1.63it/s]
  3%|▎         | 357/13317 [03:40<2:12:33,  1.63it/s]
                                                     
{'loss': 0.4671, 'grad_norm': 0.8061123490333557, 'learning_rate': 2.6801801801801803e-06, 'epoch': 0.08}

  3%|▎         | 357/13317 [03:40<2:12:33,  1.63it/s]
  3%|▎         | 358/13317 [03:40<2:12:39,  1.63it/s]
                                                     
{'loss': 0.4055, 'grad_norm': 0.6337212324142456, 'learning_rate': 2.687687687687688e-06, 'epoch': 0.08}

  3%|▎         | 358/13317 [03:40<2:12:39,  1.63it/s]
  3%|▎         | 359/13317 [03:41<2:12:37,  1.63it/s]
                                                     
{'loss': 0.4451, 'grad_norm': 0.8043761849403381, 'learning_rate': 2.6951951951951954e-06, 'epoch': 0.08}

  3%|▎         | 359/13317 [03:41<2:12:37,  1.63it/s]
  3%|▎         | 360/13317 [03:41<2:12:41,  1.63it/s]
                                                     
{'loss': 0.3945, 'grad_norm': 0.628926157951355, 'learning_rate': 2.702702702702703e-06, 'epoch': 0.08}

  3%|▎         | 360/13317 [03:42<2:12:41,  1.63it/s]
  3%|▎         | 361/13317 [03:42<2:20:11,  1.54it/s]
                                                     
{'loss': 0.515, 'grad_norm': 0.8958503007888794, 'learning_rate': 2.7102102102102106e-06, 'epoch': 0.08}

  3%|▎         | 361/13317 [03:42<2:20:11,  1.54it/s]
  3%|▎         | 362/13317 [03:43<2:17:53,  1.57it/s]
                                                     
{'loss': 0.4603, 'grad_norm': 0.7154403328895569, 'learning_rate': 2.7177177177177177e-06, 'epoch': 0.08}

  3%|▎         | 362/13317 [03:43<2:17:53,  1.57it/s]
  3%|▎         | 363/13317 [03:43<2:16:28,  1.58it/s]
                                                     
{'loss': 0.5884, 'grad_norm': 0.8347815871238708, 'learning_rate': 2.7252252252252253e-06, 'epoch': 0.08}

  3%|▎         | 363/13317 [03:44<2:16:28,  1.58it/s]
  3%|▎         | 364/13317 [03:44<2:15:14,  1.60it/s]
                                                     
{'loss': 0.4976, 'grad_norm': 0.8042851686477661, 'learning_rate': 2.732732732732733e-06, 'epoch': 0.08}

  3%|▎         | 364/13317 [03:44<2:15:14,  1.60it/s]
  3%|▎         | 365/13317 [03:45<2:14:24,  1.61it/s]
                                                     
{'loss': 0.421, 'grad_norm': 0.6979194283485413, 'learning_rate': 2.7402402402402405e-06, 'epoch': 0.08}

  3%|▎         | 365/13317 [03:45<2:14:24,  1.61it/s]
  3%|▎         | 366/13317 [03:45<2:13:46,  1.61it/s]
                                                     
{'loss': 0.4169, 'grad_norm': 0.6536282896995544, 'learning_rate': 2.747747747747748e-06, 'epoch': 0.08}

  3%|▎         | 366/13317 [03:45<2:13:46,  1.61it/s]
  3%|▎         | 367/13317 [03:46<2:13:20,  1.62it/s]
                                                     
{'loss': 0.4715, 'grad_norm': 0.6974200010299683, 'learning_rate': 2.755255255255255e-06, 'epoch': 0.08}

  3%|▎         | 367/13317 [03:46<2:13:20,  1.62it/s]
  3%|▎         | 368/13317 [03:46<2:13:03,  1.62it/s]
                                                     
{'loss': 0.4677, 'grad_norm': 0.6706352233886719, 'learning_rate': 2.7627627627627628e-06, 'epoch': 0.08}

  3%|▎         | 368/13317 [03:47<2:13:03,  1.62it/s]
  3%|▎         | 369/13317 [03:47<2:12:47,  1.63it/s]
                                                     
{'loss': 0.3203, 'grad_norm': 0.6404407620429993, 'learning_rate': 2.7702702702702703e-06, 'epoch': 0.08}

  3%|▎         | 369/13317 [03:47<2:12:47,  1.63it/s]
  3%|▎         | 370/13317 [03:48<2:12:37,  1.63it/s]
                                                     
{'loss': 0.3854, 'grad_norm': 0.712901771068573, 'learning_rate': 2.7777777777777783e-06, 'epoch': 0.08}

  3%|▎         | 370/13317 [03:48<2:12:37,  1.63it/s]
  3%|▎         | 371/13317 [03:48<2:12:39,  1.63it/s]
                                                     
{'loss': 0.4965, 'grad_norm': 0.820353090763092, 'learning_rate': 2.785285285285286e-06, 'epoch': 0.08}

  3%|▎         | 371/13317 [03:48<2:12:39,  1.63it/s]
  3%|▎         | 372/13317 [03:49<2:12:36,  1.63it/s]
                                                     
{'loss': 0.5023, 'grad_norm': 0.7078438997268677, 'learning_rate': 2.7927927927927926e-06, 'epoch': 0.08}

  3%|▎         | 372/13317 [03:49<2:12:36,  1.63it/s]
  3%|▎         | 373/13317 [03:50<2:12:30,  1.63it/s]
                                                     
{'loss': 0.3651, 'grad_norm': 0.7108337879180908, 'learning_rate': 2.8003003003003006e-06, 'epoch': 0.08}

  3%|▎         | 373/13317 [03:50<2:12:30,  1.63it/s]
  3%|▎         | 374/13317 [03:50<2:12:29,  1.63it/s]
                                                     
{'loss': 0.4168, 'grad_norm': 0.6355459094047546, 'learning_rate': 2.807807807807808e-06, 'epoch': 0.08}

  3%|▎         | 374/13317 [03:50<2:12:29,  1.63it/s]
  3%|▎         | 375/13317 [03:51<2:12:27,  1.63it/s]
                                                     
{'loss': 0.4343, 'grad_norm': 0.7654605507850647, 'learning_rate': 2.8153153153153158e-06, 'epoch': 0.08}

  3%|▎         | 375/13317 [03:51<2:12:27,  1.63it/s]
  3%|▎         | 376/13317 [03:51<2:12:33,  1.63it/s]
                                                     
{'loss': 0.4584, 'grad_norm': 0.7532657384872437, 'learning_rate': 2.8228228228228234e-06, 'epoch': 0.08}

  3%|▎         | 376/13317 [03:52<2:12:33,  1.63it/s]
  3%|▎         | 377/13317 [03:52<2:12:23,  1.63it/s]
                                                     
{'loss': 0.4836, 'grad_norm': 0.7189052104949951, 'learning_rate': 2.8303303303303305e-06, 'epoch': 0.08}

  3%|▎         | 377/13317 [03:52<2:12:23,  1.63it/s]
  3%|▎         | 378/13317 [03:53<2:12:24,  1.63it/s]
                                                     
{'loss': 0.4595, 'grad_norm': 0.8073223829269409, 'learning_rate': 2.837837837837838e-06, 'epoch': 0.09}

  3%|▎         | 378/13317 [03:53<2:12:24,  1.63it/s]
  3%|▎         | 379/13317 [03:53<2:12:19,  1.63it/s]
                                                     
{'loss': 0.4275, 'grad_norm': 0.742399275302887, 'learning_rate': 2.8453453453453457e-06, 'epoch': 0.09}

  3%|▎         | 379/13317 [03:53<2:12:19,  1.63it/s]
  3%|▎         | 380/13317 [03:54<2:12:21,  1.63it/s]
                                                     
{'loss': 0.4115, 'grad_norm': 0.9317102432250977, 'learning_rate': 2.8528528528528532e-06, 'epoch': 0.09}

  3%|▎         | 380/13317 [03:54<2:12:21,  1.63it/s]
  3%|▎         | 381/13317 [03:54<2:12:25,  1.63it/s]
                                                     
{'loss': 0.4723, 'grad_norm': 0.8024287819862366, 'learning_rate': 2.860360360360361e-06, 'epoch': 0.09}

  3%|▎         | 381/13317 [03:55<2:12:25,  1.63it/s]
  3%|▎         | 382/13317 [03:55<2:12:23,  1.63it/s]
                                                     
{'loss': 0.4736, 'grad_norm': 0.6790726780891418, 'learning_rate': 2.867867867867868e-06, 'epoch': 0.09}

  3%|▎         | 382/13317 [03:55<2:12:23,  1.63it/s]
  3%|▎         | 383/13317 [03:56<2:12:15,  1.63it/s]
                                                     
{'loss': 0.4309, 'grad_norm': 0.8539626598358154, 'learning_rate': 2.8753753753753755e-06, 'epoch': 0.09}

  3%|▎         | 383/13317 [03:56<2:12:15,  1.63it/s]
  3%|▎         | 384/13317 [03:56<2:12:19,  1.63it/s]
                                                     
{'loss': 0.3837, 'grad_norm': 0.7407821416854858, 'learning_rate': 2.882882882882883e-06, 'epoch': 0.09}

  3%|▎         | 384/13317 [03:56<2:12:19,  1.63it/s]
  3%|▎         | 385/13317 [03:57<2:12:15,  1.63it/s]
                                                     
{'loss': 0.4586, 'grad_norm': 0.8013657331466675, 'learning_rate': 2.8903903903903907e-06, 'epoch': 0.09}

  3%|▎         | 385/13317 [03:57<2:12:15,  1.63it/s]
  3%|▎         | 386/13317 [03:58<2:12:18,  1.63it/s]
                                                     
{'loss': 0.4814, 'grad_norm': 0.6932796239852905, 'learning_rate': 2.8978978978978983e-06, 'epoch': 0.09}

  3%|▎         | 386/13317 [03:58<2:12:18,  1.63it/s]
  3%|▎         | 387/13317 [03:58<2:12:16,  1.63it/s]
                                                     
{'loss': 0.345, 'grad_norm': 0.7515282034873962, 'learning_rate': 2.9054054054054054e-06, 'epoch': 0.09}

  3%|▎         | 387/13317 [03:58<2:12:16,  1.63it/s]
  3%|▎         | 388/13317 [03:59<2:12:15,  1.63it/s]
                                                     
{'loss': 0.4296, 'grad_norm': 0.7212568521499634, 'learning_rate': 2.912912912912913e-06, 'epoch': 0.09}

  3%|▎         | 388/13317 [03:59<2:12:15,  1.63it/s]
  3%|▎         | 389/13317 [03:59<2:12:11,  1.63it/s]
                                                     
{'loss': 0.4345, 'grad_norm': 0.6890455484390259, 'learning_rate': 2.9204204204204206e-06, 'epoch': 0.09}

  3%|▎         | 389/13317 [04:00<2:12:11,  1.63it/s]
  3%|▎         | 390/13317 [04:00<2:12:11,  1.63it/s]
                                                     
{'loss': 0.3573, 'grad_norm': 0.6826176047325134, 'learning_rate': 2.927927927927928e-06, 'epoch': 0.09}

  3%|▎         | 390/13317 [04:00<2:12:11,  1.63it/s]
  3%|▎         | 391/13317 [04:01<2:12:18,  1.63it/s]
                                                     
{'loss': 0.4455, 'grad_norm': 0.7777101993560791, 'learning_rate': 2.9354354354354357e-06, 'epoch': 0.09}

  3%|▎         | 391/13317 [04:01<2:12:18,  1.63it/s]
  3%|▎         | 392/13317 [04:01<2:12:13,  1.63it/s]
                                                     
{'loss': 0.4204, 'grad_norm': 0.6666024923324585, 'learning_rate': 2.942942942942943e-06, 'epoch': 0.09}

  3%|▎         | 392/13317 [04:01<2:12:13,  1.63it/s]
  3%|▎         | 393/13317 [04:02<2:12:13,  1.63it/s]
                                                     
{'loss': 0.455, 'grad_norm': 0.7357118725776672, 'learning_rate': 2.9504504504504504e-06, 'epoch': 0.09}

  3%|▎         | 393/13317 [04:02<2:12:13,  1.63it/s]
  3%|▎         | 394/13317 [04:02<2:12:08,  1.63it/s]
                                                     
{'loss': 0.3786, 'grad_norm': 0.6265907883644104, 'learning_rate': 2.957957957957958e-06, 'epoch': 0.09}

  3%|▎         | 394/13317 [04:03<2:12:08,  1.63it/s]
  3%|▎         | 395/13317 [04:03<2:12:03,  1.63it/s]
                                                     
{'loss': 0.3602, 'grad_norm': 0.6242502927780151, 'learning_rate': 2.9654654654654656e-06, 'epoch': 0.09}

  3%|▎         | 395/13317 [04:03<2:12:03,  1.63it/s]
  3%|▎         | 396/13317 [04:04<2:12:17,  1.63it/s]
                                                     
{'loss': 0.4138, 'grad_norm': 0.7104613184928894, 'learning_rate': 2.9729729729729736e-06, 'epoch': 0.09}

  3%|▎         | 396/13317 [04:04<2:12:17,  1.63it/s]
  3%|▎         | 397/13317 [04:04<2:12:11,  1.63it/s]
                                                     
{'loss': 0.4007, 'grad_norm': 0.688345730304718, 'learning_rate': 2.980480480480481e-06, 'epoch': 0.09}

  3%|▎         | 397/13317 [04:04<2:12:11,  1.63it/s]
  3%|▎         | 398/13317 [04:05<2:12:08,  1.63it/s]
                                                     
{'loss': 0.4401, 'grad_norm': 0.62779700756073, 'learning_rate': 2.987987987987988e-06, 'epoch': 0.09}

  3%|▎         | 398/13317 [04:05<2:12:08,  1.63it/s]
  3%|▎         | 399/13317 [04:06<2:12:04,  1.63it/s]
                                                     
{'loss': 0.4056, 'grad_norm': 0.7132279276847839, 'learning_rate': 2.995495495495496e-06, 'epoch': 0.09}

  3%|▎         | 399/13317 [04:06<2:12:04,  1.63it/s]
  3%|▎         | 400/13317 [04:06<2:12:03,  1.63it/s]
                                                     
{'loss': 0.4577, 'grad_norm': 0.6780898571014404, 'learning_rate': 3.0030030030030034e-06, 'epoch': 0.09}

  3%|▎         | 400/13317 [04:06<2:12:03,  1.63it/s]
  3%|▎         | 401/13317 [04:07<2:12:24,  1.63it/s]
                                                     
{'loss': 0.4987, 'grad_norm': 0.7893285751342773, 'learning_rate': 3.010510510510511e-06, 'epoch': 0.09}

  3%|▎         | 401/13317 [04:07<2:12:24,  1.63it/s]
  3%|▎         | 402/13317 [04:07<2:12:23,  1.63it/s]
                                                     
{'loss': 0.4325, 'grad_norm': 0.6596247553825378, 'learning_rate': 3.0180180180180186e-06, 'epoch': 0.09}

  3%|▎         | 402/13317 [04:07<2:12:23,  1.63it/s]
  3%|▎         | 403/13317 [04:08<2:12:20,  1.63it/s]
                                                     
{'loss': 0.3937, 'grad_norm': 0.7547261714935303, 'learning_rate': 3.0255255255255257e-06, 'epoch': 0.09}

  3%|▎         | 403/13317 [04:08<2:12:20,  1.63it/s]
  3%|▎         | 404/13317 [04:09<2:12:15,  1.63it/s]
                                                     
{'loss': 0.5235, 'grad_norm': 0.7699780464172363, 'learning_rate': 3.0330330330330333e-06, 'epoch': 0.09}

  3%|▎         | 404/13317 [04:09<2:12:15,  1.63it/s]
  3%|▎         | 405/13317 [04:09<2:12:09,  1.63it/s]
                                                     
{'loss': 0.5291, 'grad_norm': 0.7426941394805908, 'learning_rate': 3.040540540540541e-06, 'epoch': 0.09}

  3%|▎         | 405/13317 [04:09<2:12:09,  1.63it/s]
  3%|▎         | 406/13317 [04:10<2:12:09,  1.63it/s]
                                                     
{'loss': 0.4079, 'grad_norm': 0.7774872183799744, 'learning_rate': 3.0480480480480485e-06, 'epoch': 0.09}

  3%|▎         | 406/13317 [04:10<2:12:09,  1.63it/s]
  3%|▎         | 407/13317 [04:10<2:12:05,  1.63it/s]
                                                     
{'loss': 0.4196, 'grad_norm': 0.767114520072937, 'learning_rate': 3.055555555555556e-06, 'epoch': 0.09}

  3%|▎         | 407/13317 [04:11<2:12:05,  1.63it/s]
  3%|▎         | 408/13317 [04:11<2:12:43,  1.62it/s]
                                                     
{'loss': 0.4518, 'grad_norm': 0.7248908877372742, 'learning_rate': 3.063063063063063e-06, 'epoch': 0.09}

  3%|▎         | 408/13317 [04:11<2:12:43,  1.62it/s]
  3%|▎         | 409/13317 [04:12<2:12:31,  1.62it/s]
                                                     
{'loss': 0.4427, 'grad_norm': 0.7351310849189758, 'learning_rate': 3.0705705705705708e-06, 'epoch': 0.09}

  3%|▎         | 409/13317 [04:12<2:12:31,  1.62it/s]
  3%|▎         | 410/13317 [04:12<2:12:20,  1.63it/s]
                                                     
{'loss': 0.4752, 'grad_norm': 0.7025326490402222, 'learning_rate': 3.0780780780780783e-06, 'epoch': 0.09}

  3%|▎         | 410/13317 [04:12<2:12:20,  1.63it/s]
  3%|▎         | 411/13317 [04:13<2:12:17,  1.63it/s]
                                                     
{'loss': 0.4543, 'grad_norm': 0.6488361358642578, 'learning_rate': 3.085585585585586e-06, 'epoch': 0.09}

  3%|▎         | 411/13317 [04:13<2:12:17,  1.63it/s]
  3%|▎         | 412/13317 [04:14<2:12:12,  1.63it/s]
                                                     
{'loss': 0.4685, 'grad_norm': 0.6951853632926941, 'learning_rate': 3.0930930930930935e-06, 'epoch': 0.09}

  3%|▎         | 412/13317 [04:14<2:12:12,  1.63it/s]
  3%|▎         | 413/13317 [04:14<2:12:15,  1.63it/s]
                                                     
{'loss': 0.4958, 'grad_norm': 0.8334890604019165, 'learning_rate': 3.1006006006006006e-06, 'epoch': 0.09}

  3%|▎         | 413/13317 [04:14<2:12:15,  1.63it/s]
  3%|▎         | 414/13317 [04:15<2:12:06,  1.63it/s]
                                                     
{'loss': 0.3908, 'grad_norm': 0.6970667243003845, 'learning_rate': 3.1081081081081082e-06, 'epoch': 0.09}

  3%|▎         | 414/13317 [04:15<2:12:06,  1.63it/s]
  3%|▎         | 415/13317 [04:15<2:12:01,  1.63it/s]
                                                     
{'loss': 0.5229, 'grad_norm': 0.7065516710281372, 'learning_rate': 3.1156156156156158e-06, 'epoch': 0.09}

  3%|▎         | 415/13317 [04:15<2:12:01,  1.63it/s]
  3%|▎         | 416/13317 [04:16<2:12:05,  1.63it/s]
                                                     
{'loss': 0.455, 'grad_norm': 0.7638353705406189, 'learning_rate': 3.1231231231231234e-06, 'epoch': 0.09}

  3%|▎         | 416/13317 [04:16<2:12:05,  1.63it/s]
  3%|▎         | 417/13317 [04:17<2:12:03,  1.63it/s]
                                                     
{'loss': 0.3615, 'grad_norm': 0.6620998978614807, 'learning_rate': 3.130630630630631e-06, 'epoch': 0.09}

  3%|▎         | 417/13317 [04:17<2:12:03,  1.63it/s]
  3%|▎         | 418/13317 [04:17<2:12:03,  1.63it/s]
                                                     
{'loss': 0.5016, 'grad_norm': 0.7566660642623901, 'learning_rate': 3.138138138138138e-06, 'epoch': 0.09}

  3%|▎         | 418/13317 [04:17<2:12:03,  1.63it/s]
  3%|▎         | 419/13317 [04:18<2:11:57,  1.63it/s]
                                                     
{'loss': 0.3935, 'grad_norm': 0.672184407711029, 'learning_rate': 3.1456456456456457e-06, 'epoch': 0.09}

  3%|▎         | 419/13317 [04:18<2:11:57,  1.63it/s]
  3%|▎         | 420/13317 [04:18<2:11:51,  1.63it/s]
                                                     
{'loss': 0.5348, 'grad_norm': 0.797938346862793, 'learning_rate': 3.1531531531531532e-06, 'epoch': 0.09}

  3%|▎         | 420/13317 [04:19<2:11:51,  1.63it/s]
  3%|▎         | 421/13317 [04:19<2:12:04,  1.63it/s]
                                                     
{'loss': 0.5303, 'grad_norm': 0.8910072445869446, 'learning_rate': 3.160660660660661e-06, 'epoch': 0.09}

  3%|▎         | 421/13317 [04:19<2:12:04,  1.63it/s]
  3%|▎         | 422/13317 [04:20<2:11:55,  1.63it/s]
                                                     
{'loss': 0.4015, 'grad_norm': 0.6455042362213135, 'learning_rate': 3.168168168168169e-06, 'epoch': 0.1}

  3%|▎         | 422/13317 [04:20<2:11:55,  1.63it/s]
  3%|▎         | 423/13317 [04:20<2:11:52,  1.63it/s]
                                                     
{'loss': 0.5236, 'grad_norm': 0.6573291420936584, 'learning_rate': 3.1756756756756755e-06, 'epoch': 0.1}

  3%|▎         | 423/13317 [04:20<2:11:52,  1.63it/s]
  3%|▎         | 424/13317 [04:21<2:11:49,  1.63it/s]
                                                     
{'loss': 0.4813, 'grad_norm': 0.7061600089073181, 'learning_rate': 3.183183183183183e-06, 'epoch': 0.1}

  3%|▎         | 424/13317 [04:21<2:11:49,  1.63it/s]
  3%|▎         | 425/13317 [04:21<2:11:48,  1.63it/s]
                                                     
{'loss': 0.4781, 'grad_norm': 0.706832230091095, 'learning_rate': 3.190690690690691e-06, 'epoch': 0.1}

  3%|▎         | 425/13317 [04:22<2:11:48,  1.63it/s]
  3%|▎         | 426/13317 [04:22<2:11:54,  1.63it/s]
                                                     
{'loss': 0.4983, 'grad_norm': 0.7267263531684875, 'learning_rate': 3.1981981981981987e-06, 'epoch': 0.1}

  3%|▎         | 426/13317 [04:22<2:11:54,  1.63it/s]
  3%|▎         | 427/13317 [04:23<2:11:51,  1.63it/s]
                                                     
{'loss': 0.4213, 'grad_norm': 0.6855952739715576, 'learning_rate': 3.2057057057057063e-06, 'epoch': 0.1}

  3%|▎         | 427/13317 [04:23<2:11:51,  1.63it/s]
  3%|▎         | 428/13317 [04:23<2:11:52,  1.63it/s]
                                                     
{'loss': 0.4775, 'grad_norm': 0.6827059388160706, 'learning_rate': 3.2132132132132134e-06, 'epoch': 0.1}

  3%|▎         | 428/13317 [04:23<2:11:52,  1.63it/s]
  3%|▎         | 429/13317 [04:24<2:11:49,  1.63it/s]
                                                     
{'loss': 0.4742, 'grad_norm': 0.6486400365829468, 'learning_rate': 3.220720720720721e-06, 'epoch': 0.1}

  3%|▎         | 429/13317 [04:24<2:11:49,  1.63it/s]
  3%|▎         | 430/13317 [04:25<2:11:46,  1.63it/s]
                                                     
{'loss': 0.458, 'grad_norm': 0.7116115689277649, 'learning_rate': 3.2282282282282286e-06, 'epoch': 0.1}

  3%|▎         | 430/13317 [04:25<2:11:46,  1.63it/s]
  3%|▎         | 431/13317 [04:25<2:11:57,  1.63it/s]
                                                     
{'loss': 0.5099, 'grad_norm': 0.6915836334228516, 'learning_rate': 3.235735735735736e-06, 'epoch': 0.1}

  3%|▎         | 431/13317 [04:25<2:11:57,  1.63it/s]
  3%|▎         | 432/13317 [04:26<2:11:51,  1.63it/s]
                                                     
{'loss': 0.459, 'grad_norm': 0.7435068488121033, 'learning_rate': 3.2432432432432437e-06, 'epoch': 0.1}

  3%|▎         | 432/13317 [04:26<2:11:51,  1.63it/s]
  3%|▎         | 433/13317 [04:26<2:11:49,  1.63it/s]
                                                     
{'loss': 0.438, 'grad_norm': 0.6927868127822876, 'learning_rate': 3.250750750750751e-06, 'epoch': 0.1}

  3%|▎         | 433/13317 [04:27<2:11:49,  1.63it/s]
  3%|▎         | 434/13317 [04:27<2:11:51,  1.63it/s]
                                                     
{'loss': 0.408, 'grad_norm': 0.7656463980674744, 'learning_rate': 3.2582582582582584e-06, 'epoch': 0.1}

  3%|▎         | 434/13317 [04:27<2:11:51,  1.63it/s]
  3%|▎         | 435/13317 [04:28<2:11:45,  1.63it/s]
                                                     
{'loss': 0.4215, 'grad_norm': 0.6323496103286743, 'learning_rate': 3.265765765765766e-06, 'epoch': 0.1}

  3%|▎         | 435/13317 [04:28<2:11:45,  1.63it/s]
  3%|▎         | 436/13317 [04:28<2:11:53,  1.63it/s]
                                                     
{'loss': 0.4551, 'grad_norm': 0.7722992300987244, 'learning_rate': 3.2732732732732736e-06, 'epoch': 0.1}

  3%|▎         | 436/13317 [04:28<2:11:53,  1.63it/s]
  3%|▎         | 437/13317 [04:29<2:11:49,  1.63it/s]
                                                     
{'loss': 0.4665, 'grad_norm': 0.8740445375442505, 'learning_rate': 3.280780780780781e-06, 'epoch': 0.1}

  3%|▎         | 437/13317 [04:29<2:11:49,  1.63it/s]
  3%|▎         | 438/13317 [04:29<2:11:44,  1.63it/s]
                                                     
{'loss': 0.3806, 'grad_norm': 0.579095184803009, 'learning_rate': 3.2882882882882887e-06, 'epoch': 0.1}

  3%|▎         | 438/13317 [04:30<2:11:44,  1.63it/s]
  3%|▎         | 439/13317 [04:30<2:11:45,  1.63it/s]
                                                     
{'loss': 0.4427, 'grad_norm': 0.6989012360572815, 'learning_rate': 3.295795795795796e-06, 'epoch': 0.1}

  3%|▎         | 439/13317 [04:30<2:11:45,  1.63it/s]
  3%|▎         | 440/13317 [04:31<2:11:40,  1.63it/s]
                                                     
{'loss': 0.4532, 'grad_norm': 0.6261010766029358, 'learning_rate': 3.3033033033033035e-06, 'epoch': 0.1}

  3%|▎         | 440/13317 [04:31<2:11:40,  1.63it/s]
  3%|▎         | 441/13317 [04:31<2:11:46,  1.63it/s]
                                                     
{'loss': 0.4531, 'grad_norm': 0.8126707673072815, 'learning_rate': 3.310810810810811e-06, 'epoch': 0.1}

  3%|▎         | 441/13317 [04:31<2:11:46,  1.63it/s]
  3%|▎         | 442/13317 [04:32<2:11:39,  1.63it/s]
                                                     
{'loss': 0.4213, 'grad_norm': 0.762205183506012, 'learning_rate': 3.3183183183183186e-06, 'epoch': 0.1}

  3%|▎         | 442/13317 [04:32<2:11:39,  1.63it/s]
  3%|▎         | 443/13317 [04:33<2:11:40,  1.63it/s]
                                                     
{'loss': 0.4377, 'grad_norm': 0.6316590905189514, 'learning_rate': 3.325825825825826e-06, 'epoch': 0.1}

  3%|▎         | 443/13317 [04:33<2:11:40,  1.63it/s]
  3%|▎         | 444/13317 [04:33<2:11:43,  1.63it/s]
                                                     
{'loss': 0.4483, 'grad_norm': 0.7545338869094849, 'learning_rate': 3.3333333333333333e-06, 'epoch': 0.1}

  3%|▎         | 444/13317 [04:33<2:11:43,  1.63it/s]
  3%|▎         | 445/13317 [04:34<2:11:38,  1.63it/s]
                                                     
{'loss': 0.4915, 'grad_norm': 0.7908391356468201, 'learning_rate': 3.340840840840841e-06, 'epoch': 0.1}

  3%|▎         | 445/13317 [04:34<2:11:38,  1.63it/s]
  3%|▎         | 446/13317 [04:34<2:11:55,  1.63it/s]
                                                     
{'loss': 0.6086, 'grad_norm': 0.7524520754814148, 'learning_rate': 3.3483483483483485e-06, 'epoch': 0.1}

  3%|▎         | 446/13317 [04:35<2:11:55,  1.63it/s]
  3%|▎         | 447/13317 [04:35<2:11:46,  1.63it/s]
                                                     
{'loss': 0.4132, 'grad_norm': 0.6261481046676636, 'learning_rate': 3.3558558558558565e-06, 'epoch': 0.1}

  3%|▎         | 447/13317 [04:35<2:11:46,  1.63it/s]
  3%|▎         | 448/13317 [04:36<2:11:41,  1.63it/s]
                                                     
{'loss': 0.4198, 'grad_norm': 0.7558276653289795, 'learning_rate': 3.363363363363364e-06, 'epoch': 0.1}

  3%|▎         | 448/13317 [04:36<2:11:41,  1.63it/s]
  3%|▎         | 449/13317 [04:36<2:11:43,  1.63it/s]
                                                     
{'loss': 0.421, 'grad_norm': 0.6811279654502869, 'learning_rate': 3.3708708708708708e-06, 'epoch': 0.1}

  3%|▎         | 449/13317 [04:36<2:11:43,  1.63it/s]
  3%|▎         | 450/13317 [04:37<2:11:40,  1.63it/s]
                                                     
{'loss': 0.4403, 'grad_norm': 0.704143226146698, 'learning_rate': 3.3783783783783788e-06, 'epoch': 0.1}

  3%|▎         | 450/13317 [04:37<2:11:40,  1.63it/s]
  3%|▎         | 451/13317 [04:37<2:11:48,  1.63it/s]
                                                     
{'loss': 0.3303, 'grad_norm': 0.6244333982467651, 'learning_rate': 3.3858858858858863e-06, 'epoch': 0.1}

  3%|▎         | 451/13317 [04:38<2:11:48,  1.63it/s]
  3%|▎         | 452/13317 [04:38<2:11:42,  1.63it/s]
                                                     
{'loss': 0.5078, 'grad_norm': 0.6552481055259705, 'learning_rate': 3.393393393393394e-06, 'epoch': 0.1}

  3%|▎         | 452/13317 [04:38<2:11:42,  1.63it/s]
  3%|▎         | 453/13317 [04:39<2:11:37,  1.63it/s]
                                                     
{'loss': 0.4777, 'grad_norm': 0.723028302192688, 'learning_rate': 3.4009009009009015e-06, 'epoch': 0.1}

  3%|▎         | 453/13317 [04:39<2:11:37,  1.63it/s]
  3%|▎         | 454/13317 [04:39<2:11:34,  1.63it/s]
                                                     
{'loss': 0.4813, 'grad_norm': 0.6408616304397583, 'learning_rate': 3.4084084084084086e-06, 'epoch': 0.1}

  3%|▎         | 454/13317 [04:39<2:11:34,  1.63it/s]
  3%|▎         | 455/13317 [04:40<2:11:31,  1.63it/s]
                                                     
{'loss': 0.4815, 'grad_norm': 0.7385659217834473, 'learning_rate': 3.4159159159159162e-06, 'epoch': 0.1}

  3%|▎         | 455/13317 [04:40<2:11:31,  1.63it/s]
  3%|▎         | 456/13317 [04:41<2:11:38,  1.63it/s]
                                                     
{'loss': 0.4326, 'grad_norm': 0.6278891563415527, 'learning_rate': 3.423423423423424e-06, 'epoch': 0.1}

  3%|▎         | 456/13317 [04:41<2:11:38,  1.63it/s]
  3%|▎         | 457/13317 [04:41<2:11:35,  1.63it/s]
                                                     
{'loss': 0.3893, 'grad_norm': 0.5875856280326843, 'learning_rate': 3.4309309309309314e-06, 'epoch': 0.1}

  3%|▎         | 457/13317 [04:41<2:11:35,  1.63it/s]
  3%|▎         | 458/13317 [04:42<2:11:36,  1.63it/s]
                                                     
{'loss': 0.3643, 'grad_norm': 0.67644864320755, 'learning_rate': 3.438438438438439e-06, 'epoch': 0.1}

  3%|▎         | 458/13317 [04:42<2:11:36,  1.63it/s]
  3%|▎         | 459/13317 [04:42<2:11:33,  1.63it/s]
                                                     
{'loss': 0.3107, 'grad_norm': 0.5073651671409607, 'learning_rate': 3.445945945945946e-06, 'epoch': 0.1}

  3%|▎         | 459/13317 [04:43<2:11:33,  1.63it/s]
  3%|▎         | 460/13317 [04:43<2:11:32,  1.63it/s]
                                                     
{'loss': 0.541, 'grad_norm': 0.81984943151474, 'learning_rate': 3.4534534534534537e-06, 'epoch': 0.1}

  3%|▎         | 460/13317 [04:43<2:11:32,  1.63it/s]
  3%|▎         | 461/13317 [04:44<2:11:43,  1.63it/s]
                                                     
{'loss': 0.5197, 'grad_norm': 0.8332152366638184, 'learning_rate': 3.4609609609609612e-06, 'epoch': 0.1}

  3%|▎         | 461/13317 [04:44<2:11:43,  1.63it/s]
  3%|▎         | 462/13317 [04:44<2:11:43,  1.63it/s]
                                                     
{'loss': 0.5032, 'grad_norm': 0.7231094837188721, 'learning_rate': 3.468468468468469e-06, 'epoch': 0.1}

  3%|▎         | 462/13317 [04:44<2:11:43,  1.63it/s]
  3%|▎         | 463/13317 [04:45<2:11:36,  1.63it/s]
                                                     
{'loss': 0.4249, 'grad_norm': 0.6658948659896851, 'learning_rate': 3.4759759759759764e-06, 'epoch': 0.1}

  3%|▎         | 463/13317 [04:45<2:11:36,  1.63it/s]
  3%|▎         | 464/13317 [04:45<2:11:32,  1.63it/s]
                                                     
{'loss': 0.5033, 'grad_norm': 0.7102115750312805, 'learning_rate': 3.4834834834834835e-06, 'epoch': 0.1}

  3%|▎         | 464/13317 [04:46<2:11:32,  1.63it/s]
  3%|▎         | 465/13317 [04:46<2:11:32,  1.63it/s]
                                                     
{'loss': 0.512, 'grad_norm': 0.8554460406303406, 'learning_rate': 3.490990990990991e-06, 'epoch': 0.1}

  3%|▎         | 465/13317 [04:46<2:11:32,  1.63it/s]
  3%|▎         | 466/13317 [04:47<2:11:36,  1.63it/s]
                                                     
{'loss': 0.4825, 'grad_norm': 0.6698629856109619, 'learning_rate': 3.4984984984984987e-06, 'epoch': 0.1}

  3%|▎         | 466/13317 [04:47<2:11:36,  1.63it/s]
  4%|▎         | 467/13317 [04:47<2:11:31,  1.63it/s]
                                                     
{'loss': 0.5179, 'grad_norm': 0.8527231216430664, 'learning_rate': 3.5060060060060063e-06, 'epoch': 0.11}

  4%|▎         | 467/13317 [04:47<2:11:31,  1.63it/s]
  4%|▎         | 468/13317 [04:48<2:11:29,  1.63it/s]
                                                     
{'loss': 0.5558, 'grad_norm': 0.8860906958580017, 'learning_rate': 3.513513513513514e-06, 'epoch': 0.11}

  4%|▎         | 468/13317 [04:48<2:11:29,  1.63it/s]
  4%|▎         | 469/13317 [04:49<2:11:25,  1.63it/s]
                                                     
{'loss': 0.4335, 'grad_norm': 0.7079217433929443, 'learning_rate': 3.521021021021021e-06, 'epoch': 0.11}

  4%|▎         | 469/13317 [04:49<2:11:25,  1.63it/s]
  4%|▎         | 470/13317 [04:49<2:11:25,  1.63it/s]
                                                     
{'loss': 0.5229, 'grad_norm': 0.7085310220718384, 'learning_rate': 3.5285285285285286e-06, 'epoch': 0.11}

  4%|▎         | 470/13317 [04:49<2:11:25,  1.63it/s]
  4%|▎         | 471/13317 [04:50<2:11:28,  1.63it/s]
                                                     
{'loss': 0.4051, 'grad_norm': 0.765766441822052, 'learning_rate': 3.536036036036036e-06, 'epoch': 0.11}

  4%|▎         | 471/13317 [04:50<2:11:28,  1.63it/s]
  4%|▎         | 472/13317 [04:50<2:11:23,  1.63it/s]
                                                     
{'loss': 0.4148, 'grad_norm': 0.7836328148841858, 'learning_rate': 3.5435435435435437e-06, 'epoch': 0.11}

  4%|▎         | 472/13317 [04:50<2:11:23,  1.63it/s]
  4%|▎         | 473/13317 [04:51<2:11:20,  1.63it/s]
                                                     
{'loss': 0.4018, 'grad_norm': 0.6780300140380859, 'learning_rate': 3.5510510510510517e-06, 'epoch': 0.11}

  4%|▎         | 473/13317 [04:51<2:11:20,  1.63it/s]
  4%|▎         | 474/13317 [04:52<2:11:25,  1.63it/s]
                                                     
{'loss': 0.4146, 'grad_norm': 0.6662124991416931, 'learning_rate': 3.5585585585585584e-06, 'epoch': 0.11}

  4%|▎         | 474/13317 [04:52<2:11:25,  1.63it/s]
  4%|▎         | 475/13317 [04:52<2:11:21,  1.63it/s]
                                                     
{'loss': 0.3877, 'grad_norm': 0.6870988607406616, 'learning_rate': 3.566066066066066e-06, 'epoch': 0.11}

  4%|▎         | 475/13317 [04:52<2:11:21,  1.63it/s]
  4%|▎         | 476/13317 [04:53<2:11:26,  1.63it/s]
                                                     
{'loss': 0.3708, 'grad_norm': 0.6329861879348755, 'learning_rate': 3.573573573573574e-06, 'epoch': 0.11}

  4%|▎         | 476/13317 [04:53<2:11:26,  1.63it/s]
  4%|▎         | 477/13317 [04:53<2:11:24,  1.63it/s]
                                                     
{'loss': 0.419, 'grad_norm': 0.7007266879081726, 'learning_rate': 3.5810810810810816e-06, 'epoch': 0.11}

  4%|▎         | 477/13317 [04:54<2:11:24,  1.63it/s]
  4%|▎         | 478/13317 [04:54<2:11:24,  1.63it/s]
                                                     
{'loss': 0.466, 'grad_norm': 0.7247040271759033, 'learning_rate': 3.588588588588589e-06, 'epoch': 0.11}

  4%|▎         | 478/13317 [04:54<2:11:24,  1.63it/s]
  4%|▎         | 479/13317 [04:55<2:11:16,  1.63it/s]
                                                     
{'loss': 0.4634, 'grad_norm': 0.7045004367828369, 'learning_rate': 3.5960960960960967e-06, 'epoch': 0.11}

  4%|▎         | 479/13317 [04:55<2:11:16,  1.63it/s]
  4%|▎         | 480/13317 [04:55<2:11:17,  1.63it/s]
                                                     
{'loss': 0.4946, 'grad_norm': 0.793982982635498, 'learning_rate': 3.603603603603604e-06, 'epoch': 0.11}

  4%|▎         | 480/13317 [04:55<2:11:17,  1.63it/s]
  4%|▎         | 481/13317 [04:56<2:11:25,  1.63it/s]
                                                     
{'loss': 0.3892, 'grad_norm': 0.6738253235816956, 'learning_rate': 3.6111111111111115e-06, 'epoch': 0.11}

  4%|▎         | 481/13317 [04:56<2:11:25,  1.63it/s]
  4%|▎         | 482/13317 [04:56<2:11:22,  1.63it/s]
                                                     
{'loss': 0.3949, 'grad_norm': 0.6139069199562073, 'learning_rate': 3.618618618618619e-06, 'epoch': 0.11}

  4%|▎         | 482/13317 [04:57<2:11:22,  1.63it/s]
  4%|▎         | 483/13317 [04:57<2:11:17,  1.63it/s]
                                                     
{'loss': 0.4763, 'grad_norm': 0.8812891244888306, 'learning_rate': 3.6261261261261266e-06, 'epoch': 0.11}

  4%|▎         | 483/13317 [04:57<2:11:17,  1.63it/s]
  4%|▎         | 484/13317 [04:58<2:11:16,  1.63it/s]
                                                     
{'loss': 0.4176, 'grad_norm': 0.658255934715271, 'learning_rate': 3.633633633633634e-06, 'epoch': 0.11}

  4%|▎         | 484/13317 [04:58<2:11:16,  1.63it/s]
  4%|▎         | 485/13317 [04:58<2:11:17,  1.63it/s]
                                                     
{'loss': 0.3784, 'grad_norm': 0.5892598628997803, 'learning_rate': 3.6411411411411413e-06, 'epoch': 0.11}

  4%|▎         | 485/13317 [04:58<2:11:17,  1.63it/s]
  4%|▎         | 486/13317 [04:59<2:11:54,  1.62it/s]
                                                     
{'loss': 0.3932, 'grad_norm': 0.6718635559082031, 'learning_rate': 3.648648648648649e-06, 'epoch': 0.11}

  4%|▎         | 486/13317 [04:59<2:11:54,  1.62it/s]
  4%|▎         | 487/13317 [05:00<2:11:45,  1.62it/s]
                                                     
{'loss': 0.4165, 'grad_norm': 0.7494131922721863, 'learning_rate': 3.6561561561561565e-06, 'epoch': 0.11}

  4%|▎         | 487/13317 [05:00<2:11:45,  1.62it/s]
  4%|▎         | 488/13317 [05:00<2:11:37,  1.62it/s]
                                                     
{'loss': 0.4688, 'grad_norm': 0.6745133996009827, 'learning_rate': 3.663663663663664e-06, 'epoch': 0.11}

  4%|▎         | 488/13317 [05:00<2:11:37,  1.62it/s]
  4%|▎         | 489/13317 [05:01<2:11:30,  1.63it/s]
                                                     
{'loss': 0.4248, 'grad_norm': 0.698136568069458, 'learning_rate': 3.6711711711711716e-06, 'epoch': 0.11}

  4%|▎         | 489/13317 [05:01<2:11:30,  1.63it/s]
  4%|▎         | 490/13317 [05:01<2:11:27,  1.63it/s]
                                                     
{'loss': 0.4287, 'grad_norm': 0.6066250205039978, 'learning_rate': 3.6786786786786788e-06, 'epoch': 0.11}

  4%|▎         | 490/13317 [05:02<2:11:27,  1.63it/s]
  4%|▎         | 491/13317 [05:02<2:11:34,  1.62it/s]
                                                     
{'loss': 0.4886, 'grad_norm': 0.7168146967887878, 'learning_rate': 3.6861861861861863e-06, 'epoch': 0.11}

  4%|▎         | 491/13317 [05:02<2:11:34,  1.62it/s]
  4%|▎         | 492/13317 [05:03<2:11:22,  1.63it/s]
                                                     
{'loss': 0.4494, 'grad_norm': 0.7278876304626465, 'learning_rate': 3.693693693693694e-06, 'epoch': 0.11}

  4%|▎         | 492/13317 [05:03<2:11:22,  1.63it/s]
  4%|▎         | 493/13317 [05:03<2:11:29,  1.63it/s]
                                                     
{'loss': 0.3935, 'grad_norm': 0.7472261786460876, 'learning_rate': 3.7012012012012015e-06, 'epoch': 0.11}

  4%|▎         | 493/13317 [05:03<2:11:29,  1.63it/s]
  4%|▎         | 494/13317 [05:04<2:11:25,  1.63it/s]
                                                     
{'loss': 0.4268, 'grad_norm': 0.6780577301979065, 'learning_rate': 3.708708708708709e-06, 'epoch': 0.11}

  4%|▎         | 494/13317 [05:04<2:11:25,  1.63it/s]
  4%|▎         | 495/13317 [05:04<2:11:19,  1.63it/s]
                                                     
{'loss': 0.4721, 'grad_norm': 0.7228316068649292, 'learning_rate': 3.7162162162162162e-06, 'epoch': 0.11}

  4%|▎         | 495/13317 [05:05<2:11:19,  1.63it/s]
  4%|▎         | 496/13317 [05:05<2:11:17,  1.63it/s]
                                                     
{'loss': 0.3736, 'grad_norm': 0.65328449010849, 'learning_rate': 3.723723723723724e-06, 'epoch': 0.11}

  4%|▎         | 496/13317 [05:05<2:11:17,  1.63it/s]
  4%|▎         | 497/13317 [05:06<2:11:13,  1.63it/s]
                                                     
{'loss': 0.4561, 'grad_norm': 0.7401781678199768, 'learning_rate': 3.7312312312312314e-06, 'epoch': 0.11}

  4%|▎         | 497/13317 [05:06<2:11:13,  1.63it/s]
  4%|▎         | 498/13317 [05:06<2:11:23,  1.63it/s]
                                                     
{'loss': 0.3764, 'grad_norm': 0.6361309289932251, 'learning_rate': 3.7387387387387394e-06, 'epoch': 0.11}

  4%|▎         | 498/13317 [05:06<2:11:23,  1.63it/s]
  4%|▎         | 499/13317 [05:07<2:11:20,  1.63it/s]
                                                     
{'loss': 0.4862, 'grad_norm': 0.6940679550170898, 'learning_rate': 3.746246246246247e-06, 'epoch': 0.11}

  4%|▎         | 499/13317 [05:07<2:11:20,  1.63it/s]
  4%|▍         | 500/13317 [05:08<2:11:14,  1.63it/s]
                                                     
{'loss': 0.4615, 'grad_norm': 0.7067543864250183, 'learning_rate': 3.7537537537537537e-06, 'epoch': 0.11}

  4%|▍         | 500/13317 [05:08<2:11:14,  1.63it/s]

  0%|          | 0/13 [00:00<?, ?it/s]

 15%|█▌        | 2/13 [00:00<00:00, 13.35it/s]

 31%|███       | 4/13 [00:00<00:01,  8.40it/s]

 38%|███▊      | 5/13 [00:00<00:01,  7.80it/s]

 46%|████▌     | 6/13 [00:00<00:00,  7.43it/s]

 54%|█████▍    | 7/13 [00:00<00:00,  7.18it/s]

 62%|██████▏   | 8/13 [00:01<00:00,  7.02it/s]

 69%|██████▉   | 9/13 [00:01<00:00,  6.91it/s]

 77%|███████▋  | 10/13 [00:01<00:00,  6.84it/s]

 85%|████████▍ | 11/13 [00:01<00:00,  6.79it/s]

 92%|█████████▏| 12/13 [00:01<00:00,  6.75it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]
                                                     

                                               
{'eval_loss': 0.852842390537262, 'eval_runtime': 1.9519, 'eval_samples_per_second': 102.463, 'eval_steps_per_second': 6.66, 'epoch': 0.11}

  4%|▍         | 500/13317 [05:10<2:11:14,  1.63it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]

                                               
  4%|▍         | 501/13317 [05:10<4:16:32,  1.20s/it]
                                                     
{'loss': 0.4481, 'grad_norm': 0.8868950009346008, 'learning_rate': 3.7612612612612612e-06, 'epoch': 0.11}

  4%|▍         | 501/13317 [05:10<4:16:32,  1.20s/it]
  4%|▍         | 502/13317 [05:11<3:38:55,  1.02s/it]
                                                     
{'loss': 0.3702, 'grad_norm': 0.6206368803977966, 'learning_rate': 3.7687687687687692e-06, 'epoch': 0.11}

  4%|▍         | 502/13317 [05:11<3:38:55,  1.02s/it]
  4%|▍         | 503/13317 [05:11<3:12:35,  1.11it/s]
                                                     
{'loss': 0.4287, 'grad_norm': 0.6510940790176392, 'learning_rate': 3.776276276276277e-06, 'epoch': 0.11}

  4%|▍         | 503/13317 [05:11<3:12:35,  1.11it/s]
  4%|▍         | 504/13317 [05:12<2:54:05,  1.23it/s]
                                                     
{'loss': 0.3768, 'grad_norm': 0.6568686962127686, 'learning_rate': 3.7837837837837844e-06, 'epoch': 0.11}

  4%|▍         | 504/13317 [05:12<2:54:05,  1.23it/s]
  4%|▍         | 505/13317 [05:13<2:41:11,  1.32it/s]
                                                     
{'loss': 0.4612, 'grad_norm': 0.7525666952133179, 'learning_rate': 3.7912912912912915e-06, 'epoch': 0.11}

  4%|▍         | 505/13317 [05:13<2:41:11,  1.32it/s]
  4%|▍         | 506/13317 [05:13<2:32:08,  1.40it/s]
                                                     
{'loss': 0.3992, 'grad_norm': 0.6893332600593567, 'learning_rate': 3.798798798798799e-06, 'epoch': 0.11}

  4%|▍         | 506/13317 [05:13<2:32:08,  1.40it/s]
  4%|▍         | 507/13317 [05:14<2:25:42,  1.47it/s]
                                                     
{'loss': 0.4962, 'grad_norm': 0.7228710055351257, 'learning_rate': 3.8063063063063067e-06, 'epoch': 0.11}

  4%|▍         | 507/13317 [05:14<2:25:42,  1.47it/s]
  4%|▍         | 508/13317 [05:14<2:21:21,  1.51it/s]
                                                     
{'loss': 0.4447, 'grad_norm': 0.6795083284378052, 'learning_rate': 3.8138138138138143e-06, 'epoch': 0.11}

  4%|▍         | 508/13317 [05:15<2:21:21,  1.51it/s]
  4%|▍         | 509/13317 [05:15<2:18:11,  1.54it/s]
                                                     
{'loss': 0.412, 'grad_norm': 0.6653012633323669, 'learning_rate': 3.821321321321322e-06, 'epoch': 0.11}

  4%|▍         | 509/13317 [05:15<2:18:11,  1.54it/s]
  4%|▍         | 510/13317 [05:16<2:16:01,  1.57it/s]
                                                     
{'loss': 0.5315, 'grad_norm': 0.721488893032074, 'learning_rate': 3.828828828828829e-06, 'epoch': 0.11}

  4%|▍         | 510/13317 [05:16<2:16:01,  1.57it/s]
  4%|▍         | 511/13317 [05:16<2:14:29,  1.59it/s]
                                                     
{'loss': 0.5047, 'grad_norm': 0.673145055770874, 'learning_rate': 3.836336336336336e-06, 'epoch': 0.12}

  4%|▍         | 511/13317 [05:16<2:14:29,  1.59it/s]
  4%|▍         | 512/13317 [05:17<2:13:26,  1.60it/s]
                                                     
{'loss': 0.407, 'grad_norm': 0.7387295365333557, 'learning_rate': 3.843843843843844e-06, 'epoch': 0.12}

  4%|▍         | 512/13317 [05:17<2:13:26,  1.60it/s]
  4%|▍         | 513/13317 [05:18<2:12:49,  1.61it/s]
                                                     
{'loss': 0.3565, 'grad_norm': 0.6859047412872314, 'learning_rate': 3.851351351351352e-06, 'epoch': 0.12}

  4%|▍         | 513/13317 [05:18<2:12:49,  1.61it/s]
  4%|▍         | 514/13317 [05:18<2:12:19,  1.61it/s]
                                                     
{'loss': 0.3785, 'grad_norm': 0.6660869121551514, 'learning_rate': 3.858858858858859e-06, 'epoch': 0.12}

  4%|▍         | 514/13317 [05:18<2:12:19,  1.61it/s]
  4%|▍         | 515/13317 [05:19<2:11:56,  1.62it/s]
                                                     
{'loss': 0.425, 'grad_norm': 0.7481129765510559, 'learning_rate': 3.8663663663663664e-06, 'epoch': 0.12}

  4%|▍         | 515/13317 [05:19<2:11:56,  1.62it/s]
  4%|▍         | 516/13317 [05:19<2:11:33,  1.62it/s]
                                                     
{'loss': 0.4433, 'grad_norm': 0.668999195098877, 'learning_rate': 3.8738738738738744e-06, 'epoch': 0.12}

  4%|▍         | 516/13317 [05:19<2:11:33,  1.62it/s]
  4%|▍         | 517/13317 [05:20<2:11:21,  1.62it/s]
                                                     
{'loss': 0.4276, 'grad_norm': 0.841803252696991, 'learning_rate': 3.881381381381382e-06, 'epoch': 0.12}

  4%|▍         | 517/13317 [05:20<2:11:21,  1.62it/s]
  4%|▍         | 518/13317 [05:21<2:11:16,  1.62it/s]
                                                     
{'loss': 0.4293, 'grad_norm': 0.7674733400344849, 'learning_rate': 3.88888888888889e-06, 'epoch': 0.12}

  4%|▍         | 518/13317 [05:21<2:11:16,  1.62it/s]
  4%|▍         | 519/13317 [05:21<2:11:04,  1.63it/s]
                                                     
{'loss': 0.4566, 'grad_norm': 0.7696067094802856, 'learning_rate': 3.896396396396397e-06, 'epoch': 0.12}

  4%|▍         | 519/13317 [05:21<2:11:04,  1.63it/s]
  4%|▍         | 520/13317 [05:22<2:10:56,  1.63it/s]
                                                     
{'loss': 0.5686, 'grad_norm': 0.8306894898414612, 'learning_rate': 3.903903903903904e-06, 'epoch': 0.12}

  4%|▍         | 520/13317 [05:22<2:10:56,  1.63it/s]
  4%|▍         | 521/13317 [05:22<2:10:55,  1.63it/s]
                                                     
{'loss': 0.3467, 'grad_norm': 0.6511797308921814, 'learning_rate': 3.911411411411412e-06, 'epoch': 0.12}

  4%|▍         | 521/13317 [05:23<2:10:55,  1.63it/s]
  4%|▍         | 522/13317 [05:23<2:10:51,  1.63it/s]
                                                     
{'loss': 0.5382, 'grad_norm': 0.7276720404624939, 'learning_rate': 3.918918918918919e-06, 'epoch': 0.12}

  4%|▍         | 522/13317 [05:23<2:10:51,  1.63it/s]
  4%|▍         | 523/13317 [05:24<2:10:50,  1.63it/s]
                                                     
{'loss': 0.4015, 'grad_norm': 0.700407087802887, 'learning_rate': 3.926426426426427e-06, 'epoch': 0.12}

  4%|▍         | 523/13317 [05:24<2:10:50,  1.63it/s]
  4%|▍         | 524/13317 [05:24<2:10:47,  1.63it/s]
                                                     
{'loss': 0.383, 'grad_norm': 0.7672831416130066, 'learning_rate': 3.933933933933934e-06, 'epoch': 0.12}

  4%|▍         | 524/13317 [05:24<2:10:47,  1.63it/s]
  4%|▍         | 525/13317 [05:25<2:10:44,  1.63it/s]
                                                     
{'loss': 0.4383, 'grad_norm': 0.7947280406951904, 'learning_rate': 3.941441441441442e-06, 'epoch': 0.12}

  4%|▍         | 525/13317 [05:25<2:10:44,  1.63it/s]
  4%|▍         | 526/13317 [05:25<2:10:47,  1.63it/s]
                                                     
{'loss': 0.4673, 'grad_norm': 0.7911531925201416, 'learning_rate': 3.948948948948949e-06, 'epoch': 0.12}

  4%|▍         | 526/13317 [05:26<2:10:47,  1.63it/s]
  4%|▍         | 527/13317 [05:26<2:10:51,  1.63it/s]
                                                     
{'loss': 0.374, 'grad_norm': 0.7088866233825684, 'learning_rate': 3.9564564564564565e-06, 'epoch': 0.12}

  4%|▍         | 527/13317 [05:26<2:10:51,  1.63it/s]
  4%|▍         | 528/13317 [05:27<2:10:51,  1.63it/s]
                                                     
{'loss': 0.4959, 'grad_norm': 0.7523390650749207, 'learning_rate': 3.9639639639639645e-06, 'epoch': 0.12}

  4%|▍         | 528/13317 [05:27<2:10:51,  1.63it/s]
  4%|▍         | 529/13317 [05:27<2:10:53,  1.63it/s]
                                                     
{'loss': 0.3291, 'grad_norm': 0.5942575931549072, 'learning_rate': 3.971471471471472e-06, 'epoch': 0.12}

  4%|▍         | 529/13317 [05:27<2:10:53,  1.63it/s]
  4%|▍         | 530/13317 [05:28<2:10:49,  1.63it/s]
                                                     
{'loss': 0.3817, 'grad_norm': 0.7468729019165039, 'learning_rate': 3.97897897897898e-06, 'epoch': 0.12}

  4%|▍         | 530/13317 [05:28<2:10:49,  1.63it/s]
  4%|▍         | 531/13317 [05:29<2:10:53,  1.63it/s]
                                                     
{'loss': 0.4637, 'grad_norm': 0.8114937543869019, 'learning_rate': 3.986486486486487e-06, 'epoch': 0.12}

  4%|▍         | 531/13317 [05:29<2:10:53,  1.63it/s]
  4%|▍         | 532/13317 [05:29<2:10:46,  1.63it/s]
                                                     
{'loss': 0.4076, 'grad_norm': 0.6661086082458496, 'learning_rate': 3.993993993993994e-06, 'epoch': 0.12}

  4%|▍         | 532/13317 [05:29<2:10:46,  1.63it/s]
  4%|▍         | 533/13317 [05:30<2:10:49,  1.63it/s]
                                                     
{'loss': 0.3118, 'grad_norm': 0.5750447511672974, 'learning_rate': 4.001501501501502e-06, 'epoch': 0.12}

  4%|▍         | 533/13317 [05:30<2:10:49,  1.63it/s]
  4%|▍         | 534/13317 [05:30<2:10:45,  1.63it/s]
                                                     
{'loss': 0.4776, 'grad_norm': 0.8399438858032227, 'learning_rate': 4.009009009009009e-06, 'epoch': 0.12}

  4%|▍         | 534/13317 [05:31<2:10:45,  1.63it/s]
  4%|▍         | 535/13317 [05:31<2:10:40,  1.63it/s]
                                                     
{'loss': 0.4317, 'grad_norm': 0.8334818482398987, 'learning_rate': 4.016516516516517e-06, 'epoch': 0.12}

  4%|▍         | 535/13317 [05:31<2:10:40,  1.63it/s]
  4%|▍         | 536/13317 [05:32<2:10:44,  1.63it/s]
                                                     
{'loss': 0.5005, 'grad_norm': 0.7687023878097534, 'learning_rate': 4.024024024024024e-06, 'epoch': 0.12}

  4%|▍         | 536/13317 [05:32<2:10:44,  1.63it/s]
  4%|▍         | 537/13317 [05:32<2:10:41,  1.63it/s]
                                                     
{'loss': 0.3564, 'grad_norm': 0.5965541005134583, 'learning_rate': 4.031531531531531e-06, 'epoch': 0.12}

  4%|▍         | 537/13317 [05:32<2:10:41,  1.63it/s]
  4%|▍         | 538/13317 [05:33<2:10:37,  1.63it/s]
                                                     
{'loss': 0.4285, 'grad_norm': 0.7562461495399475, 'learning_rate': 4.039039039039039e-06, 'epoch': 0.12}

  4%|▍         | 538/13317 [05:33<2:10:37,  1.63it/s]
  4%|▍         | 539/13317 [05:33<2:10:38,  1.63it/s]
                                                     
{'loss': 0.4012, 'grad_norm': 0.6406046748161316, 'learning_rate': 4.046546546546547e-06, 'epoch': 0.12}

  4%|▍         | 539/13317 [05:34<2:10:38,  1.63it/s]
  4%|▍         | 540/13317 [05:34<2:10:36,  1.63it/s]
                                                     
{'loss': 0.3135, 'grad_norm': 0.5608196258544922, 'learning_rate': 4.0540540540540545e-06, 'epoch': 0.12}

  4%|▍         | 540/13317 [05:34<2:10:36,  1.63it/s]
  4%|▍         | 541/13317 [05:35<2:10:37,  1.63it/s]
                                                     
{'loss': 0.3511, 'grad_norm': 0.6716158986091614, 'learning_rate': 4.061561561561562e-06, 'epoch': 0.12}

  4%|▍         | 541/13317 [05:35<2:10:37,  1.63it/s]
  4%|▍         | 542/13317 [05:35<2:10:34,  1.63it/s]
                                                     
{'loss': 0.4015, 'grad_norm': 0.7718603014945984, 'learning_rate': 4.06906906906907e-06, 'epoch': 0.12}

  4%|▍         | 542/13317 [05:35<2:10:34,  1.63it/s]
  4%|▍         | 543/13317 [05:36<2:10:37,  1.63it/s]
                                                     
{'loss': 0.4021, 'grad_norm': 0.7509663105010986, 'learning_rate': 4.076576576576577e-06, 'epoch': 0.12}

  4%|▍         | 543/13317 [05:36<2:10:37,  1.63it/s]
  4%|▍         | 544/13317 [05:37<2:10:38,  1.63it/s]
                                                     
{'loss': 0.3918, 'grad_norm': 0.6871640086174011, 'learning_rate': 4.084084084084085e-06, 'epoch': 0.12}

  4%|▍         | 544/13317 [05:37<2:10:38,  1.63it/s]
  4%|▍         | 545/13317 [05:37<2:10:33,  1.63it/s]
                                                     
{'loss': 0.4213, 'grad_norm': 0.7010743618011475, 'learning_rate': 4.091591591591592e-06, 'epoch': 0.12}

  4%|▍         | 545/13317 [05:37<2:10:33,  1.63it/s]
  4%|▍         | 546/13317 [05:38<2:10:32,  1.63it/s]
                                                     
{'loss': 0.4039, 'grad_norm': 0.7625281810760498, 'learning_rate': 4.099099099099099e-06, 'epoch': 0.12}

  4%|▍         | 546/13317 [05:38<2:10:32,  1.63it/s]
  4%|▍         | 547/13317 [05:38<2:10:36,  1.63it/s]
                                                     
{'loss': 0.3555, 'grad_norm': 0.6066325902938843, 'learning_rate': 4.106606606606607e-06, 'epoch': 0.12}

  4%|▍         | 547/13317 [05:38<2:10:36,  1.63it/s]
  4%|▍         | 548/13317 [05:39<2:10:39,  1.63it/s]
                                                     
{'loss': 0.4377, 'grad_norm': 0.718162477016449, 'learning_rate': 4.114114114114114e-06, 'epoch': 0.12}

  4%|▍         | 548/13317 [05:39<2:10:39,  1.63it/s]
  4%|▍         | 549/13317 [05:40<2:10:36,  1.63it/s]
                                                     
{'loss': 0.482, 'grad_norm': 0.6883980631828308, 'learning_rate': 4.121621621621622e-06, 'epoch': 0.12}

  4%|▍         | 549/13317 [05:40<2:10:36,  1.63it/s]
  4%|▍         | 550/13317 [05:40<2:10:32,  1.63it/s]
                                                     
{'loss': 0.3921, 'grad_norm': 0.6029106974601746, 'learning_rate': 4.129129129129129e-06, 'epoch': 0.12}

  4%|▍         | 550/13317 [05:40<2:10:32,  1.63it/s]
  4%|▍         | 551/13317 [05:41<2:10:31,  1.63it/s]
                                                     
{'loss': 0.4079, 'grad_norm': 0.6776401996612549, 'learning_rate': 4.1366366366366366e-06, 'epoch': 0.12}

  4%|▍         | 551/13317 [05:41<2:10:31,  1.63it/s]
  4%|▍         | 552/13317 [05:41<2:10:31,  1.63it/s]
                                                     
{'loss': 0.5462, 'grad_norm': 0.7695199251174927, 'learning_rate': 4.1441441441441446e-06, 'epoch': 0.12}

  4%|▍         | 552/13317 [05:42<2:10:31,  1.63it/s]
  4%|▍         | 553/13317 [05:42<2:10:28,  1.63it/s]
                                                     
{'loss': 0.4149, 'grad_norm': 0.6587623953819275, 'learning_rate': 4.151651651651652e-06, 'epoch': 0.12}

  4%|▍         | 553/13317 [05:42<2:10:28,  1.63it/s]
  4%|▍         | 554/13317 [05:43<2:10:28,  1.63it/s]
                                                     
{'loss': 0.4113, 'grad_norm': 0.7647483944892883, 'learning_rate': 4.15915915915916e-06, 'epoch': 0.12}

  4%|▍         | 554/13317 [05:43<2:10:28,  1.63it/s]
  4%|▍         | 555/13317 [05:43<2:10:30,  1.63it/s]
                                                     
{'loss': 0.4039, 'grad_norm': 0.6421744227409363, 'learning_rate': 4.166666666666667e-06, 'epoch': 0.13}

  4%|▍         | 555/13317 [05:43<2:10:30,  1.63it/s]
  4%|▍         | 556/13317 [05:44<2:10:29,  1.63it/s]
                                                     
{'loss': 0.3835, 'grad_norm': 0.6301008462905884, 'learning_rate': 4.174174174174174e-06, 'epoch': 0.13}

  4%|▍         | 556/13317 [05:44<2:10:29,  1.63it/s]
  4%|▍         | 557/13317 [05:44<2:10:30,  1.63it/s]
                                                     
{'loss': 0.5426, 'grad_norm': 0.7693248987197876, 'learning_rate': 4.181681681681682e-06, 'epoch': 0.13}

  4%|▍         | 557/13317 [05:45<2:10:30,  1.63it/s]
  4%|▍         | 558/13317 [05:45<2:10:30,  1.63it/s]
                                                     
{'loss': 0.4974, 'grad_norm': 0.7547498941421509, 'learning_rate': 4.189189189189189e-06, 'epoch': 0.13}

  4%|▍         | 558/13317 [05:45<2:10:30,  1.63it/s]
  4%|▍         | 559/13317 [05:46<2:10:29,  1.63it/s]
                                                     
{'loss': 0.3836, 'grad_norm': 0.742424726486206, 'learning_rate': 4.196696696696697e-06, 'epoch': 0.13}

  4%|▍         | 559/13317 [05:46<2:10:29,  1.63it/s]
  4%|▍         | 560/13317 [05:46<2:10:25,  1.63it/s]
                                                     
{'loss': 0.3999, 'grad_norm': 0.7679975032806396, 'learning_rate': 4.204204204204204e-06, 'epoch': 0.13}

  4%|▍         | 560/13317 [05:46<2:10:25,  1.63it/s]
  4%|▍         | 561/13317 [05:47<2:10:25,  1.63it/s]
                                                     
{'loss': 0.3708, 'grad_norm': 0.6942506432533264, 'learning_rate': 4.2117117117117115e-06, 'epoch': 0.13}

  4%|▍         | 561/13317 [05:47<2:10:25,  1.63it/s]
  4%|▍         | 562/13317 [05:48<2:10:34,  1.63it/s]
                                                     
{'loss': 0.3409, 'grad_norm': 0.6488205194473267, 'learning_rate': 4.2192192192192195e-06, 'epoch': 0.13}

  4%|▍         | 562/13317 [05:48<2:10:34,  1.63it/s]
  4%|▍         | 563/13317 [05:48<2:10:29,  1.63it/s]
                                                     
{'loss': 0.4367, 'grad_norm': 0.6213116645812988, 'learning_rate': 4.226726726726727e-06, 'epoch': 0.13}

  4%|▍         | 563/13317 [05:48<2:10:29,  1.63it/s]
  4%|▍         | 564/13317 [05:49<2:10:29,  1.63it/s]
                                                     
{'loss': 0.5009, 'grad_norm': 0.7521598935127258, 'learning_rate': 4.234234234234235e-06, 'epoch': 0.13}

  4%|▍         | 564/13317 [05:49<2:10:29,  1.63it/s]
  4%|▍         | 565/13317 [05:49<2:10:30,  1.63it/s]
                                                     
{'loss': 0.3721, 'grad_norm': 0.6627850532531738, 'learning_rate': 4.241741741741743e-06, 'epoch': 0.13}

  4%|▍         | 565/13317 [05:50<2:10:30,  1.63it/s]
  4%|▍         | 566/13317 [05:50<2:10:35,  1.63it/s]
                                                     
{'loss': 0.3884, 'grad_norm': 0.7788907289505005, 'learning_rate': 4.24924924924925e-06, 'epoch': 0.13}

  4%|▍         | 566/13317 [05:50<2:10:35,  1.63it/s]
  4%|▍         | 567/13317 [05:51<2:10:28,  1.63it/s]
                                                     
{'loss': 0.3924, 'grad_norm': 0.8588048219680786, 'learning_rate': 4.256756756756757e-06, 'epoch': 0.13}

  4%|▍         | 567/13317 [05:51<2:10:28,  1.63it/s]
  4%|▍         | 568/13317 [05:51<2:10:25,  1.63it/s]
                                                     
{'loss': 0.419, 'grad_norm': 0.5870054960250854, 'learning_rate': 4.264264264264265e-06, 'epoch': 0.13}

  4%|▍         | 568/13317 [05:51<2:10:25,  1.63it/s]
  4%|▍         | 569/13317 [05:52<2:10:24,  1.63it/s]
                                                     
{'loss': 0.3642, 'grad_norm': 0.6916638612747192, 'learning_rate': 4.271771771771772e-06, 'epoch': 0.13}

  4%|▍         | 569/13317 [05:52<2:10:24,  1.63it/s]
  4%|▍         | 570/13317 [05:52<2:10:26,  1.63it/s]
                                                     
{'loss': 0.4308, 'grad_norm': 0.7599396705627441, 'learning_rate': 4.27927927927928e-06, 'epoch': 0.13}

  4%|▍         | 570/13317 [05:53<2:10:26,  1.63it/s]
  4%|▍         | 571/13317 [05:53<2:10:27,  1.63it/s]
                                                     
{'loss': 0.557, 'grad_norm': 0.8604689240455627, 'learning_rate': 4.286786786786787e-06, 'epoch': 0.13}

  4%|▍         | 571/13317 [05:53<2:10:27,  1.63it/s]
  4%|▍         | 572/13317 [05:54<2:10:26,  1.63it/s]
                                                     
{'loss': 0.4983, 'grad_norm': 0.7239753007888794, 'learning_rate': 4.294294294294294e-06, 'epoch': 0.13}

  4%|▍         | 572/13317 [05:54<2:10:26,  1.63it/s]
  4%|▍         | 573/13317 [05:54<2:10:30,  1.63it/s]
                                                     
{'loss': 0.3905, 'grad_norm': 0.7629039883613586, 'learning_rate': 4.301801801801802e-06, 'epoch': 0.13}

  4%|▍         | 573/13317 [05:54<2:10:30,  1.63it/s]
  4%|▍         | 574/13317 [05:55<2:10:28,  1.63it/s]
                                                     
{'loss': 0.4654, 'grad_norm': 0.7137024998664856, 'learning_rate': 4.3093093093093095e-06, 'epoch': 0.13}

  4%|▍         | 574/13317 [05:55<2:10:28,  1.63it/s]
  4%|▍         | 575/13317 [05:56<2:10:26,  1.63it/s]
                                                     
{'loss': 0.3836, 'grad_norm': 0.6769149899482727, 'learning_rate': 4.3168168168168175e-06, 'epoch': 0.13}

  4%|▍         | 575/13317 [05:56<2:10:26,  1.63it/s]
  4%|▍         | 576/13317 [05:56<2:10:23,  1.63it/s]
                                                     
{'loss': 0.4755, 'grad_norm': 0.7132650017738342, 'learning_rate': 4.324324324324325e-06, 'epoch': 0.13}

  4%|▍         | 576/13317 [05:56<2:10:23,  1.63it/s]
  4%|▍         | 577/13317 [05:57<2:10:18,  1.63it/s]
                                                     
{'loss': 0.4236, 'grad_norm': 0.6844512820243835, 'learning_rate': 4.331831831831832e-06, 'epoch': 0.13}

  4%|▍         | 577/13317 [05:57<2:10:18,  1.63it/s]
  4%|▍         | 578/13317 [05:57<2:10:32,  1.63it/s]
                                                     
{'loss': 0.3707, 'grad_norm': 0.6702597737312317, 'learning_rate': 4.33933933933934e-06, 'epoch': 0.13}

  4%|▍         | 578/13317 [05:58<2:10:32,  1.63it/s]
  4%|▍         | 579/13317 [05:58<2:10:23,  1.63it/s]
                                                     
{'loss': 0.413, 'grad_norm': 0.7079041004180908, 'learning_rate': 4.346846846846847e-06, 'epoch': 0.13}

  4%|▍         | 579/13317 [05:58<2:10:23,  1.63it/s]
  4%|▍         | 580/13317 [05:59<2:10:20,  1.63it/s]
                                                     
{'loss': 0.4657, 'grad_norm': 0.6552077531814575, 'learning_rate': 4.354354354354355e-06, 'epoch': 0.13}

  4%|▍         | 580/13317 [05:59<2:10:20,  1.63it/s]
  4%|▍         | 581/13317 [05:59<2:10:18,  1.63it/s]
                                                     
{'loss': 0.3489, 'grad_norm': 0.562629222869873, 'learning_rate': 4.361861861861862e-06, 'epoch': 0.13}

  4%|▍         | 581/13317 [05:59<2:10:18,  1.63it/s]
  4%|▍         | 582/13317 [06:00<2:10:20,  1.63it/s]
                                                     
{'loss': 0.3764, 'grad_norm': 0.6495241522789001, 'learning_rate': 4.369369369369369e-06, 'epoch': 0.13}

  4%|▍         | 582/13317 [06:00<2:10:20,  1.63it/s]
  4%|▍         | 583/13317 [06:00<2:10:29,  1.63it/s]
                                                     
{'loss': 0.4115, 'grad_norm': 0.6811219453811646, 'learning_rate': 4.376876876876877e-06, 'epoch': 0.13}

  4%|▍         | 583/13317 [06:01<2:10:29,  1.63it/s]
  4%|▍         | 584/13317 [06:01<2:10:22,  1.63it/s]
                                                     
{'loss': 0.3646, 'grad_norm': 0.7039616703987122, 'learning_rate': 4.384384384384384e-06, 'epoch': 0.13}

  4%|▍         | 584/13317 [06:01<2:10:22,  1.63it/s]
  4%|▍         | 585/13317 [06:02<2:10:13,  1.63it/s]
                                                     
{'loss': 0.4052, 'grad_norm': 0.6591152548789978, 'learning_rate': 4.391891891891892e-06, 'epoch': 0.13}

  4%|▍         | 585/13317 [06:02<2:10:13,  1.63it/s]
  4%|▍         | 586/13317 [06:02<2:10:18,  1.63it/s]
                                                     
{'loss': 0.3769, 'grad_norm': 0.6417102217674255, 'learning_rate': 4.3993993993993996e-06, 'epoch': 0.13}

  4%|▍         | 586/13317 [06:02<2:10:18,  1.63it/s]
  4%|▍         | 587/13317 [06:03<2:10:17,  1.63it/s]
                                                     
{'loss': 0.387, 'grad_norm': 0.6521098613739014, 'learning_rate': 4.406906906906907e-06, 'epoch': 0.13}

  4%|▍         | 587/13317 [06:03<2:10:17,  1.63it/s]
  4%|▍         | 588/13317 [06:04<2:10:26,  1.63it/s]
                                                     
{'loss': 0.3956, 'grad_norm': 0.6957395672798157, 'learning_rate': 4.414414414414415e-06, 'epoch': 0.13}

  4%|▍         | 588/13317 [06:04<2:10:26,  1.63it/s]
  4%|▍         | 589/13317 [06:04<2:10:17,  1.63it/s]
                                                     
{'loss': 0.4759, 'grad_norm': 0.7646422386169434, 'learning_rate': 4.421921921921922e-06, 'epoch': 0.13}

  4%|▍         | 589/13317 [06:04<2:10:17,  1.63it/s]
  4%|▍         | 590/13317 [06:05<2:10:14,  1.63it/s]
                                                     
{'loss': 0.2878, 'grad_norm': 0.5918605327606201, 'learning_rate': 4.42942942942943e-06, 'epoch': 0.13}

  4%|▍         | 590/13317 [06:05<2:10:14,  1.63it/s]
  4%|▍         | 591/13317 [06:05<2:10:12,  1.63it/s]
                                                     
{'loss': 0.4514, 'grad_norm': 0.8569434285163879, 'learning_rate': 4.436936936936938e-06, 'epoch': 0.13}

  4%|▍         | 591/13317 [06:06<2:10:12,  1.63it/s]
  4%|▍         | 592/13317 [06:06<2:10:09,  1.63it/s]
                                                     
{'loss': 0.3612, 'grad_norm': 0.5796396732330322, 'learning_rate': 4.444444444444444e-06, 'epoch': 0.13}

  4%|▍         | 592/13317 [06:06<2:10:09,  1.63it/s]
  4%|▍         | 593/13317 [06:07<2:10:18,  1.63it/s]
                                                     
{'loss': 0.457, 'grad_norm': 0.6724703907966614, 'learning_rate': 4.451951951951952e-06, 'epoch': 0.13}

  4%|▍         | 593/13317 [06:07<2:10:18,  1.63it/s]
  4%|▍         | 594/13317 [06:07<2:10:14,  1.63it/s]
                                                     
{'loss': 0.3436, 'grad_norm': 0.662566602230072, 'learning_rate': 4.45945945945946e-06, 'epoch': 0.13}

  4%|▍         | 594/13317 [06:07<2:10:14,  1.63it/s]
  4%|▍         | 595/13317 [06:08<2:10:05,  1.63it/s]
                                                     
{'loss': 0.3422, 'grad_norm': 0.6361055970191956, 'learning_rate': 4.466966966966967e-06, 'epoch': 0.13}

  4%|▍         | 595/13317 [06:08<2:10:05,  1.63it/s]
  4%|▍         | 596/13317 [06:08<2:10:14,  1.63it/s]
                                                     
{'loss': 0.3612, 'grad_norm': 0.5670613646507263, 'learning_rate': 4.474474474474475e-06, 'epoch': 0.13}

  4%|▍         | 596/13317 [06:09<2:10:14,  1.63it/s]
  4%|▍         | 597/13317 [06:09<2:10:11,  1.63it/s]
                                                     
{'loss': 0.4056, 'grad_norm': 0.7257280945777893, 'learning_rate': 4.4819819819819824e-06, 'epoch': 0.13}

  4%|▍         | 597/13317 [06:09<2:10:11,  1.63it/s]
  4%|▍         | 598/13317 [06:10<2:10:23,  1.63it/s]
                                                     
{'loss': 0.6114, 'grad_norm': 0.8918281197547913, 'learning_rate': 4.48948948948949e-06, 'epoch': 0.13}

  4%|▍         | 598/13317 [06:10<2:10:23,  1.63it/s]
  4%|▍         | 599/13317 [06:10<2:10:19,  1.63it/s]
                                                     
{'loss': 0.3274, 'grad_norm': 0.7188802361488342, 'learning_rate': 4.496996996996998e-06, 'epoch': 0.13}

  4%|▍         | 599/13317 [06:10<2:10:19,  1.63it/s]
  5%|▍         | 600/13317 [06:11<2:10:17,  1.63it/s]
                                                     
{'loss': 0.4072, 'grad_norm': 0.6751221418380737, 'learning_rate': 4.504504504504505e-06, 'epoch': 0.14}

  5%|▍         | 600/13317 [06:11<2:10:17,  1.63it/s]
  5%|▍         | 601/13317 [06:12<2:10:46,  1.62it/s]
                                                     
{'loss': 0.4516, 'grad_norm': 0.7631073594093323, 'learning_rate': 4.512012012012013e-06, 'epoch': 0.14}

  5%|▍         | 601/13317 [06:12<2:10:46,  1.62it/s]
  5%|▍         | 602/13317 [06:12<2:10:37,  1.62it/s]
                                                     
{'loss': 0.4466, 'grad_norm': 0.7512320280075073, 'learning_rate': 4.51951951951952e-06, 'epoch': 0.14}

  5%|▍         | 602/13317 [06:12<2:10:37,  1.62it/s]
  5%|▍         | 603/13317 [06:13<2:10:33,  1.62it/s]
                                                     
{'loss': 0.3985, 'grad_norm': 0.6817054152488708, 'learning_rate': 4.527027027027027e-06, 'epoch': 0.14}

  5%|▍         | 603/13317 [06:13<2:10:33,  1.62it/s]
  5%|▍         | 604/13317 [06:13<2:10:22,  1.63it/s]
                                                     
{'loss': 0.3483, 'grad_norm': 0.6410658955574036, 'learning_rate': 4.534534534534535e-06, 'epoch': 0.14}

  5%|▍         | 604/13317 [06:14<2:10:22,  1.63it/s]
  5%|▍         | 605/13317 [06:14<2:10:16,  1.63it/s]
                                                     
{'loss': 0.4605, 'grad_norm': 0.8276411890983582, 'learning_rate': 4.542042042042042e-06, 'epoch': 0.14}

  5%|▍         | 605/13317 [06:14<2:10:16,  1.63it/s]
  5%|▍         | 606/13317 [06:15<2:10:17,  1.63it/s]
                                                     
{'loss': 0.4297, 'grad_norm': 0.7772295475006104, 'learning_rate': 4.54954954954955e-06, 'epoch': 0.14}

  5%|▍         | 606/13317 [06:15<2:10:17,  1.63it/s]
  5%|▍         | 607/13317 [06:15<2:10:08,  1.63it/s]
                                                     
{'loss': 0.3793, 'grad_norm': 0.6447915434837341, 'learning_rate': 4.557057057057057e-06, 'epoch': 0.14}

  5%|▍         | 607/13317 [06:15<2:10:08,  1.63it/s]
  5%|▍         | 608/13317 [06:16<2:10:41,  1.62it/s]
                                                     
{'loss': 0.407, 'grad_norm': 0.6365005373954773, 'learning_rate': 4.5645645645645645e-06, 'epoch': 0.14}

  5%|▍         | 608/13317 [06:16<2:10:41,  1.62it/s]
  5%|▍         | 609/13317 [06:16<2:10:30,  1.62it/s]
                                                     
{'loss': 0.3757, 'grad_norm': 0.6875844597816467, 'learning_rate': 4.5720720720720725e-06, 'epoch': 0.14}

  5%|▍         | 609/13317 [06:17<2:10:30,  1.62it/s]
  5%|▍         | 610/13317 [06:17<2:10:14,  1.63it/s]
                                                     
{'loss': 0.464, 'grad_norm': 0.7170318961143494, 'learning_rate': 4.57957957957958e-06, 'epoch': 0.14}

  5%|▍         | 610/13317 [06:17<2:10:14,  1.63it/s]
  5%|▍         | 611/13317 [06:18<2:10:05,  1.63it/s]
                                                     
{'loss': 0.4004, 'grad_norm': 0.6795490980148315, 'learning_rate': 4.587087087087088e-06, 'epoch': 0.14}

  5%|▍         | 611/13317 [06:18<2:10:05,  1.63it/s]
  5%|▍         | 612/13317 [06:18<2:10:01,  1.63it/s]
                                                     
{'loss': 0.4999, 'grad_norm': 0.8649230599403381, 'learning_rate': 4.594594594594596e-06, 'epoch': 0.14}

  5%|▍         | 612/13317 [06:18<2:10:01,  1.63it/s]
  5%|▍         | 613/13317 [06:19<2:10:23,  1.62it/s]
                                                     
{'loss': 0.4786, 'grad_norm': 0.7106096744537354, 'learning_rate': 4.602102102102102e-06, 'epoch': 0.14}

  5%|▍         | 613/13317 [06:19<2:10:23,  1.62it/s]
  5%|▍         | 614/13317 [06:20<2:10:17,  1.62it/s]
                                                     
{'loss': 0.4, 'grad_norm': 0.6837033033370972, 'learning_rate': 4.60960960960961e-06, 'epoch': 0.14}

  5%|▍         | 614/13317 [06:20<2:10:17,  1.62it/s]
  5%|▍         | 615/13317 [06:20<2:10:18,  1.62it/s]
                                                     
{'loss': 0.4027, 'grad_norm': 0.694284200668335, 'learning_rate': 4.617117117117118e-06, 'epoch': 0.14}

  5%|▍         | 615/13317 [06:20<2:10:18,  1.62it/s]
  5%|▍         | 616/13317 [06:21<2:10:19,  1.62it/s]
                                                     
{'loss': 0.3387, 'grad_norm': 0.5538964867591858, 'learning_rate': 4.624624624624625e-06, 'epoch': 0.14}

  5%|▍         | 616/13317 [06:21<2:10:19,  1.62it/s]
  5%|▍         | 617/13317 [06:21<2:10:07,  1.63it/s]
                                                     
{'loss': 0.619, 'grad_norm': 0.8222833275794983, 'learning_rate': 4.632132132132133e-06, 'epoch': 0.14}

  5%|▍         | 617/13317 [06:22<2:10:07,  1.63it/s]
  5%|▍         | 618/13317 [06:22<2:10:09,  1.63it/s]
                                                     
{'loss': 0.3462, 'grad_norm': 0.6526497602462769, 'learning_rate': 4.63963963963964e-06, 'epoch': 0.14}

  5%|▍         | 618/13317 [06:22<2:10:09,  1.63it/s]
  5%|▍         | 619/13317 [06:23<2:10:02,  1.63it/s]
                                                     
{'loss': 0.4162, 'grad_norm': 0.6797808408737183, 'learning_rate': 4.647147147147147e-06, 'epoch': 0.14}

  5%|▍         | 619/13317 [06:23<2:10:02,  1.63it/s]
  5%|▍         | 620/13317 [06:23<2:09:58,  1.63it/s]
                                                     
{'loss': 0.5414, 'grad_norm': 0.9431416988372803, 'learning_rate': 4.654654654654655e-06, 'epoch': 0.14}

  5%|▍         | 620/13317 [06:23<2:09:58,  1.63it/s]
  5%|▍         | 621/13317 [06:24<2:09:55,  1.63it/s]
                                                     
{'loss': 0.423, 'grad_norm': 0.7169815301895142, 'learning_rate': 4.6621621621621625e-06, 'epoch': 0.14}

  5%|▍         | 621/13317 [06:24<2:09:55,  1.63it/s]
  5%|▍         | 622/13317 [06:24<2:09:53,  1.63it/s]
                                                     
{'loss': 0.376, 'grad_norm': 0.6444219350814819, 'learning_rate': 4.6696696696696705e-06, 'epoch': 0.14}

  5%|▍         | 622/13317 [06:25<2:09:53,  1.63it/s]
  5%|▍         | 623/13317 [06:25<2:09:49,  1.63it/s]
                                                     
{'loss': 0.4086, 'grad_norm': 0.6835696697235107, 'learning_rate': 4.677177177177178e-06, 'epoch': 0.14}

  5%|▍         | 623/13317 [06:25<2:09:49,  1.63it/s]
  5%|▍         | 624/13317 [06:26<2:09:47,  1.63it/s]
                                                     
{'loss': 0.3088, 'grad_norm': 0.5782841444015503, 'learning_rate': 4.684684684684685e-06, 'epoch': 0.14}

  5%|▍         | 624/13317 [06:26<2:09:47,  1.63it/s]
  5%|▍         | 625/13317 [06:26<2:09:45,  1.63it/s]
                                                     
{'loss': 0.4238, 'grad_norm': 0.679583728313446, 'learning_rate': 4.692192192192193e-06, 'epoch': 0.14}

  5%|▍         | 625/13317 [06:26<2:09:45,  1.63it/s]
  5%|▍         | 626/13317 [06:27<2:09:43,  1.63it/s]
                                                     
{'loss': 0.3927, 'grad_norm': 0.6506307125091553, 'learning_rate': 4.6996996996997e-06, 'epoch': 0.14}

  5%|▍         | 626/13317 [06:27<2:09:43,  1.63it/s]
  5%|▍         | 627/13317 [06:28<2:09:44,  1.63it/s]
                                                     
{'loss': 0.3642, 'grad_norm': 0.6891327500343323, 'learning_rate': 4.707207207207208e-06, 'epoch': 0.14}

  5%|▍         | 627/13317 [06:28<2:09:44,  1.63it/s]
  5%|▍         | 628/13317 [06:28<2:09:43,  1.63it/s]
                                                     
{'loss': 0.5459, 'grad_norm': 0.8559671640396118, 'learning_rate': 4.714714714714715e-06, 'epoch': 0.14}

  5%|▍         | 628/13317 [06:28<2:09:43,  1.63it/s]
  5%|▍         | 629/13317 [06:29<2:09:41,  1.63it/s]
                                                     
{'loss': 0.3167, 'grad_norm': 0.5458211898803711, 'learning_rate': 4.722222222222222e-06, 'epoch': 0.14}

  5%|▍         | 629/13317 [06:29<2:09:41,  1.63it/s]
  5%|▍         | 630/13317 [06:29<2:09:44,  1.63it/s]
                                                     
{'loss': 0.4173, 'grad_norm': 0.6909137964248657, 'learning_rate': 4.72972972972973e-06, 'epoch': 0.14}

  5%|▍         | 630/13317 [06:29<2:09:44,  1.63it/s]
  5%|▍         | 631/13317 [06:30<2:09:46,  1.63it/s]
                                                     
{'loss': 0.4001, 'grad_norm': 0.6511400938034058, 'learning_rate': 4.7372372372372374e-06, 'epoch': 0.14}

  5%|▍         | 631/13317 [06:30<2:09:46,  1.63it/s]
  5%|▍         | 632/13317 [06:31<2:09:42,  1.63it/s]
                                                     
{'loss': 0.5238, 'grad_norm': 0.7925885319709778, 'learning_rate': 4.7447447447447454e-06, 'epoch': 0.14}

  5%|▍         | 632/13317 [06:31<2:09:42,  1.63it/s]
  5%|▍         | 633/13317 [06:31<2:09:44,  1.63it/s]
                                                     
{'loss': 0.4064, 'grad_norm': 0.6806579828262329, 'learning_rate': 4.7522522522522526e-06, 'epoch': 0.14}

  5%|▍         | 633/13317 [06:31<2:09:44,  1.63it/s]
  5%|▍         | 634/13317 [06:32<2:09:42,  1.63it/s]
                                                     
{'loss': 0.3691, 'grad_norm': 0.6237165927886963, 'learning_rate': 4.75975975975976e-06, 'epoch': 0.14}

  5%|▍         | 634/13317 [06:32<2:09:42,  1.63it/s]
  5%|▍         | 635/13317 [06:32<2:09:38,  1.63it/s]
                                                     
{'loss': 0.3252, 'grad_norm': 0.6305434107780457, 'learning_rate': 4.767267267267268e-06, 'epoch': 0.14}

  5%|▍         | 635/13317 [06:33<2:09:38,  1.63it/s]
  5%|▍         | 636/13317 [06:33<2:09:41,  1.63it/s]
                                                     
{'loss': 0.3186, 'grad_norm': 0.6293449401855469, 'learning_rate': 4.774774774774775e-06, 'epoch': 0.14}

  5%|▍         | 636/13317 [06:33<2:09:41,  1.63it/s]
  5%|▍         | 637/13317 [06:34<2:09:37,  1.63it/s]
                                                     
{'loss': 0.4542, 'grad_norm': 0.7509760856628418, 'learning_rate': 4.782282282282283e-06, 'epoch': 0.14}

  5%|▍         | 637/13317 [06:34<2:09:37,  1.63it/s]
  5%|▍         | 638/13317 [06:34<2:09:40,  1.63it/s]
                                                     
{'loss': 0.4172, 'grad_norm': 0.7404630780220032, 'learning_rate': 4.78978978978979e-06, 'epoch': 0.14}

  5%|▍         | 638/13317 [06:34<2:09:40,  1.63it/s]
  5%|▍         | 639/13317 [06:35<2:09:43,  1.63it/s]
                                                     
{'loss': 0.4078, 'grad_norm': 0.6759922504425049, 'learning_rate': 4.797297297297297e-06, 'epoch': 0.14}

  5%|▍         | 639/13317 [06:35<2:09:43,  1.63it/s]
  5%|▍         | 640/13317 [06:35<2:09:37,  1.63it/s]
                                                     
{'loss': 0.4511, 'grad_norm': 0.7170428037643433, 'learning_rate': 4.804804804804805e-06, 'epoch': 0.14}

  5%|▍         | 640/13317 [06:36<2:09:37,  1.63it/s]
  5%|▍         | 641/13317 [06:36<2:09:40,  1.63it/s]
                                                     
{'loss': 0.4878, 'grad_norm': 0.7855735421180725, 'learning_rate': 4.812312312312313e-06, 'epoch': 0.14}

  5%|▍         | 641/13317 [06:36<2:09:40,  1.63it/s]
  5%|▍         | 642/13317 [06:37<2:09:43,  1.63it/s]
                                                     
{'loss': 0.3515, 'grad_norm': 0.6572322249412537, 'learning_rate': 4.81981981981982e-06, 'epoch': 0.14}

  5%|▍         | 642/13317 [06:37<2:09:43,  1.63it/s]
  5%|▍         | 643/13317 [06:37<2:09:40,  1.63it/s]
                                                     
{'loss': 0.3865, 'grad_norm': 0.6224420070648193, 'learning_rate': 4.8273273273273275e-06, 'epoch': 0.14}

  5%|▍         | 643/13317 [06:37<2:09:40,  1.63it/s]
  5%|▍         | 644/13317 [06:38<2:09:37,  1.63it/s]
                                                     
{'loss': 0.4937, 'grad_norm': 0.8594943881034851, 'learning_rate': 4.8348348348348355e-06, 'epoch': 0.15}

  5%|▍         | 644/13317 [06:38<2:09:37,  1.63it/s]
  5%|▍         | 645/13317 [06:39<2:09:31,  1.63it/s]
                                                     
{'loss': 0.408, 'grad_norm': 0.6594147682189941, 'learning_rate': 4.842342342342343e-06, 'epoch': 0.15}

  5%|▍         | 645/13317 [06:39<2:09:31,  1.63it/s]
  5%|▍         | 646/13317 [06:39<2:09:27,  1.63it/s]
                                                     
{'loss': 0.3868, 'grad_norm': 0.7192913889884949, 'learning_rate': 4.849849849849851e-06, 'epoch': 0.15}

  5%|▍         | 646/13317 [06:39<2:09:27,  1.63it/s]
  5%|▍         | 647/13317 [06:40<2:09:33,  1.63it/s]
                                                     
{'loss': 0.49, 'grad_norm': 0.6879770159721375, 'learning_rate': 4.857357357357358e-06, 'epoch': 0.15}

  5%|▍         | 647/13317 [06:40<2:09:33,  1.63it/s]
  5%|▍         | 648/13317 [06:40<2:09:34,  1.63it/s]
                                                     
{'loss': 0.348, 'grad_norm': 0.6410459280014038, 'learning_rate': 4.864864864864866e-06, 'epoch': 0.15}

  5%|▍         | 648/13317 [06:41<2:09:34,  1.63it/s]
  5%|▍         | 649/13317 [06:41<2:09:30,  1.63it/s]
                                                     
{'loss': 0.3937, 'grad_norm': 0.7045912742614746, 'learning_rate': 4.872372372372373e-06, 'epoch': 0.15}

  5%|▍         | 649/13317 [06:41<2:09:30,  1.63it/s]
  5%|▍         | 650/13317 [06:42<2:09:29,  1.63it/s]
                                                     
{'loss': 0.3776, 'grad_norm': 0.5418573617935181, 'learning_rate': 4.87987987987988e-06, 'epoch': 0.15}

  5%|▍         | 650/13317 [06:42<2:09:29,  1.63it/s]
  5%|▍         | 651/13317 [06:42<2:09:29,  1.63it/s]
                                                     
{'loss': 0.3679, 'grad_norm': 0.6477158665657043, 'learning_rate': 4.887387387387388e-06, 'epoch': 0.15}

  5%|▍         | 651/13317 [06:42<2:09:29,  1.63it/s]
  5%|▍         | 652/13317 [06:43<2:09:24,  1.63it/s]
                                                     
{'loss': 0.3814, 'grad_norm': 0.793807864189148, 'learning_rate': 4.894894894894895e-06, 'epoch': 0.15}

  5%|▍         | 652/13317 [06:43<2:09:24,  1.63it/s]
  5%|▍         | 653/13317 [06:43<2:09:26,  1.63it/s]
                                                     
{'loss': 0.504, 'grad_norm': 0.6894840598106384, 'learning_rate': 4.902402402402403e-06, 'epoch': 0.15}

  5%|▍         | 653/13317 [06:44<2:09:26,  1.63it/s]
  5%|▍         | 654/13317 [06:44<2:09:28,  1.63it/s]
                                                     
{'loss': 0.3331, 'grad_norm': 0.6427252888679504, 'learning_rate': 4.90990990990991e-06, 'epoch': 0.15}

  5%|▍         | 654/13317 [06:44<2:09:28,  1.63it/s]
  5%|▍         | 655/13317 [06:45<2:09:31,  1.63it/s]
                                                     
{'loss': 0.3706, 'grad_norm': 0.8315549492835999, 'learning_rate': 4.9174174174174175e-06, 'epoch': 0.15}

  5%|▍         | 655/13317 [06:45<2:09:31,  1.63it/s]
  5%|▍         | 656/13317 [06:45<2:09:31,  1.63it/s]
                                                     
{'loss': 0.3572, 'grad_norm': 0.6528039574623108, 'learning_rate': 4.9249249249249255e-06, 'epoch': 0.15}

  5%|▍         | 656/13317 [06:45<2:09:31,  1.63it/s]
  5%|▍         | 657/13317 [06:46<2:09:33,  1.63it/s]
                                                     
{'loss': 0.3957, 'grad_norm': 0.725016713142395, 'learning_rate': 4.932432432432433e-06, 'epoch': 0.15}

  5%|▍         | 657/13317 [06:46<2:09:33,  1.63it/s]
  5%|▍         | 658/13317 [06:47<2:09:36,  1.63it/s]
                                                     
{'loss': 0.4074, 'grad_norm': 0.6253482699394226, 'learning_rate': 4.939939939939941e-06, 'epoch': 0.15}

  5%|▍         | 658/13317 [06:47<2:09:36,  1.63it/s]
  5%|▍         | 659/13317 [06:47<2:09:35,  1.63it/s]
                                                     
{'loss': 0.4264, 'grad_norm': 0.7041759490966797, 'learning_rate': 4.947447447447448e-06, 'epoch': 0.15}

  5%|▍         | 659/13317 [06:47<2:09:35,  1.63it/s]
  5%|▍         | 660/13317 [06:48<2:09:27,  1.63it/s]
                                                     
{'loss': 0.4899, 'grad_norm': 0.8274503946304321, 'learning_rate': 4.954954954954955e-06, 'epoch': 0.15}

  5%|▍         | 660/13317 [06:48<2:09:27,  1.63it/s]
  5%|▍         | 661/13317 [06:48<2:09:24,  1.63it/s]
                                                     
{'loss': 0.3736, 'grad_norm': 0.6121860146522522, 'learning_rate': 4.962462462462463e-06, 'epoch': 0.15}

  5%|▍         | 661/13317 [06:49<2:09:24,  1.63it/s]
  5%|▍         | 662/13317 [06:49<2:09:19,  1.63it/s]
                                                     
{'loss': 0.3762, 'grad_norm': 0.7112452983856201, 'learning_rate': 4.96996996996997e-06, 'epoch': 0.15}

  5%|▍         | 662/13317 [06:49<2:09:19,  1.63it/s]
  5%|▍         | 663/13317 [06:50<2:09:18,  1.63it/s]
                                                     
{'loss': 0.3994, 'grad_norm': 0.7543418407440186, 'learning_rate': 4.977477477477478e-06, 'epoch': 0.15}

  5%|▍         | 663/13317 [06:50<2:09:18,  1.63it/s]
  5%|▍         | 664/13317 [06:50<2:09:17,  1.63it/s]
                                                     
{'loss': 0.4145, 'grad_norm': 0.7494813203811646, 'learning_rate': 4.984984984984985e-06, 'epoch': 0.15}

  5%|▍         | 664/13317 [06:50<2:09:17,  1.63it/s]
  5%|▍         | 665/13317 [06:51<2:09:13,  1.63it/s]
                                                     
{'loss': 0.4096, 'grad_norm': 0.6854363083839417, 'learning_rate': 4.992492492492492e-06, 'epoch': 0.15}

  5%|▍         | 665/13317 [06:51<2:09:13,  1.63it/s]
  5%|▌         | 666/13317 [06:51<2:09:14,  1.63it/s]
                                                     
{'loss': 0.4875, 'grad_norm': 1.243247151374817, 'learning_rate': 5e-06, 'epoch': 0.15}

  5%|▌         | 666/13317 [06:52<2:09:14,  1.63it/s]
  5%|▌         | 667/13317 [06:52<2:09:16,  1.63it/s]
                                                     
{'loss': 0.3935, 'grad_norm': 0.6826725006103516, 'learning_rate': 5.0075075075075076e-06, 'epoch': 0.15}

  5%|▌         | 667/13317 [06:52<2:09:16,  1.63it/s]
  5%|▌         | 668/13317 [06:53<2:09:14,  1.63it/s]
                                                     
{'loss': 0.4136, 'grad_norm': 0.737432062625885, 'learning_rate': 5.0150150150150156e-06, 'epoch': 0.15}

  5%|▌         | 668/13317 [06:53<2:09:14,  1.63it/s]
  5%|▌         | 669/13317 [06:53<2:09:38,  1.63it/s]
                                                     
{'loss': 0.4016, 'grad_norm': 0.6825419664382935, 'learning_rate': 5.022522522522523e-06, 'epoch': 0.15}

  5%|▌         | 669/13317 [06:53<2:09:38,  1.63it/s]
  5%|▌         | 670/13317 [06:54<2:09:32,  1.63it/s]
                                                     
{'loss': 0.4014, 'grad_norm': 0.6437894701957703, 'learning_rate': 5.030030030030031e-06, 'epoch': 0.15}

  5%|▌         | 670/13317 [06:54<2:09:32,  1.63it/s]
  5%|▌         | 671/13317 [06:55<2:09:27,  1.63it/s]
                                                     
{'loss': 0.4783, 'grad_norm': 0.8737983703613281, 'learning_rate': 5.037537537537538e-06, 'epoch': 0.15}

  5%|▌         | 671/13317 [06:55<2:09:27,  1.63it/s]
  5%|▌         | 672/13317 [06:55<2:09:21,  1.63it/s]
                                                     
{'loss': 0.4039, 'grad_norm': 0.7414088249206543, 'learning_rate': 5.045045045045045e-06, 'epoch': 0.15}

  5%|▌         | 672/13317 [06:55<2:09:21,  1.63it/s]
  5%|▌         | 673/13317 [06:56<2:09:20,  1.63it/s]
                                                     
{'loss': 0.5214, 'grad_norm': 0.7128754258155823, 'learning_rate': 5.052552552552553e-06, 'epoch': 0.15}

  5%|▌         | 673/13317 [06:56<2:09:20,  1.63it/s]
  5%|▌         | 674/13317 [06:56<2:09:15,  1.63it/s]
                                                     
{'loss': 0.3904, 'grad_norm': 0.6714414954185486, 'learning_rate': 5.06006006006006e-06, 'epoch': 0.15}

  5%|▌         | 674/13317 [06:56<2:09:15,  1.63it/s]
  5%|▌         | 675/13317 [06:57<2:09:19,  1.63it/s]
                                                     
{'loss': 0.4171, 'grad_norm': 0.7112318277359009, 'learning_rate': 5.067567567567568e-06, 'epoch': 0.15}

  5%|▌         | 675/13317 [06:57<2:09:19,  1.63it/s]
  5%|▌         | 676/13317 [06:58<2:09:25,  1.63it/s]
                                                     
{'loss': 0.377, 'grad_norm': 0.715488851070404, 'learning_rate': 5.075075075075075e-06, 'epoch': 0.15}

  5%|▌         | 676/13317 [06:58<2:09:25,  1.63it/s]
  5%|▌         | 677/13317 [06:58<2:09:18,  1.63it/s]
                                                     
{'loss': 0.4036, 'grad_norm': 0.7381907105445862, 'learning_rate': 5.0825825825825825e-06, 'epoch': 0.15}

  5%|▌         | 677/13317 [06:58<2:09:18,  1.63it/s]
  5%|▌         | 678/13317 [06:59<2:09:22,  1.63it/s]
                                                     
{'loss': 0.3882, 'grad_norm': 0.7151139974594116, 'learning_rate': 5.0900900900900905e-06, 'epoch': 0.15}

  5%|▌         | 678/13317 [06:59<2:09:22,  1.63it/s]
  5%|▌         | 679/13317 [06:59<2:09:20,  1.63it/s]
                                                     
{'loss': 0.3787, 'grad_norm': 0.7776292562484741, 'learning_rate': 5.097597597597598e-06, 'epoch': 0.15}

  5%|▌         | 679/13317 [07:00<2:09:20,  1.63it/s]
  5%|▌         | 680/13317 [07:00<2:09:19,  1.63it/s]
                                                     
{'loss': 0.3958, 'grad_norm': 0.7129265069961548, 'learning_rate': 5.105105105105106e-06, 'epoch': 0.15}

  5%|▌         | 680/13317 [07:00<2:09:19,  1.63it/s]
  5%|▌         | 681/13317 [07:01<2:09:17,  1.63it/s]
                                                     
{'loss': 0.4595, 'grad_norm': 0.6947645545005798, 'learning_rate': 5.112612612612613e-06, 'epoch': 0.15}

  5%|▌         | 681/13317 [07:01<2:09:17,  1.63it/s]
  5%|▌         | 682/13317 [07:01<2:09:15,  1.63it/s]
                                                     
{'loss': 0.4393, 'grad_norm': 0.6935344934463501, 'learning_rate': 5.12012012012012e-06, 'epoch': 0.15}

  5%|▌         | 682/13317 [07:01<2:09:15,  1.63it/s]
  5%|▌         | 683/13317 [07:02<2:09:24,  1.63it/s]
                                                     
{'loss': 0.3462, 'grad_norm': 0.6865419745445251, 'learning_rate': 5.127627627627628e-06, 'epoch': 0.15}

  5%|▌         | 683/13317 [07:02<2:09:24,  1.63it/s]
  5%|▌         | 684/13317 [07:02<2:09:20,  1.63it/s]
                                                     
{'loss': 0.5063, 'grad_norm': 0.6948108673095703, 'learning_rate': 5.135135135135135e-06, 'epoch': 0.15}

  5%|▌         | 684/13317 [07:03<2:09:20,  1.63it/s]
  5%|▌         | 685/13317 [07:03<2:09:27,  1.63it/s]
                                                     
{'loss': 0.4346, 'grad_norm': 0.6918430924415588, 'learning_rate': 5.142642642642643e-06, 'epoch': 0.15}

  5%|▌         | 685/13317 [07:03<2:09:27,  1.63it/s]
  5%|▌         | 686/13317 [07:04<2:09:13,  1.63it/s]
                                                     
{'loss': 0.468, 'grad_norm': 0.9648342132568359, 'learning_rate': 5.15015015015015e-06, 'epoch': 0.15}

  5%|▌         | 686/13317 [07:04<2:09:13,  1.63it/s]
  5%|▌         | 687/13317 [07:04<2:09:11,  1.63it/s]
                                                     
{'loss': 0.3394, 'grad_norm': 0.6268414258956909, 'learning_rate': 5.157657657657657e-06, 'epoch': 0.15}

  5%|▌         | 687/13317 [07:04<2:09:11,  1.63it/s]
  5%|▌         | 688/13317 [07:05<2:09:26,  1.63it/s]
                                                     
{'loss': 0.3721, 'grad_norm': 0.7131672501564026, 'learning_rate': 5.165165165165165e-06, 'epoch': 0.15}

  5%|▌         | 688/13317 [07:05<2:09:26,  1.63it/s]
  5%|▌         | 689/13317 [07:06<2:09:20,  1.63it/s]
                                                     
{'loss': 0.5541, 'grad_norm': 0.791860044002533, 'learning_rate': 5.1726726726726725e-06, 'epoch': 0.16}

  5%|▌         | 689/13317 [07:06<2:09:20,  1.63it/s]
  5%|▌         | 690/13317 [07:06<2:09:12,  1.63it/s]
                                                     
{'loss': 0.4139, 'grad_norm': 0.8333534598350525, 'learning_rate': 5.180180180180181e-06, 'epoch': 0.16}

  5%|▌         | 690/13317 [07:06<2:09:12,  1.63it/s]
  5%|▌         | 691/13317 [07:07<2:09:07,  1.63it/s]
                                                     
{'loss': 0.4428, 'grad_norm': 0.6658928990364075, 'learning_rate': 5.187687687687688e-06, 'epoch': 0.16}

  5%|▌         | 691/13317 [07:07<2:09:07,  1.63it/s]
  5%|▌         | 692/13317 [07:07<2:09:08,  1.63it/s]
                                                     
{'loss': 0.409, 'grad_norm': 0.7971087694168091, 'learning_rate': 5.195195195195195e-06, 'epoch': 0.16}

  5%|▌         | 692/13317 [07:08<2:09:08,  1.63it/s]
  5%|▌         | 693/13317 [07:08<2:09:05,  1.63it/s]
                                                     
{'loss': 0.4165, 'grad_norm': 0.821128249168396, 'learning_rate': 5.202702702702704e-06, 'epoch': 0.16}

  5%|▌         | 693/13317 [07:08<2:09:05,  1.63it/s]
  5%|▌         | 694/13317 [07:09<2:09:06,  1.63it/s]
                                                     
{'loss': 0.4396, 'grad_norm': 0.7325173020362854, 'learning_rate': 5.21021021021021e-06, 'epoch': 0.16}

  5%|▌         | 694/13317 [07:09<2:09:06,  1.63it/s]
  5%|▌         | 695/13317 [07:09<2:09:06,  1.63it/s]
                                                     
{'loss': 0.3571, 'grad_norm': 0.6991190314292908, 'learning_rate': 5.217717717717719e-06, 'epoch': 0.16}

  5%|▌         | 695/13317 [07:09<2:09:06,  1.63it/s]
  5%|▌         | 696/13317 [07:10<2:09:03,  1.63it/s]
                                                     
{'loss': 0.4184, 'grad_norm': 0.7503544092178345, 'learning_rate': 5.225225225225226e-06, 'epoch': 0.16}

  5%|▌         | 696/13317 [07:10<2:09:03,  1.63it/s]
  5%|▌         | 697/13317 [07:10<2:08:58,  1.63it/s]
                                                     
{'loss': 0.3864, 'grad_norm': 0.7109355926513672, 'learning_rate': 5.232732732732732e-06, 'epoch': 0.16}

  5%|▌         | 697/13317 [07:11<2:08:58,  1.63it/s]
  5%|▌         | 698/13317 [07:11<2:09:03,  1.63it/s]
                                                     
{'loss': 0.2955, 'grad_norm': 0.5675808787345886, 'learning_rate': 5.240240240240241e-06, 'epoch': 0.16}

  5%|▌         | 698/13317 [07:11<2:09:03,  1.63it/s]
  5%|▌         | 699/13317 [07:12<2:09:07,  1.63it/s]
                                                     
{'loss': 0.4071, 'grad_norm': 0.7245252132415771, 'learning_rate': 5.247747747747748e-06, 'epoch': 0.16}

  5%|▌         | 699/13317 [07:12<2:09:07,  1.63it/s]
  5%|▌         | 700/13317 [07:12<2:09:06,  1.63it/s]
                                                     
{'loss': 0.4177, 'grad_norm': 0.6785301566123962, 'learning_rate': 5.255255255255256e-06, 'epoch': 0.16}

  5%|▌         | 700/13317 [07:12<2:09:06,  1.63it/s]
  5%|▌         | 701/13317 [07:13<2:09:06,  1.63it/s]
                                                     
{'loss': 0.4703, 'grad_norm': 0.6547427177429199, 'learning_rate': 5.262762762762763e-06, 'epoch': 0.16}

  5%|▌         | 701/13317 [07:13<2:09:06,  1.63it/s]
  5%|▌         | 702/13317 [07:14<2:09:03,  1.63it/s]
                                                     
{'loss': 0.4705, 'grad_norm': 0.7608590722084045, 'learning_rate': 5.2702702702702705e-06, 'epoch': 0.16}

  5%|▌         | 702/13317 [07:14<2:09:03,  1.63it/s]
  5%|▌         | 703/13317 [07:14<2:08:59,  1.63it/s]
                                                     
{'loss': 0.3471, 'grad_norm': 0.6363322734832764, 'learning_rate': 5.2777777777777785e-06, 'epoch': 0.16}

  5%|▌         | 703/13317 [07:14<2:08:59,  1.63it/s]
  5%|▌         | 704/13317 [07:15<2:09:03,  1.63it/s]
                                                     
{'loss': 0.3986, 'grad_norm': 0.666003942489624, 'learning_rate': 5.285285285285286e-06, 'epoch': 0.16}

  5%|▌         | 704/13317 [07:15<2:09:03,  1.63it/s]
  5%|▌         | 705/13317 [07:15<2:08:54,  1.63it/s]
                                                     
{'loss': 0.3985, 'grad_norm': 0.8007597327232361, 'learning_rate': 5.292792792792794e-06, 'epoch': 0.16}

  5%|▌         | 705/13317 [07:16<2:08:54,  1.63it/s]
  5%|▌         | 706/13317 [07:16<2:08:52,  1.63it/s]
                                                     
{'loss': 0.3469, 'grad_norm': 0.7084166407585144, 'learning_rate': 5.300300300300301e-06, 'epoch': 0.16}

  5%|▌         | 706/13317 [07:16<2:08:52,  1.63it/s]
  5%|▌         | 707/13317 [07:17<2:08:54,  1.63it/s]
                                                     
{'loss': 0.3723, 'grad_norm': 0.6259096264839172, 'learning_rate': 5.307807807807808e-06, 'epoch': 0.16}

  5%|▌         | 707/13317 [07:17<2:08:54,  1.63it/s]
  5%|▌         | 708/13317 [07:17<2:08:53,  1.63it/s]
                                                     
{'loss': 0.3675, 'grad_norm': 0.690226674079895, 'learning_rate': 5.315315315315316e-06, 'epoch': 0.16}

  5%|▌         | 708/13317 [07:17<2:08:53,  1.63it/s]
  5%|▌         | 709/13317 [07:18<2:08:48,  1.63it/s]
                                                     
{'loss': 0.5041, 'grad_norm': 0.87501060962677, 'learning_rate': 5.322822822822823e-06, 'epoch': 0.16}

  5%|▌         | 709/13317 [07:18<2:08:48,  1.63it/s]
  5%|▌         | 710/13317 [07:18<2:08:49,  1.63it/s]
                                                     
{'loss': 0.3493, 'grad_norm': 0.6763721108436584, 'learning_rate': 5.330330330330331e-06, 'epoch': 0.16}

  5%|▌         | 710/13317 [07:19<2:08:49,  1.63it/s]
  5%|▌         | 711/13317 [07:19<2:08:49,  1.63it/s]
                                                     
{'loss': 0.3261, 'grad_norm': 0.6570829749107361, 'learning_rate': 5.337837837837838e-06, 'epoch': 0.16}

  5%|▌         | 711/13317 [07:19<2:08:49,  1.63it/s]
  5%|▌         | 712/13317 [07:20<2:08:51,  1.63it/s]
                                                     
{'loss': 0.4437, 'grad_norm': 0.6715841293334961, 'learning_rate': 5.345345345345346e-06, 'epoch': 0.16}

  5%|▌         | 712/13317 [07:20<2:08:51,  1.63it/s]
  5%|▌         | 713/13317 [07:20<2:09:06,  1.63it/s]
                                                     
{'loss': 0.3694, 'grad_norm': 0.6278824806213379, 'learning_rate': 5.3528528528528534e-06, 'epoch': 0.16}

  5%|▌         | 713/13317 [07:20<2:09:06,  1.63it/s]
  5%|▌         | 714/13317 [07:21<2:09:10,  1.63it/s]
                                                     
{'loss': 0.442, 'grad_norm': 0.6397892236709595, 'learning_rate': 5.360360360360361e-06, 'epoch': 0.16}

  5%|▌         | 714/13317 [07:21<2:09:10,  1.63it/s]
  5%|▌         | 715/13317 [07:22<2:09:11,  1.63it/s]
                                                     
{'loss': 0.4232, 'grad_norm': 0.6600443720817566, 'learning_rate': 5.367867867867869e-06, 'epoch': 0.16}

  5%|▌         | 715/13317 [07:22<2:09:11,  1.63it/s]
  5%|▌         | 716/13317 [07:22<2:09:07,  1.63it/s]
                                                     
{'loss': 0.3667, 'grad_norm': 0.7339807748794556, 'learning_rate': 5.375375375375376e-06, 'epoch': 0.16}

  5%|▌         | 716/13317 [07:22<2:09:07,  1.63it/s]
  5%|▌         | 717/13317 [07:23<2:08:58,  1.63it/s]
                                                     
{'loss': 0.3643, 'grad_norm': 0.6893627047538757, 'learning_rate': 5.382882882882884e-06, 'epoch': 0.16}

  5%|▌         | 717/13317 [07:23<2:08:58,  1.63it/s]
  5%|▌         | 718/13317 [07:23<2:09:01,  1.63it/s]
                                                     
{'loss': 0.3488, 'grad_norm': 0.6567084789276123, 'learning_rate': 5.390390390390391e-06, 'epoch': 0.16}

  5%|▌         | 718/13317 [07:23<2:09:01,  1.63it/s]
  5%|▌         | 719/13317 [07:24<2:08:57,  1.63it/s]
                                                     
{'loss': 0.3814, 'grad_norm': 0.6846104264259338, 'learning_rate': 5.397897897897898e-06, 'epoch': 0.16}

  5%|▌         | 719/13317 [07:24<2:08:57,  1.63it/s]
  5%|▌         | 720/13317 [07:25<2:08:56,  1.63it/s]
                                                     
{'loss': 0.4569, 'grad_norm': 0.7495478987693787, 'learning_rate': 5.405405405405406e-06, 'epoch': 0.16}

  5%|▌         | 720/13317 [07:25<2:08:56,  1.63it/s]
  5%|▌         | 721/13317 [07:25<2:08:52,  1.63it/s]
                                                     
{'loss': 0.3727, 'grad_norm': 0.6346465349197388, 'learning_rate': 5.412912912912913e-06, 'epoch': 0.16}

  5%|▌         | 721/13317 [07:25<2:08:52,  1.63it/s]
  5%|▌         | 722/13317 [07:26<2:08:51,  1.63it/s]
                                                     
{'loss': 0.3461, 'grad_norm': 0.6609312891960144, 'learning_rate': 5.420420420420421e-06, 'epoch': 0.16}

  5%|▌         | 722/13317 [07:26<2:08:51,  1.63it/s]
  5%|▌         | 723/13317 [07:26<2:08:57,  1.63it/s]
                                                     
{'loss': 0.3365, 'grad_norm': 0.627494215965271, 'learning_rate': 5.427927927927928e-06, 'epoch': 0.16}

  5%|▌         | 723/13317 [07:27<2:08:57,  1.63it/s]
  5%|▌         | 724/13317 [07:27<2:08:51,  1.63it/s]
                                                     
{'loss': 0.3274, 'grad_norm': 0.7485214471817017, 'learning_rate': 5.4354354354354355e-06, 'epoch': 0.16}

  5%|▌         | 724/13317 [07:27<2:08:51,  1.63it/s]
  5%|▌         | 725/13317 [07:28<2:08:50,  1.63it/s]
                                                     
{'loss': 0.3641, 'grad_norm': 0.6534873247146606, 'learning_rate': 5.4429429429429435e-06, 'epoch': 0.16}

  5%|▌         | 725/13317 [07:28<2:08:50,  1.63it/s]
  5%|▌         | 726/13317 [07:28<2:08:45,  1.63it/s]
                                                     
{'loss': 0.3003, 'grad_norm': 0.6177656054496765, 'learning_rate': 5.450450450450451e-06, 'epoch': 0.16}

  5%|▌         | 726/13317 [07:28<2:08:45,  1.63it/s]
  5%|▌         | 727/13317 [07:29<2:08:42,  1.63it/s]
                                                     
{'loss': 0.3217, 'grad_norm': 0.6563719511032104, 'learning_rate': 5.457957957957959e-06, 'epoch': 0.16}

  5%|▌         | 727/13317 [07:29<2:08:42,  1.63it/s]
  5%|▌         | 728/13317 [07:29<2:08:46,  1.63it/s]
                                                     
{'loss': 0.3467, 'grad_norm': 0.7234529256820679, 'learning_rate': 5.465465465465466e-06, 'epoch': 0.16}

  5%|▌         | 728/13317 [07:30<2:08:46,  1.63it/s]
  5%|▌         | 729/13317 [07:30<2:08:42,  1.63it/s]
                                                     
{'loss': 0.3683, 'grad_norm': 0.7340179681777954, 'learning_rate': 5.472972972972973e-06, 'epoch': 0.16}

  5%|▌         | 729/13317 [07:30<2:08:42,  1.63it/s]
  5%|▌         | 730/13317 [07:31<2:08:39,  1.63it/s]
                                                     
{'loss': 0.3968, 'grad_norm': 0.6459847688674927, 'learning_rate': 5.480480480480481e-06, 'epoch': 0.16}

  5%|▌         | 730/13317 [07:31<2:08:39,  1.63it/s]
  5%|▌         | 731/13317 [07:31<2:08:38,  1.63it/s]
                                                     
{'loss': 0.3868, 'grad_norm': 0.7490301132202148, 'learning_rate': 5.487987987987988e-06, 'epoch': 0.16}

  5%|▌         | 731/13317 [07:31<2:08:38,  1.63it/s]
  5%|▌         | 732/13317 [07:32<2:08:36,  1.63it/s]
                                                     
{'loss': 0.3661, 'grad_norm': 0.676509439945221, 'learning_rate': 5.495495495495496e-06, 'epoch': 0.16}

  5%|▌         | 732/13317 [07:32<2:08:36,  1.63it/s]
  6%|▌         | 733/13317 [07:33<2:08:36,  1.63it/s]
                                                     
{'loss': 0.4172, 'grad_norm': 0.7960578799247742, 'learning_rate': 5.503003003003003e-06, 'epoch': 0.17}

  6%|▌         | 733/13317 [07:33<2:08:36,  1.63it/s]
  6%|▌         | 734/13317 [07:33<2:08:35,  1.63it/s]
                                                     
{'loss': 0.4364, 'grad_norm': 0.6828281879425049, 'learning_rate': 5.51051051051051e-06, 'epoch': 0.17}

  6%|▌         | 734/13317 [07:33<2:08:35,  1.63it/s]
  6%|▌         | 735/13317 [07:34<2:08:32,  1.63it/s]
                                                     
{'loss': 0.3235, 'grad_norm': 0.6209385991096497, 'learning_rate': 5.518018018018018e-06, 'epoch': 0.17}

  6%|▌         | 735/13317 [07:34<2:08:32,  1.63it/s]
  6%|▌         | 736/13317 [07:34<2:08:31,  1.63it/s]
                                                     
{'loss': 0.4158, 'grad_norm': 0.7349317669868469, 'learning_rate': 5.5255255255255255e-06, 'epoch': 0.17}

  6%|▌         | 736/13317 [07:35<2:08:31,  1.63it/s]
  6%|▌         | 737/13317 [07:35<2:08:34,  1.63it/s]
                                                     
{'loss': 0.3324, 'grad_norm': 0.6702031493186951, 'learning_rate': 5.533033033033034e-06, 'epoch': 0.17}

  6%|▌         | 737/13317 [07:35<2:08:34,  1.63it/s]
  6%|▌         | 738/13317 [07:36<2:08:34,  1.63it/s]
                                                     
{'loss': 0.385, 'grad_norm': 0.8253604769706726, 'learning_rate': 5.540540540540541e-06, 'epoch': 0.17}

  6%|▌         | 738/13317 [07:36<2:08:34,  1.63it/s]
  6%|▌         | 739/13317 [07:36<2:08:35,  1.63it/s]
                                                     
{'loss': 0.4611, 'grad_norm': 0.8851189613342285, 'learning_rate': 5.548048048048048e-06, 'epoch': 0.17}

  6%|▌         | 739/13317 [07:36<2:08:35,  1.63it/s]
  6%|▌         | 740/13317 [07:37<2:08:31,  1.63it/s]
                                                     
{'loss': 0.3376, 'grad_norm': 0.6776289343833923, 'learning_rate': 5.555555555555557e-06, 'epoch': 0.17}

  6%|▌         | 740/13317 [07:37<2:08:31,  1.63it/s]
  6%|▌         | 741/13317 [07:37<2:08:34,  1.63it/s]
                                                     
{'loss': 0.3763, 'grad_norm': 0.6437105536460876, 'learning_rate': 5.563063063063063e-06, 'epoch': 0.17}

  6%|▌         | 741/13317 [07:38<2:08:34,  1.63it/s]
  6%|▌         | 742/13317 [07:38<2:08:38,  1.63it/s]
                                                     
{'loss': 0.3769, 'grad_norm': 0.7221785187721252, 'learning_rate': 5.570570570570572e-06, 'epoch': 0.17}

  6%|▌         | 742/13317 [07:38<2:08:38,  1.63it/s]
  6%|▌         | 743/13317 [07:39<2:08:33,  1.63it/s]
                                                     
{'loss': 0.3652, 'grad_norm': 0.6708287000656128, 'learning_rate': 5.578078078078079e-06, 'epoch': 0.17}

  6%|▌         | 743/13317 [07:39<2:08:33,  1.63it/s]
  6%|▌         | 744/13317 [07:39<2:08:41,  1.63it/s]
                                                     
{'loss': 0.4381, 'grad_norm': 0.7338582277297974, 'learning_rate': 5.585585585585585e-06, 'epoch': 0.17}

  6%|▌         | 744/13317 [07:39<2:08:41,  1.63it/s]
  6%|▌         | 745/13317 [07:40<2:08:39,  1.63it/s]
                                                     
{'loss': 0.3528, 'grad_norm': 0.6765159368515015, 'learning_rate': 5.593093093093094e-06, 'epoch': 0.17}

  6%|▌         | 745/13317 [07:40<2:08:39,  1.63it/s]
  6%|▌         | 746/13317 [07:41<2:08:35,  1.63it/s]
                                                     
{'loss': 0.3892, 'grad_norm': 0.6958214044570923, 'learning_rate': 5.600600600600601e-06, 'epoch': 0.17}

  6%|▌         | 746/13317 [07:41<2:08:35,  1.63it/s]
  6%|▌         | 747/13317 [07:41<2:08:34,  1.63it/s]
                                                     
{'loss': 0.4037, 'grad_norm': 0.6560460925102234, 'learning_rate': 5.608108108108109e-06, 'epoch': 0.17}

  6%|▌         | 747/13317 [07:41<2:08:34,  1.63it/s]
  6%|▌         | 748/13317 [07:42<2:08:34,  1.63it/s]
                                                     
{'loss': 0.2945, 'grad_norm': 0.5592767000198364, 'learning_rate': 5.615615615615616e-06, 'epoch': 0.17}

  6%|▌         | 748/13317 [07:42<2:08:34,  1.63it/s]
  6%|▌         | 749/13317 [07:42<2:08:33,  1.63it/s]
                                                     
{'loss': 0.3368, 'grad_norm': 0.7583639621734619, 'learning_rate': 5.623123123123123e-06, 'epoch': 0.17}

  6%|▌         | 749/13317 [07:43<2:08:33,  1.63it/s]
  6%|▌         | 750/13317 [07:43<2:08:29,  1.63it/s]
                                                     
{'loss': 0.508, 'grad_norm': 0.8161595463752747, 'learning_rate': 5.6306306306306316e-06, 'epoch': 0.17}

  6%|▌         | 750/13317 [07:43<2:08:29,  1.63it/s]
  6%|▌         | 751/13317 [07:44<2:08:28,  1.63it/s]
                                                     
{'loss': 0.3546, 'grad_norm': 0.6512833833694458, 'learning_rate': 5.638138138138139e-06, 'epoch': 0.17}

  6%|▌         | 751/13317 [07:44<2:08:28,  1.63it/s]
  6%|▌         | 752/13317 [07:44<2:08:27,  1.63it/s]
                                                     
{'loss': 0.4433, 'grad_norm': 0.7555944919586182, 'learning_rate': 5.645645645645647e-06, 'epoch': 0.17}

  6%|▌         | 752/13317 [07:44<2:08:27,  1.63it/s]
  6%|▌         | 753/13317 [07:45<2:08:28,  1.63it/s]
                                                     
{'loss': 0.2935, 'grad_norm': 0.5690192580223083, 'learning_rate': 5.653153153153154e-06, 'epoch': 0.17}

  6%|▌         | 753/13317 [07:45<2:08:28,  1.63it/s]
  6%|▌         | 754/13317 [07:45<2:08:23,  1.63it/s]
                                                     
{'loss': 0.3801, 'grad_norm': 0.851008951663971, 'learning_rate': 5.660660660660661e-06, 'epoch': 0.17}

  6%|▌         | 754/13317 [07:46<2:08:23,  1.63it/s]
  6%|▌         | 755/13317 [07:46<2:08:34,  1.63it/s]
                                                     
{'loss': 0.3834, 'grad_norm': 0.8074430227279663, 'learning_rate': 5.668168168168169e-06, 'epoch': 0.17}

  6%|▌         | 755/13317 [07:46<2:08:34,  1.63it/s]
  6%|▌         | 756/13317 [07:47<2:08:25,  1.63it/s]
                                                     
{'loss': 0.3153, 'grad_norm': 0.634027361869812, 'learning_rate': 5.675675675675676e-06, 'epoch': 0.17}

  6%|▌         | 756/13317 [07:47<2:08:25,  1.63it/s]
  6%|▌         | 757/13317 [07:47<2:08:21,  1.63it/s]
                                                     
{'loss': 0.3693, 'grad_norm': 0.7385318279266357, 'learning_rate': 5.683183183183184e-06, 'epoch': 0.17}

  6%|▌         | 757/13317 [07:47<2:08:21,  1.63it/s]
  6%|▌         | 758/13317 [07:48<2:08:20,  1.63it/s]
                                                     
{'loss': 0.3639, 'grad_norm': 0.6998007297515869, 'learning_rate': 5.690690690690691e-06, 'epoch': 0.17}

  6%|▌         | 758/13317 [07:48<2:08:20,  1.63it/s]
  6%|▌         | 759/13317 [07:49<2:08:19,  1.63it/s]
                                                     
{'loss': 0.4092, 'grad_norm': 0.7250273823738098, 'learning_rate': 5.6981981981981985e-06, 'epoch': 0.17}

  6%|▌         | 759/13317 [07:49<2:08:19,  1.63it/s]
  6%|▌         | 760/13317 [07:49<2:08:17,  1.63it/s]
                                                     
{'loss': 0.3626, 'grad_norm': 0.6068693995475769, 'learning_rate': 5.7057057057057065e-06, 'epoch': 0.17}

  6%|▌         | 760/13317 [07:49<2:08:17,  1.63it/s]
  6%|▌         | 761/13317 [07:50<2:08:19,  1.63it/s]
                                                     
{'loss': 0.343, 'grad_norm': 0.6097556352615356, 'learning_rate': 5.713213213213214e-06, 'epoch': 0.17}

  6%|▌         | 761/13317 [07:50<2:08:19,  1.63it/s]
  6%|▌         | 762/13317 [07:50<2:08:22,  1.63it/s]
                                                     
{'loss': 0.4615, 'grad_norm': 0.7616616487503052, 'learning_rate': 5.720720720720722e-06, 'epoch': 0.17}

  6%|▌         | 762/13317 [07:50<2:08:22,  1.63it/s]
  6%|▌         | 763/13317 [07:51<2:08:19,  1.63it/s]
                                                     
{'loss': 0.3936, 'grad_norm': 0.7703631520271301, 'learning_rate': 5.728228228228229e-06, 'epoch': 0.17}

  6%|▌         | 763/13317 [07:51<2:08:19,  1.63it/s]
  6%|▌         | 764/13317 [07:52<2:08:22,  1.63it/s]
                                                     
{'loss': 0.2944, 'grad_norm': 0.5981501936912537, 'learning_rate': 5.735735735735736e-06, 'epoch': 0.17}

  6%|▌         | 764/13317 [07:52<2:08:22,  1.63it/s]
  6%|▌         | 765/13317 [07:52<2:08:21,  1.63it/s]
                                                     
{'loss': 0.3167, 'grad_norm': 0.5580065846443176, 'learning_rate': 5.743243243243244e-06, 'epoch': 0.17}

  6%|▌         | 765/13317 [07:52<2:08:21,  1.63it/s]
  6%|▌         | 766/13317 [07:53<2:08:17,  1.63it/s]
                                                     
{'loss': 0.456, 'grad_norm': 0.7869643568992615, 'learning_rate': 5.750750750750751e-06, 'epoch': 0.17}

  6%|▌         | 766/13317 [07:53<2:08:17,  1.63it/s]
  6%|▌         | 767/13317 [07:53<2:08:15,  1.63it/s]
                                                     
{'loss': 0.3477, 'grad_norm': 0.6792856454849243, 'learning_rate': 5.758258258258259e-06, 'epoch': 0.17}

  6%|▌         | 767/13317 [07:54<2:08:15,  1.63it/s]
  6%|▌         | 768/13317 [07:54<2:08:17,  1.63it/s]
                                                     
{'loss': 0.3762, 'grad_norm': 0.6729170680046082, 'learning_rate': 5.765765765765766e-06, 'epoch': 0.17}

  6%|▌         | 768/13317 [07:54<2:08:17,  1.63it/s]
  6%|▌         | 769/13317 [07:55<2:08:17,  1.63it/s]
                                                     
{'loss': 0.3861, 'grad_norm': 0.7379255294799805, 'learning_rate': 5.773273273273273e-06, 'epoch': 0.17}

  6%|▌         | 769/13317 [07:55<2:08:17,  1.63it/s]
  6%|▌         | 770/13317 [07:55<2:08:14,  1.63it/s]
                                                     
{'loss': 0.3842, 'grad_norm': 0.7053853273391724, 'learning_rate': 5.780780780780781e-06, 'epoch': 0.17}

  6%|▌         | 770/13317 [07:55<2:08:14,  1.63it/s]
  6%|▌         | 771/13317 [07:56<2:08:18,  1.63it/s]
                                                     
{'loss': 0.4094, 'grad_norm': 0.8121079206466675, 'learning_rate': 5.7882882882882885e-06, 'epoch': 0.17}

  6%|▌         | 771/13317 [07:56<2:08:18,  1.63it/s]
  6%|▌         | 772/13317 [07:56<2:08:15,  1.63it/s]
                                                     
{'loss': 0.2943, 'grad_norm': 0.7223427891731262, 'learning_rate': 5.7957957957957965e-06, 'epoch': 0.17}

  6%|▌         | 772/13317 [07:57<2:08:15,  1.63it/s]
  6%|▌         | 773/13317 [07:57<2:08:15,  1.63it/s]
                                                     
{'loss': 0.3111, 'grad_norm': 0.615266740322113, 'learning_rate': 5.803303303303304e-06, 'epoch': 0.17}

  6%|▌         | 773/13317 [07:57<2:08:15,  1.63it/s]
  6%|▌         | 774/13317 [07:58<2:08:17,  1.63it/s]
                                                     
{'loss': 0.3607, 'grad_norm': 0.6517868638038635, 'learning_rate': 5.810810810810811e-06, 'epoch': 0.17}

  6%|▌         | 774/13317 [07:58<2:08:17,  1.63it/s]
  6%|▌         | 775/13317 [07:58<2:08:15,  1.63it/s]
                                                     
{'loss': 0.3808, 'grad_norm': 0.6881788372993469, 'learning_rate': 5.818318318318319e-06, 'epoch': 0.17}

  6%|▌         | 775/13317 [07:58<2:08:15,  1.63it/s]
  6%|▌         | 776/13317 [07:59<2:08:11,  1.63it/s]
                                                     
{'loss': 0.4186, 'grad_norm': 0.687802255153656, 'learning_rate': 5.825825825825826e-06, 'epoch': 0.17}

  6%|▌         | 776/13317 [07:59<2:08:11,  1.63it/s]
  6%|▌         | 777/13317 [08:00<2:08:12,  1.63it/s]
                                                     
{'loss': 0.4114, 'grad_norm': 0.6994653940200806, 'learning_rate': 5.833333333333334e-06, 'epoch': 0.18}

  6%|▌         | 777/13317 [08:00<2:08:12,  1.63it/s]
  6%|▌         | 778/13317 [08:00<2:08:11,  1.63it/s]
                                                     
{'loss': 0.3525, 'grad_norm': 0.6956170797348022, 'learning_rate': 5.840840840840841e-06, 'epoch': 0.18}

  6%|▌         | 778/13317 [08:00<2:08:11,  1.63it/s]
  6%|▌         | 779/13317 [08:01<2:08:11,  1.63it/s]
                                                     
{'loss': 0.3513, 'grad_norm': 0.6223928332328796, 'learning_rate': 5.848348348348348e-06, 'epoch': 0.18}

  6%|▌         | 779/13317 [08:01<2:08:11,  1.63it/s]
  6%|▌         | 780/13317 [08:01<2:08:10,  1.63it/s]
                                                     
{'loss': 0.4884, 'grad_norm': 0.8188555836677551, 'learning_rate': 5.855855855855856e-06, 'epoch': 0.18}

  6%|▌         | 780/13317 [08:02<2:08:10,  1.63it/s]
  6%|▌         | 781/13317 [08:02<2:08:10,  1.63it/s]
                                                     
{'loss': 0.3467, 'grad_norm': 0.7624197602272034, 'learning_rate': 5.863363363363363e-06, 'epoch': 0.18}

  6%|▌         | 781/13317 [08:02<2:08:10,  1.63it/s]
  6%|▌         | 782/13317 [08:03<2:08:07,  1.63it/s]
                                                     
{'loss': 0.3409, 'grad_norm': 0.6536375284194946, 'learning_rate': 5.870870870870871e-06, 'epoch': 0.18}

  6%|▌         | 782/13317 [08:03<2:08:07,  1.63it/s]
  6%|▌         | 783/13317 [08:03<2:08:07,  1.63it/s]
                                                     
{'loss': 0.3448, 'grad_norm': 0.6221846342086792, 'learning_rate': 5.8783783783783786e-06, 'epoch': 0.18}

  6%|▌         | 783/13317 [08:03<2:08:07,  1.63it/s]
  6%|▌         | 784/13317 [08:04<2:08:14,  1.63it/s]
                                                     
{'loss': 0.4287, 'grad_norm': 0.7807203531265259, 'learning_rate': 5.885885885885886e-06, 'epoch': 0.18}

  6%|▌         | 784/13317 [08:04<2:08:14,  1.63it/s]
  6%|▌         | 785/13317 [08:04<2:08:21,  1.63it/s]
                                                     
{'loss': 0.2693, 'grad_norm': 0.6960160136222839, 'learning_rate': 5.893393393393394e-06, 'epoch': 0.18}

  6%|▌         | 785/13317 [08:05<2:08:21,  1.63it/s]
  6%|▌         | 786/13317 [08:05<2:08:18,  1.63it/s]
                                                     
{'loss': 0.4669, 'grad_norm': 1.0643281936645508, 'learning_rate': 5.900900900900901e-06, 'epoch': 0.18}

  6%|▌         | 786/13317 [08:05<2:08:18,  1.63it/s]
  6%|▌         | 787/13317 [08:06<2:08:17,  1.63it/s]
                                                     
{'loss': 0.3894, 'grad_norm': 0.6567772030830383, 'learning_rate': 5.908408408408409e-06, 'epoch': 0.18}

  6%|▌         | 787/13317 [08:06<2:08:17,  1.63it/s]
  6%|▌         | 788/13317 [08:06<2:08:12,  1.63it/s]
                                                     
{'loss': 0.3454, 'grad_norm': 0.6543930172920227, 'learning_rate': 5.915915915915916e-06, 'epoch': 0.18}

  6%|▌         | 788/13317 [08:06<2:08:12,  1.63it/s]
  6%|▌         | 789/13317 [08:07<2:08:07,  1.63it/s]
                                                     
{'loss': 0.3495, 'grad_norm': 0.7554091811180115, 'learning_rate': 5.923423423423423e-06, 'epoch': 0.18}

  6%|▌         | 789/13317 [08:07<2:08:07,  1.63it/s]
  6%|▌         | 790/13317 [08:08<2:08:06,  1.63it/s]
                                                     
{'loss': 0.3812, 'grad_norm': 0.7074887752532959, 'learning_rate': 5.930930930930931e-06, 'epoch': 0.18}

  6%|▌         | 790/13317 [08:08<2:08:06,  1.63it/s]
  6%|▌         | 791/13317 [08:08<2:08:05,  1.63it/s]
                                                     
{'loss': 0.3311, 'grad_norm': 0.8220980763435364, 'learning_rate': 5.938438438438438e-06, 'epoch': 0.18}

  6%|▌         | 791/13317 [08:08<2:08:05,  1.63it/s]
  6%|▌         | 792/13317 [08:09<2:08:05,  1.63it/s]
                                                     
{'loss': 0.3755, 'grad_norm': 0.6586796641349792, 'learning_rate': 5.945945945945947e-06, 'epoch': 0.18}

  6%|▌         | 792/13317 [08:09<2:08:05,  1.63it/s]
  6%|▌         | 793/13317 [08:09<2:08:07,  1.63it/s]
                                                     
{'loss': 0.3929, 'grad_norm': 0.8116493821144104, 'learning_rate': 5.9534534534534534e-06, 'epoch': 0.18}

  6%|▌         | 793/13317 [08:10<2:08:07,  1.63it/s]
  6%|▌         | 794/13317 [08:10<2:08:06,  1.63it/s]
                                                     
{'loss': 0.4123, 'grad_norm': 0.6948558688163757, 'learning_rate': 5.960960960960962e-06, 'epoch': 0.18}

  6%|▌         | 794/13317 [08:10<2:08:06,  1.63it/s]
  6%|▌         | 795/13317 [08:11<2:08:04,  1.63it/s]
                                                     
{'loss': 0.3151, 'grad_norm': 0.587511420249939, 'learning_rate': 5.9684684684684694e-06, 'epoch': 0.18}

  6%|▌         | 795/13317 [08:11<2:08:04,  1.63it/s]
  6%|▌         | 796/13317 [08:11<2:08:11,  1.63it/s]
                                                     
{'loss': 0.3103, 'grad_norm': 0.5377517938613892, 'learning_rate': 5.975975975975976e-06, 'epoch': 0.18}

  6%|▌         | 796/13317 [08:11<2:08:11,  1.63it/s]
  6%|▌         | 797/13317 [08:12<2:08:06,  1.63it/s]
                                                     
{'loss': 0.3502, 'grad_norm': 0.7332735061645508, 'learning_rate': 5.983483483483485e-06, 'epoch': 0.18}

  6%|▌         | 797/13317 [08:12<2:08:06,  1.63it/s]
  6%|▌         | 798/13317 [08:12<2:08:03,  1.63it/s]
                                                     
{'loss': 0.3835, 'grad_norm': 0.6342670917510986, 'learning_rate': 5.990990990990992e-06, 'epoch': 0.18}

  6%|▌         | 798/13317 [08:13<2:08:03,  1.63it/s]
  6%|▌         | 799/13317 [08:13<2:08:01,  1.63it/s]
                                                     
{'loss': 0.3666, 'grad_norm': 0.7168092727661133, 'learning_rate': 5.9984984984985e-06, 'epoch': 0.18}

  6%|▌         | 799/13317 [08:13<2:08:01,  1.63it/s]
  6%|▌         | 800/13317 [08:14<2:08:02,  1.63it/s]
                                                     
{'loss': 0.4096, 'grad_norm': 0.6674797534942627, 'learning_rate': 6.006006006006007e-06, 'epoch': 0.18}

  6%|▌         | 800/13317 [08:14<2:08:02,  1.63it/s]
  6%|▌         | 801/13317 [08:14<2:08:04,  1.63it/s]
                                                     
{'loss': 0.3728, 'grad_norm': 0.7308337688446045, 'learning_rate': 6.013513513513514e-06, 'epoch': 0.18}

  6%|▌         | 801/13317 [08:14<2:08:04,  1.63it/s]
  6%|▌         | 802/13317 [08:15<2:07:59,  1.63it/s]
                                                     
{'loss': 0.3547, 'grad_norm': 0.6438812017440796, 'learning_rate': 6.021021021021022e-06, 'epoch': 0.18}

  6%|▌         | 802/13317 [08:15<2:07:59,  1.63it/s]
  6%|▌         | 803/13317 [08:16<2:07:56,  1.63it/s]
                                                     
{'loss': 0.4685, 'grad_norm': 0.7684497833251953, 'learning_rate': 6.028528528528529e-06, 'epoch': 0.18}

  6%|▌         | 803/13317 [08:16<2:07:56,  1.63it/s]
  6%|▌         | 804/13317 [08:16<2:07:54,  1.63it/s]
                                                     
{'loss': 0.4136, 'grad_norm': 0.8395625948905945, 'learning_rate': 6.036036036036037e-06, 'epoch': 0.18}

  6%|▌         | 804/13317 [08:16<2:07:54,  1.63it/s]
  6%|▌         | 805/13317 [08:17<2:07:52,  1.63it/s]
                                                     
{'loss': 0.4183, 'grad_norm': 0.6931245923042297, 'learning_rate': 6.043543543543544e-06, 'epoch': 0.18}

  6%|▌         | 805/13317 [08:17<2:07:52,  1.63it/s]
  6%|▌         | 806/13317 [08:17<2:07:53,  1.63it/s]
                                                     
{'loss': 0.3642, 'grad_norm': 0.6280020475387573, 'learning_rate': 6.0510510510510515e-06, 'epoch': 0.18}

  6%|▌         | 806/13317 [08:17<2:07:53,  1.63it/s]
  6%|▌         | 807/13317 [08:18<2:07:51,  1.63it/s]
                                                     
{'loss': 0.3603, 'grad_norm': 0.7074242234230042, 'learning_rate': 6.0585585585585595e-06, 'epoch': 0.18}

  6%|▌         | 807/13317 [08:18<2:07:51,  1.63it/s]
  6%|▌         | 808/13317 [08:19<2:07:51,  1.63it/s]
                                                     
{'loss': 0.3433, 'grad_norm': 0.7268837690353394, 'learning_rate': 6.066066066066067e-06, 'epoch': 0.18}

  6%|▌         | 808/13317 [08:19<2:07:51,  1.63it/s]
  6%|▌         | 809/13317 [08:19<2:07:47,  1.63it/s]
                                                     
{'loss': 0.3487, 'grad_norm': 0.6924358010292053, 'learning_rate': 6.073573573573575e-06, 'epoch': 0.18}

  6%|▌         | 809/13317 [08:19<2:07:47,  1.63it/s]
  6%|▌         | 810/13317 [08:20<2:07:47,  1.63it/s]
                                                     
{'loss': 0.4592, 'grad_norm': 0.7032753229141235, 'learning_rate': 6.081081081081082e-06, 'epoch': 0.18}

  6%|▌         | 810/13317 [08:20<2:07:47,  1.63it/s]
  6%|▌         | 811/13317 [08:20<2:07:51,  1.63it/s]
                                                     
{'loss': 0.3898, 'grad_norm': 1.028041124343872, 'learning_rate': 6.088588588588589e-06, 'epoch': 0.18}

  6%|▌         | 811/13317 [08:21<2:07:51,  1.63it/s]
  6%|▌         | 812/13317 [08:21<2:07:53,  1.63it/s]
                                                     
{'loss': 0.3265, 'grad_norm': 0.685143768787384, 'learning_rate': 6.096096096096097e-06, 'epoch': 0.18}

  6%|▌         | 812/13317 [08:21<2:07:53,  1.63it/s]
  6%|▌         | 813/13317 [08:22<2:07:48,  1.63it/s]
                                                     
{'loss': 0.3154, 'grad_norm': 0.6215029358863831, 'learning_rate': 6.103603603603604e-06, 'epoch': 0.18}

  6%|▌         | 813/13317 [08:22<2:07:48,  1.63it/s]
  6%|▌         | 814/13317 [08:22<2:07:49,  1.63it/s]
                                                     
{'loss': 0.4521, 'grad_norm': 0.7478507161140442, 'learning_rate': 6.111111111111112e-06, 'epoch': 0.18}

  6%|▌         | 814/13317 [08:22<2:07:49,  1.63it/s]
  6%|▌         | 815/13317 [08:23<2:07:49,  1.63it/s]
                                                     
{'loss': 0.3286, 'grad_norm': 0.6930565237998962, 'learning_rate': 6.118618618618619e-06, 'epoch': 0.18}

  6%|▌         | 815/13317 [08:23<2:07:49,  1.63it/s]
  6%|▌         | 816/13317 [08:23<2:07:47,  1.63it/s]
                                                     
{'loss': 0.3356, 'grad_norm': 0.6922736167907715, 'learning_rate': 6.126126126126126e-06, 'epoch': 0.18}

  6%|▌         | 816/13317 [08:24<2:07:47,  1.63it/s]
  6%|▌         | 817/13317 [08:24<2:07:44,  1.63it/s]
                                                     
{'loss': 0.3731, 'grad_norm': 0.684515655040741, 'learning_rate': 6.133633633633634e-06, 'epoch': 0.18}

  6%|▌         | 817/13317 [08:24<2:07:44,  1.63it/s]
  6%|▌         | 818/13317 [08:25<2:07:42,  1.63it/s]
                                                     
{'loss': 0.4594, 'grad_norm': 0.7336033582687378, 'learning_rate': 6.1411411411411415e-06, 'epoch': 0.18}

  6%|▌         | 818/13317 [08:25<2:07:42,  1.63it/s]
  6%|▌         | 819/13317 [08:25<2:07:43,  1.63it/s]
                                                     
{'loss': 0.4842, 'grad_norm': 0.7844893336296082, 'learning_rate': 6.1486486486486495e-06, 'epoch': 0.18}

  6%|▌         | 819/13317 [08:25<2:07:43,  1.63it/s]
  6%|▌         | 820/13317 [08:26<2:07:38,  1.63it/s]
                                                     
{'loss': 0.4052, 'grad_norm': 0.7417922019958496, 'learning_rate': 6.156156156156157e-06, 'epoch': 0.18}

  6%|▌         | 820/13317 [08:26<2:07:38,  1.63it/s]
  6%|▌         | 821/13317 [08:27<2:07:38,  1.63it/s]
                                                     
{'loss': 0.3314, 'grad_norm': 0.6744697093963623, 'learning_rate': 6.163663663663664e-06, 'epoch': 0.18}

  6%|▌         | 821/13317 [08:27<2:07:38,  1.63it/s]
  6%|▌         | 822/13317 [08:27<2:07:39,  1.63it/s]
                                                     
{'loss': 0.4433, 'grad_norm': 0.9214012622833252, 'learning_rate': 6.171171171171172e-06, 'epoch': 0.19}

  6%|▌         | 822/13317 [08:27<2:07:39,  1.63it/s]
  6%|▌         | 823/13317 [08:28<2:07:40,  1.63it/s]
                                                     
{'loss': 0.3219, 'grad_norm': 0.582699716091156, 'learning_rate': 6.178678678678679e-06, 'epoch': 0.19}

  6%|▌         | 823/13317 [08:28<2:07:40,  1.63it/s]
  6%|▌         | 824/13317 [08:28<2:07:43,  1.63it/s]
                                                     
{'loss': 0.3825, 'grad_norm': 0.7160772085189819, 'learning_rate': 6.186186186186187e-06, 'epoch': 0.19}

  6%|▌         | 824/13317 [08:29<2:07:43,  1.63it/s]
  6%|▌         | 825/13317 [08:29<2:07:42,  1.63it/s]
                                                     
{'loss': 0.3002, 'grad_norm': 0.5942308902740479, 'learning_rate': 6.193693693693694e-06, 'epoch': 0.19}

  6%|▌         | 825/13317 [08:29<2:07:42,  1.63it/s]
  6%|▌         | 826/13317 [08:30<2:07:40,  1.63it/s]
                                                     
{'loss': 0.3042, 'grad_norm': 0.7947961688041687, 'learning_rate': 6.201201201201201e-06, 'epoch': 0.19}

  6%|▌         | 826/13317 [08:30<2:07:40,  1.63it/s]
  6%|▌         | 827/13317 [08:30<2:07:42,  1.63it/s]
                                                     
{'loss': 0.3825, 'grad_norm': 0.7281845808029175, 'learning_rate': 6.208708708708709e-06, 'epoch': 0.19}

  6%|▌         | 827/13317 [08:30<2:07:42,  1.63it/s]
  6%|▌         | 828/13317 [08:31<2:08:03,  1.63it/s]
                                                     
{'loss': 0.4401, 'grad_norm': 0.6841791868209839, 'learning_rate': 6.2162162162162164e-06, 'epoch': 0.19}

  6%|▌         | 828/13317 [08:31<2:08:03,  1.63it/s]
  6%|▌         | 829/13317 [08:31<2:07:50,  1.63it/s]
                                                     
{'loss': 0.4424, 'grad_norm': 0.709362804889679, 'learning_rate': 6.2237237237237244e-06, 'epoch': 0.19}

  6%|▌         | 829/13317 [08:32<2:07:50,  1.63it/s]
  6%|▌         | 830/13317 [08:32<2:07:48,  1.63it/s]
                                                     
{'loss': 0.4161, 'grad_norm': 0.6416612267494202, 'learning_rate': 6.2312312312312316e-06, 'epoch': 0.19}

  6%|▌         | 830/13317 [08:32<2:07:48,  1.63it/s]
  6%|▌         | 831/13317 [08:33<2:07:45,  1.63it/s]
                                                     
{'loss': 0.3294, 'grad_norm': 0.6661086082458496, 'learning_rate': 6.238738738738739e-06, 'epoch': 0.19}

  6%|▌         | 831/13317 [08:33<2:07:45,  1.63it/s]
  6%|▌         | 832/13317 [08:33<2:07:37,  1.63it/s]
                                                     
{'loss': 0.3234, 'grad_norm': 0.5586583614349365, 'learning_rate': 6.246246246246247e-06, 'epoch': 0.19}

  6%|▌         | 832/13317 [08:33<2:07:37,  1.63it/s]
  6%|▋         | 833/13317 [08:34<2:07:35,  1.63it/s]
                                                     
{'loss': 0.4955, 'grad_norm': 0.7482484579086304, 'learning_rate': 6.253753753753754e-06, 'epoch': 0.19}

  6%|▋         | 833/13317 [08:34<2:07:35,  1.63it/s]
  6%|▋         | 834/13317 [08:35<2:07:40,  1.63it/s]
                                                     
{'loss': 0.3214, 'grad_norm': 0.6349591016769409, 'learning_rate': 6.261261261261262e-06, 'epoch': 0.19}

  6%|▋         | 834/13317 [08:35<2:07:40,  1.63it/s]
  6%|▋         | 835/13317 [08:35<2:07:39,  1.63it/s]
                                                     
{'loss': 0.4656, 'grad_norm': 0.8047845363616943, 'learning_rate': 6.268768768768769e-06, 'epoch': 0.19}

  6%|▋         | 835/13317 [08:35<2:07:39,  1.63it/s]
  6%|▋         | 836/13317 [08:36<2:07:38,  1.63it/s]
                                                     
{'loss': 0.3567, 'grad_norm': 0.6698337197303772, 'learning_rate': 6.276276276276276e-06, 'epoch': 0.19}

  6%|▋         | 836/13317 [08:36<2:07:38,  1.63it/s]
  6%|▋         | 837/13317 [08:36<2:07:35,  1.63it/s]
                                                     
{'loss': 0.4332, 'grad_norm': 0.6770508885383606, 'learning_rate': 6.283783783783784e-06, 'epoch': 0.19}

  6%|▋         | 837/13317 [08:37<2:07:35,  1.63it/s]
  6%|▋         | 838/13317 [08:37<2:07:54,  1.63it/s]
                                                     
{'loss': 0.4183, 'grad_norm': 0.6572591066360474, 'learning_rate': 6.291291291291291e-06, 'epoch': 0.19}

  6%|▋         | 838/13317 [08:37<2:07:54,  1.63it/s]
  6%|▋         | 839/13317 [08:38<2:07:51,  1.63it/s]
                                                     
{'loss': 0.4203, 'grad_norm': 0.6870815753936768, 'learning_rate': 6.298798798798799e-06, 'epoch': 0.19}

  6%|▋         | 839/13317 [08:38<2:07:51,  1.63it/s]
  6%|▋         | 840/13317 [08:38<2:07:47,  1.63it/s]
                                                     
{'loss': 0.3919, 'grad_norm': 0.7048009037971497, 'learning_rate': 6.3063063063063065e-06, 'epoch': 0.19}

  6%|▋         | 840/13317 [08:38<2:07:47,  1.63it/s]
  6%|▋         | 841/13317 [08:39<2:07:42,  1.63it/s]
                                                     
{'loss': 0.395, 'grad_norm': 0.6204796433448792, 'learning_rate': 6.313813813813814e-06, 'epoch': 0.19}

  6%|▋         | 841/13317 [08:39<2:07:42,  1.63it/s]
  6%|▋         | 842/13317 [08:39<2:07:37,  1.63it/s]
                                                     
{'loss': 0.3997, 'grad_norm': 0.7136670351028442, 'learning_rate': 6.321321321321322e-06, 'epoch': 0.19}

  6%|▋         | 842/13317 [08:40<2:07:37,  1.63it/s]
  6%|▋         | 843/13317 [08:40<2:07:44,  1.63it/s]
                                                     
{'loss': 0.3836, 'grad_norm': 0.8171271681785583, 'learning_rate': 6.328828828828829e-06, 'epoch': 0.19}

  6%|▋         | 843/13317 [08:40<2:07:44,  1.63it/s]
  6%|▋         | 844/13317 [08:41<2:07:38,  1.63it/s]
                                                     
{'loss': 0.321, 'grad_norm': 0.65977942943573, 'learning_rate': 6.336336336336338e-06, 'epoch': 0.19}

  6%|▋         | 844/13317 [08:41<2:07:38,  1.63it/s]
  6%|▋         | 845/13317 [08:41<2:07:35,  1.63it/s]
                                                     
{'loss': 0.376, 'grad_norm': 0.7031236290931702, 'learning_rate': 6.343843843843844e-06, 'epoch': 0.19}

  6%|▋         | 845/13317 [08:41<2:07:35,  1.63it/s]
  6%|▋         | 846/13317 [08:42<2:07:34,  1.63it/s]
                                                     
{'loss': 0.3593, 'grad_norm': 0.6345517039299011, 'learning_rate': 6.351351351351351e-06, 'epoch': 0.19}

  6%|▋         | 846/13317 [08:42<2:07:34,  1.63it/s]
  6%|▋         | 847/13317 [08:43<2:07:30,  1.63it/s]
                                                     
{'loss': 0.3499, 'grad_norm': 0.7526505589485168, 'learning_rate': 6.35885885885886e-06, 'epoch': 0.19}

  6%|▋         | 847/13317 [08:43<2:07:30,  1.63it/s]
  6%|▋         | 848/13317 [08:43<2:07:29,  1.63it/s]
                                                     
{'loss': 0.4668, 'grad_norm': 0.7832536697387695, 'learning_rate': 6.366366366366366e-06, 'epoch': 0.19}

  6%|▋         | 848/13317 [08:43<2:07:29,  1.63it/s]
  6%|▋         | 849/13317 [08:44<2:07:31,  1.63it/s]
                                                     
{'loss': 0.361, 'grad_norm': 0.6706039905548096, 'learning_rate': 6.373873873873875e-06, 'epoch': 0.19}

  6%|▋         | 849/13317 [08:44<2:07:31,  1.63it/s]
  6%|▋         | 850/13317 [08:44<2:07:31,  1.63it/s]
                                                     
{'loss': 0.3681, 'grad_norm': 0.6352694630622864, 'learning_rate': 6.381381381381382e-06, 'epoch': 0.19}

  6%|▋         | 850/13317 [08:44<2:07:31,  1.63it/s]
  6%|▋         | 851/13317 [08:45<2:07:31,  1.63it/s]
                                                     
{'loss': 0.3878, 'grad_norm': 0.6255568861961365, 'learning_rate': 6.3888888888888885e-06, 'epoch': 0.19}

  6%|▋         | 851/13317 [08:45<2:07:31,  1.63it/s]
  6%|▋         | 852/13317 [08:46<2:07:29,  1.63it/s]
                                                     
{'loss': 0.379, 'grad_norm': 0.6917126774787903, 'learning_rate': 6.396396396396397e-06, 'epoch': 0.19}

  6%|▋         | 852/13317 [08:46<2:07:29,  1.63it/s]
  6%|▋         | 853/13317 [08:46<2:07:31,  1.63it/s]
                                                     
{'loss': 0.3398, 'grad_norm': 0.6805317997932434, 'learning_rate': 6.4039039039039045e-06, 'epoch': 0.19}

  6%|▋         | 853/13317 [08:46<2:07:31,  1.63it/s]
  6%|▋         | 854/13317 [08:47<2:07:28,  1.63it/s]
                                                     
{'loss': 0.3714, 'grad_norm': 0.8721371293067932, 'learning_rate': 6.4114114114114125e-06, 'epoch': 0.19}

  6%|▋         | 854/13317 [08:47<2:07:28,  1.63it/s]
  6%|▋         | 855/13317 [08:47<2:07:26,  1.63it/s]
                                                     
{'loss': 0.3648, 'grad_norm': 0.7038485407829285, 'learning_rate': 6.41891891891892e-06, 'epoch': 0.19}

  6%|▋         | 855/13317 [08:48<2:07:26,  1.63it/s]
  6%|▋         | 856/13317 [08:48<2:07:25,  1.63it/s]
                                                     
{'loss': 0.3369, 'grad_norm': 0.8113279938697815, 'learning_rate': 6.426426426426427e-06, 'epoch': 0.19}

  6%|▋         | 856/13317 [08:48<2:07:25,  1.63it/s]
  6%|▋         | 857/13317 [08:49<2:07:22,  1.63it/s]
                                                     
{'loss': 0.4513, 'grad_norm': 0.7493097186088562, 'learning_rate': 6.433933933933935e-06, 'epoch': 0.19}

  6%|▋         | 857/13317 [08:49<2:07:22,  1.63it/s]
  6%|▋         | 858/13317 [08:49<2:07:25,  1.63it/s]
                                                     
{'loss': 0.3825, 'grad_norm': 0.6996623277664185, 'learning_rate': 6.441441441441442e-06, 'epoch': 0.19}

  6%|▋         | 858/13317 [08:49<2:07:25,  1.63it/s]
  6%|▋         | 859/13317 [08:50<2:07:25,  1.63it/s]
                                                     
{'loss': 0.4165, 'grad_norm': 0.7512524127960205, 'learning_rate': 6.44894894894895e-06, 'epoch': 0.19}

  6%|▋         | 859/13317 [08:50<2:07:25,  1.63it/s]
  6%|▋         | 860/13317 [08:50<2:07:19,  1.63it/s]
                                                     
{'loss': 0.4781, 'grad_norm': 0.7439413666725159, 'learning_rate': 6.456456456456457e-06, 'epoch': 0.19}

  6%|▋         | 860/13317 [08:51<2:07:19,  1.63it/s]
  6%|▋         | 861/13317 [08:51<2:07:22,  1.63it/s]
                                                     
{'loss': 0.3196, 'grad_norm': 0.6679162979125977, 'learning_rate': 6.463963963963964e-06, 'epoch': 0.19}

  6%|▋         | 861/13317 [08:51<2:07:22,  1.63it/s]
  6%|▋         | 862/13317 [08:52<2:07:19,  1.63it/s]
                                                     
{'loss': 0.3663, 'grad_norm': 0.6537676453590393, 'learning_rate': 6.471471471471472e-06, 'epoch': 0.19}

  6%|▋         | 862/13317 [08:52<2:07:19,  1.63it/s]
  6%|▋         | 863/13317 [08:52<2:07:21,  1.63it/s]
                                                     
{'loss': 0.2758, 'grad_norm': 0.5566006898880005, 'learning_rate': 6.478978978978979e-06, 'epoch': 0.19}

  6%|▋         | 863/13317 [08:52<2:07:21,  1.63it/s]
  6%|▋         | 864/13317 [08:53<2:07:25,  1.63it/s]
                                                     
{'loss': 0.3664, 'grad_norm': 0.664745032787323, 'learning_rate': 6.486486486486487e-06, 'epoch': 0.19}

  6%|▋         | 864/13317 [08:53<2:07:25,  1.63it/s]
  6%|▋         | 865/13317 [08:54<2:07:21,  1.63it/s]
                                                     
{'loss': 0.3892, 'grad_norm': 0.834060788154602, 'learning_rate': 6.4939939939939946e-06, 'epoch': 0.19}

  6%|▋         | 865/13317 [08:54<2:07:21,  1.63it/s]
  7%|▋         | 866/13317 [08:54<2:07:17,  1.63it/s]
                                                     
{'loss': 0.4642, 'grad_norm': 0.6734587550163269, 'learning_rate': 6.501501501501502e-06, 'epoch': 0.2}

  7%|▋         | 866/13317 [08:54<2:07:17,  1.63it/s]
  7%|▋         | 867/13317 [08:55<2:07:24,  1.63it/s]
                                                     
{'loss': 0.3089, 'grad_norm': 0.7000230550765991, 'learning_rate': 6.50900900900901e-06, 'epoch': 0.2}

  7%|▋         | 867/13317 [08:55<2:07:24,  1.63it/s]
  7%|▋         | 868/13317 [08:55<2:07:19,  1.63it/s]
                                                     
{'loss': 0.3944, 'grad_norm': 0.8035679459571838, 'learning_rate': 6.516516516516517e-06, 'epoch': 0.2}

  7%|▋         | 868/13317 [08:56<2:07:19,  1.63it/s]
  7%|▋         | 869/13317 [08:56<2:07:17,  1.63it/s]
                                                     
{'loss': 0.434, 'grad_norm': 0.8334922790527344, 'learning_rate': 6.524024024024025e-06, 'epoch': 0.2}

  7%|▋         | 869/13317 [08:56<2:07:17,  1.63it/s]
  7%|▋         | 870/13317 [08:57<2:07:24,  1.63it/s]
                                                     
{'loss': 0.3385, 'grad_norm': 0.7081919312477112, 'learning_rate': 6.531531531531532e-06, 'epoch': 0.2}

  7%|▋         | 870/13317 [08:57<2:07:24,  1.63it/s]
  7%|▋         | 871/13317 [08:57<2:07:21,  1.63it/s]
                                                     
{'loss': 0.4139, 'grad_norm': 0.8252947330474854, 'learning_rate': 6.539039039039039e-06, 'epoch': 0.2}

  7%|▋         | 871/13317 [08:57<2:07:21,  1.63it/s]
  7%|▋         | 872/13317 [08:58<2:07:25,  1.63it/s]
                                                     
{'loss': 0.4622, 'grad_norm': 0.7968078851699829, 'learning_rate': 6.546546546546547e-06, 'epoch': 0.2}

  7%|▋         | 872/13317 [08:58<2:07:25,  1.63it/s]
  7%|▋         | 873/13317 [08:58<2:07:27,  1.63it/s]
                                                     
{'loss': 0.2839, 'grad_norm': 0.6121948957443237, 'learning_rate': 6.554054054054054e-06, 'epoch': 0.2}

  7%|▋         | 873/13317 [08:59<2:07:27,  1.63it/s]
  7%|▋         | 874/13317 [08:59<2:07:21,  1.63it/s]
                                                     
{'loss': 0.4673, 'grad_norm': 0.9301978945732117, 'learning_rate': 6.561561561561562e-06, 'epoch': 0.2}

  7%|▋         | 874/13317 [08:59<2:07:21,  1.63it/s]
  7%|▋         | 875/13317 [09:00<2:07:15,  1.63it/s]
                                                     
{'loss': 0.3648, 'grad_norm': 0.660483181476593, 'learning_rate': 6.5690690690690695e-06, 'epoch': 0.2}

  7%|▋         | 875/13317 [09:00<2:07:15,  1.63it/s]
  7%|▋         | 876/13317 [09:00<2:07:18,  1.63it/s]
                                                     
{'loss': 0.4286, 'grad_norm': 0.568031370639801, 'learning_rate': 6.5765765765765775e-06, 'epoch': 0.2}

  7%|▋         | 876/13317 [09:00<2:07:18,  1.63it/s]
  7%|▋         | 877/13317 [09:01<2:07:15,  1.63it/s]
                                                     
{'loss': 0.4643, 'grad_norm': 0.6859811544418335, 'learning_rate': 6.584084084084085e-06, 'epoch': 0.2}

  7%|▋         | 877/13317 [09:01<2:07:15,  1.63it/s]
  7%|▋         | 878/13317 [09:02<2:07:13,  1.63it/s]
                                                     
{'loss': 0.3797, 'grad_norm': 0.6926259994506836, 'learning_rate': 6.591591591591592e-06, 'epoch': 0.2}

  7%|▋         | 878/13317 [09:02<2:07:13,  1.63it/s]
  7%|▋         | 879/13317 [09:02<2:07:07,  1.63it/s]
                                                     
{'loss': 0.295, 'grad_norm': 0.6819000840187073, 'learning_rate': 6.5990990990991e-06, 'epoch': 0.2}

  7%|▋         | 879/13317 [09:02<2:07:07,  1.63it/s]
  7%|▋         | 880/13317 [09:03<2:07:03,  1.63it/s]
                                                     
{'loss': 0.3715, 'grad_norm': 0.7018037438392639, 'learning_rate': 6.606606606606607e-06, 'epoch': 0.2}

  7%|▋         | 880/13317 [09:03<2:07:03,  1.63it/s]
  7%|▋         | 881/13317 [09:03<2:07:08,  1.63it/s]
                                                     
{'loss': 0.3204, 'grad_norm': 0.6260318160057068, 'learning_rate': 6.614114114114115e-06, 'epoch': 0.2}

  7%|▋         | 881/13317 [09:04<2:07:08,  1.63it/s]
  7%|▋         | 882/13317 [09:04<2:07:05,  1.63it/s]
                                                     
{'loss': 0.3427, 'grad_norm': 0.673068106174469, 'learning_rate': 6.621621621621622e-06, 'epoch': 0.2}

  7%|▋         | 882/13317 [09:04<2:07:05,  1.63it/s]
  7%|▋         | 883/13317 [09:05<2:07:04,  1.63it/s]
                                                     
{'loss': 0.2613, 'grad_norm': 0.6451634168624878, 'learning_rate': 6.629129129129129e-06, 'epoch': 0.2}

  7%|▋         | 883/13317 [09:05<2:07:04,  1.63it/s]
  7%|▋         | 884/13317 [09:05<2:07:06,  1.63it/s]
                                                     
{'loss': 0.3075, 'grad_norm': 0.6564715504646301, 'learning_rate': 6.636636636636637e-06, 'epoch': 0.2}

  7%|▋         | 884/13317 [09:05<2:07:06,  1.63it/s]
  7%|▋         | 885/13317 [09:06<2:07:03,  1.63it/s]
                                                     
{'loss': 0.3426, 'grad_norm': 0.6353038549423218, 'learning_rate': 6.644144144144144e-06, 'epoch': 0.2}

  7%|▋         | 885/13317 [09:06<2:07:03,  1.63it/s]
  7%|▋         | 886/13317 [09:06<2:07:16,  1.63it/s]
                                                     
{'loss': 0.3691, 'grad_norm': 0.7826282382011414, 'learning_rate': 6.651651651651652e-06, 'epoch': 0.2}

  7%|▋         | 886/13317 [09:07<2:07:16,  1.63it/s]
  7%|▋         | 887/13317 [09:07<2:07:10,  1.63it/s]
                                                     
{'loss': 0.3335, 'grad_norm': 0.61937415599823, 'learning_rate': 6.6591591591591595e-06, 'epoch': 0.2}

  7%|▋         | 887/13317 [09:07<2:07:10,  1.63it/s]
  7%|▋         | 888/13317 [09:08<2:07:10,  1.63it/s]
                                                     
{'loss': 0.3395, 'grad_norm': 0.6862392425537109, 'learning_rate': 6.666666666666667e-06, 'epoch': 0.2}

  7%|▋         | 888/13317 [09:08<2:07:10,  1.63it/s]
  7%|▋         | 889/13317 [09:08<2:07:18,  1.63it/s]
                                                     
{'loss': 0.3357, 'grad_norm': 0.7673256397247314, 'learning_rate': 6.674174174174175e-06, 'epoch': 0.2}

  7%|▋         | 889/13317 [09:08<2:07:18,  1.63it/s]
  7%|▋         | 890/13317 [09:09<2:07:11,  1.63it/s]
                                                     
{'loss': 0.3437, 'grad_norm': 0.6050322651863098, 'learning_rate': 6.681681681681682e-06, 'epoch': 0.2}

  7%|▋         | 890/13317 [09:09<2:07:11,  1.63it/s]
  7%|▋         | 891/13317 [09:10<2:07:05,  1.63it/s]
                                                     
{'loss': 0.3718, 'grad_norm': 1.0353114604949951, 'learning_rate': 6.689189189189191e-06, 'epoch': 0.2}

  7%|▋         | 891/13317 [09:10<2:07:05,  1.63it/s]
  7%|▋         | 892/13317 [09:10<2:07:03,  1.63it/s]
                                                     
{'loss': 0.3192, 'grad_norm': 0.628167986869812, 'learning_rate': 6.696696696696697e-06, 'epoch': 0.2}

  7%|▋         | 892/13317 [09:10<2:07:03,  1.63it/s]
  7%|▋         | 893/13317 [09:11<2:07:16,  1.63it/s]
                                                     
{'loss': 0.3414, 'grad_norm': 0.6454979181289673, 'learning_rate': 6.704204204204204e-06, 'epoch': 0.2}

  7%|▋         | 893/13317 [09:11<2:07:16,  1.63it/s]
  7%|▋         | 894/13317 [09:11<2:07:09,  1.63it/s]
                                                     
{'loss': 0.3012, 'grad_norm': 0.6663264632225037, 'learning_rate': 6.711711711711713e-06, 'epoch': 0.2}

  7%|▋         | 894/13317 [09:11<2:07:09,  1.63it/s]
  7%|▋         | 895/13317 [09:12<2:07:08,  1.63it/s]
                                                     
{'loss': 0.3235, 'grad_norm': 0.7167181968688965, 'learning_rate': 6.719219219219219e-06, 'epoch': 0.2}

  7%|▋         | 895/13317 [09:12<2:07:08,  1.63it/s]
  7%|▋         | 896/13317 [09:13<2:07:06,  1.63it/s]
                                                     
{'loss': 0.3138, 'grad_norm': 0.6217464208602905, 'learning_rate': 6.726726726726728e-06, 'epoch': 0.2}

  7%|▋         | 896/13317 [09:13<2:07:06,  1.63it/s]
  7%|▋         | 897/13317 [09:13<2:07:06,  1.63it/s]
                                                     
{'loss': 0.4043, 'grad_norm': 1.0593585968017578, 'learning_rate': 6.734234234234235e-06, 'epoch': 0.2}

  7%|▋         | 897/13317 [09:13<2:07:06,  1.63it/s]
  7%|▋         | 898/13317 [09:14<2:07:13,  1.63it/s]
                                                     
{'loss': 0.2684, 'grad_norm': 0.6224299073219299, 'learning_rate': 6.7417417417417415e-06, 'epoch': 0.2}

  7%|▋         | 898/13317 [09:14<2:07:13,  1.63it/s]
  7%|▋         | 899/13317 [09:14<2:07:12,  1.63it/s]
                                                     
{'loss': 0.4188, 'grad_norm': 0.834233283996582, 'learning_rate': 6.74924924924925e-06, 'epoch': 0.2}

  7%|▋         | 899/13317 [09:15<2:07:12,  1.63it/s]
  7%|▋         | 900/13317 [09:15<2:07:05,  1.63it/s]
                                                     
{'loss': 0.4456, 'grad_norm': 0.7499524354934692, 'learning_rate': 6.7567567567567575e-06, 'epoch': 0.2}

  7%|▋         | 900/13317 [09:15<2:07:05,  1.63it/s]
  7%|▋         | 901/13317 [09:16<2:07:13,  1.63it/s]
                                                     
{'loss': 0.372, 'grad_norm': 0.6431075930595398, 'learning_rate': 6.7642642642642655e-06, 'epoch': 0.2}

  7%|▋         | 901/13317 [09:16<2:07:13,  1.63it/s]
  7%|▋         | 902/13317 [09:16<2:07:05,  1.63it/s]
                                                     
{'loss': 0.3709, 'grad_norm': 0.783943235874176, 'learning_rate': 6.771771771771773e-06, 'epoch': 0.2}

  7%|▋         | 902/13317 [09:16<2:07:05,  1.63it/s]
  7%|▋         | 903/13317 [09:17<2:07:11,  1.63it/s]
                                                     
{'loss': 0.3134, 'grad_norm': 0.6468042731285095, 'learning_rate': 6.77927927927928e-06, 'epoch': 0.2}

  7%|▋         | 903/13317 [09:17<2:07:11,  1.63it/s]
  7%|▋         | 904/13317 [09:18<2:07:05,  1.63it/s]
                                                     
{'loss': 0.3712, 'grad_norm': 0.6579153537750244, 'learning_rate': 6.786786786786788e-06, 'epoch': 0.2}

  7%|▋         | 904/13317 [09:18<2:07:05,  1.63it/s]
  7%|▋         | 905/13317 [09:18<2:06:59,  1.63it/s]
                                                     
{'loss': 0.3674, 'grad_norm': 0.7060410380363464, 'learning_rate': 6.794294294294295e-06, 'epoch': 0.2}

  7%|▋         | 905/13317 [09:18<2:06:59,  1.63it/s]
  7%|▋         | 906/13317 [09:19<2:06:56,  1.63it/s]
                                                     
{'loss': 0.3261, 'grad_norm': 0.7102665901184082, 'learning_rate': 6.801801801801803e-06, 'epoch': 0.2}

  7%|▋         | 906/13317 [09:19<2:06:56,  1.63it/s]
  7%|▋         | 907/13317 [09:19<2:06:56,  1.63it/s]
                                                     
{'loss': 0.3432, 'grad_norm': 0.7670366168022156, 'learning_rate': 6.80930930930931e-06, 'epoch': 0.2}

  7%|▋         | 907/13317 [09:19<2:06:56,  1.63it/s]
  7%|▋         | 908/13317 [09:20<2:06:56,  1.63it/s]
                                                     
{'loss': 0.345, 'grad_norm': 0.7242881655693054, 'learning_rate': 6.816816816816817e-06, 'epoch': 0.2}

  7%|▋         | 908/13317 [09:20<2:06:56,  1.63it/s]
  7%|▋         | 909/13317 [09:21<2:06:51,  1.63it/s]
                                                     
{'loss': 0.3624, 'grad_norm': 0.7889056205749512, 'learning_rate': 6.824324324324325e-06, 'epoch': 0.2}

  7%|▋         | 909/13317 [09:21<2:06:51,  1.63it/s]
  7%|▋         | 910/13317 [09:21<2:06:50,  1.63it/s]
                                                     
{'loss': 0.5939, 'grad_norm': 0.8242616057395935, 'learning_rate': 6.8318318318318324e-06, 'epoch': 0.21}

  7%|▋         | 910/13317 [09:21<2:06:50,  1.63it/s]
  7%|▋         | 911/13317 [09:22<2:06:54,  1.63it/s]
                                                     
{'loss': 0.3398, 'grad_norm': 0.6732015013694763, 'learning_rate': 6.8393393393393404e-06, 'epoch': 0.21}

  7%|▋         | 911/13317 [09:22<2:06:54,  1.63it/s]
  7%|▋         | 912/13317 [09:22<2:06:53,  1.63it/s]
                                                     
{'loss': 0.3611, 'grad_norm': 0.7651621103286743, 'learning_rate': 6.846846846846848e-06, 'epoch': 0.21}

  7%|▋         | 912/13317 [09:23<2:06:53,  1.63it/s]
  7%|▋         | 913/13317 [09:23<2:06:56,  1.63it/s]
                                                     
{'loss': 0.3729, 'grad_norm': 0.744815468788147, 'learning_rate': 6.854354354354355e-06, 'epoch': 0.21}

  7%|▋         | 913/13317 [09:23<2:06:56,  1.63it/s]
  7%|▋         | 914/13317 [09:24<2:06:54,  1.63it/s]
                                                     
{'loss': 0.3642, 'grad_norm': 0.8144098520278931, 'learning_rate': 6.861861861861863e-06, 'epoch': 0.21}

  7%|▋         | 914/13317 [09:24<2:06:54,  1.63it/s]
  7%|▋         | 915/13317 [09:24<2:06:58,  1.63it/s]
                                                     
{'loss': 0.3514, 'grad_norm': 0.6842818856239319, 'learning_rate': 6.86936936936937e-06, 'epoch': 0.21}

  7%|▋         | 915/13317 [09:24<2:06:58,  1.63it/s]
  7%|▋         | 916/13317 [09:25<2:07:04,  1.63it/s]
                                                     
{'loss': 0.3394, 'grad_norm': 0.6827661991119385, 'learning_rate': 6.876876876876878e-06, 'epoch': 0.21}

  7%|▋         | 916/13317 [09:25<2:07:04,  1.63it/s]
  7%|▋         | 917/13317 [09:25<2:06:56,  1.63it/s]
                                                     
{'loss': 0.4383, 'grad_norm': 0.8249902725219727, 'learning_rate': 6.884384384384385e-06, 'epoch': 0.21}

  7%|▋         | 917/13317 [09:26<2:06:56,  1.63it/s]
  7%|▋         | 918/13317 [09:26<2:06:47,  1.63it/s]
                                                     
{'loss': 0.3643, 'grad_norm': 0.6308428645133972, 'learning_rate': 6.891891891891892e-06, 'epoch': 0.21}

  7%|▋         | 918/13317 [09:26<2:06:47,  1.63it/s]
  7%|▋         | 919/13317 [09:27<2:06:49,  1.63it/s]
                                                     
{'loss': 0.414, 'grad_norm': 0.687248170375824, 'learning_rate': 6.8993993993994e-06, 'epoch': 0.21}

  7%|▋         | 919/13317 [09:27<2:06:49,  1.63it/s]
  7%|▋         | 920/13317 [09:27<2:06:47,  1.63it/s]
                                                     
{'loss': 0.3284, 'grad_norm': 0.5704975724220276, 'learning_rate': 6.906906906906907e-06, 'epoch': 0.21}

  7%|▋         | 920/13317 [09:27<2:06:47,  1.63it/s]
  7%|▋         | 921/13317 [09:28<2:06:51,  1.63it/s]
                                                     
{'loss': 0.3539, 'grad_norm': 0.6432682275772095, 'learning_rate': 6.914414414414415e-06, 'epoch': 0.21}

  7%|▋         | 921/13317 [09:28<2:06:51,  1.63it/s]
  7%|▋         | 922/13317 [09:29<2:06:49,  1.63it/s]
                                                     
{'loss': 0.3614, 'grad_norm': 0.6627680659294128, 'learning_rate': 6.9219219219219225e-06, 'epoch': 0.21}

  7%|▋         | 922/13317 [09:29<2:06:49,  1.63it/s]
  7%|▋         | 923/13317 [09:29<2:06:51,  1.63it/s]
                                                     
{'loss': 0.3472, 'grad_norm': 0.6721769571304321, 'learning_rate': 6.92942942942943e-06, 'epoch': 0.21}

  7%|▋         | 923/13317 [09:29<2:06:51,  1.63it/s]
  7%|▋         | 924/13317 [09:30<2:06:48,  1.63it/s]
                                                     
{'loss': 0.3064, 'grad_norm': 0.645616352558136, 'learning_rate': 6.936936936936938e-06, 'epoch': 0.21}

  7%|▋         | 924/13317 [09:30<2:06:48,  1.63it/s]
  7%|▋         | 925/13317 [09:30<2:06:45,  1.63it/s]
                                                     
{'loss': 0.3687, 'grad_norm': 0.7163966298103333, 'learning_rate': 6.944444444444445e-06, 'epoch': 0.21}

  7%|▋         | 925/13317 [09:31<2:06:45,  1.63it/s]
  7%|▋         | 926/13317 [09:31<2:06:52,  1.63it/s]
                                                     
{'loss': 0.4255, 'grad_norm': 0.8482557535171509, 'learning_rate': 6.951951951951953e-06, 'epoch': 0.21}

  7%|▋         | 926/13317 [09:31<2:06:52,  1.63it/s]
  7%|▋         | 927/13317 [09:32<2:06:50,  1.63it/s]
                                                     
{'loss': 0.3705, 'grad_norm': 0.8032557964324951, 'learning_rate': 6.95945945945946e-06, 'epoch': 0.21}

  7%|▋         | 927/13317 [09:32<2:06:50,  1.63it/s]
  7%|▋         | 928/13317 [09:32<2:06:48,  1.63it/s]
                                                     
{'loss': 0.229, 'grad_norm': 0.4821937084197998, 'learning_rate': 6.966966966966967e-06, 'epoch': 0.21}

  7%|▋         | 928/13317 [09:32<2:06:48,  1.63it/s]
  7%|▋         | 929/13317 [09:33<2:06:49,  1.63it/s]
                                                     
{'loss': 0.3223, 'grad_norm': 0.621852457523346, 'learning_rate': 6.974474474474475e-06, 'epoch': 0.21}

  7%|▋         | 929/13317 [09:33<2:06:49,  1.63it/s]
  7%|▋         | 930/13317 [09:33<2:06:43,  1.63it/s]
                                                     
{'loss': 0.2893, 'grad_norm': 0.661959171295166, 'learning_rate': 6.981981981981982e-06, 'epoch': 0.21}

  7%|▋         | 930/13317 [09:34<2:06:43,  1.63it/s]
  7%|▋         | 931/13317 [09:34<2:06:48,  1.63it/s]
                                                     
{'loss': 0.3337, 'grad_norm': 0.7707591652870178, 'learning_rate': 6.98948948948949e-06, 'epoch': 0.21}

  7%|▋         | 931/13317 [09:34<2:06:48,  1.63it/s]
  7%|▋         | 932/13317 [09:35<2:06:46,  1.63it/s]
                                                     
{'loss': 0.376, 'grad_norm': 0.6556852459907532, 'learning_rate': 6.996996996996997e-06, 'epoch': 0.21}

  7%|▋         | 932/13317 [09:35<2:06:46,  1.63it/s]
  7%|▋         | 933/13317 [09:35<2:06:48,  1.63it/s]
                                                     
{'loss': 0.3193, 'grad_norm': 0.5875115394592285, 'learning_rate': 7.0045045045045045e-06, 'epoch': 0.21}

  7%|▋         | 933/13317 [09:35<2:06:48,  1.63it/s]
  7%|▋         | 934/13317 [09:36<2:06:40,  1.63it/s]
                                                     
{'loss': 0.405, 'grad_norm': 0.6346341371536255, 'learning_rate': 7.0120120120120125e-06, 'epoch': 0.21}

  7%|▋         | 934/13317 [09:36<2:06:40,  1.63it/s]
  7%|▋         | 935/13317 [09:37<2:06:37,  1.63it/s]
                                                     
{'loss': 0.2805, 'grad_norm': 0.6096067428588867, 'learning_rate': 7.01951951951952e-06, 'epoch': 0.21}

  7%|▋         | 935/13317 [09:37<2:06:37,  1.63it/s]
  7%|▋         | 936/13317 [09:37<2:06:43,  1.63it/s]
                                                     
{'loss': 0.4508, 'grad_norm': 0.7483050227165222, 'learning_rate': 7.027027027027028e-06, 'epoch': 0.21}

  7%|▋         | 936/13317 [09:37<2:06:43,  1.63it/s]
  7%|▋         | 937/13317 [09:38<2:06:40,  1.63it/s]
                                                     
{'loss': 0.3715, 'grad_norm': 0.7354792356491089, 'learning_rate': 7.034534534534535e-06, 'epoch': 0.21}

  7%|▋         | 937/13317 [09:38<2:06:40,  1.63it/s]
  7%|▋         | 938/13317 [09:38<2:06:42,  1.63it/s]
                                                     
{'loss': 0.3731, 'grad_norm': 0.7064424157142639, 'learning_rate': 7.042042042042042e-06, 'epoch': 0.21}

  7%|▋         | 938/13317 [09:39<2:06:42,  1.63it/s]
  7%|▋         | 939/13317 [09:39<2:06:42,  1.63it/s]
                                                     
{'loss': 0.3328, 'grad_norm': 0.7056183815002441, 'learning_rate': 7.04954954954955e-06, 'epoch': 0.21}

  7%|▋         | 939/13317 [09:39<2:06:42,  1.63it/s]
  7%|▋         | 940/13317 [09:40<2:06:38,  1.63it/s]
                                                     
{'loss': 0.3572, 'grad_norm': 0.6621873378753662, 'learning_rate': 7.057057057057057e-06, 'epoch': 0.21}

  7%|▋         | 940/13317 [09:40<2:06:38,  1.63it/s]
  7%|▋         | 941/13317 [09:40<2:06:45,  1.63it/s]
                                                     
{'loss': 0.3594, 'grad_norm': 0.7086957097053528, 'learning_rate': 7.064564564564565e-06, 'epoch': 0.21}

  7%|▋         | 941/13317 [09:40<2:06:45,  1.63it/s]
  7%|▋         | 942/13317 [09:41<2:06:41,  1.63it/s]
                                                     
{'loss': 0.3556, 'grad_norm': 0.6905097365379333, 'learning_rate': 7.072072072072072e-06, 'epoch': 0.21}

  7%|▋         | 942/13317 [09:41<2:06:41,  1.63it/s]
  7%|▋         | 943/13317 [09:41<2:06:37,  1.63it/s]
                                                     
{'loss': 0.311, 'grad_norm': 0.5690462589263916, 'learning_rate': 7.079579579579579e-06, 'epoch': 0.21}

  7%|▋         | 943/13317 [09:42<2:06:37,  1.63it/s]
  7%|▋         | 944/13317 [09:42<2:06:36,  1.63it/s]
                                                     
{'loss': 0.4679, 'grad_norm': 0.7635220289230347, 'learning_rate': 7.087087087087087e-06, 'epoch': 0.21}

  7%|▋         | 944/13317 [09:42<2:06:36,  1.63it/s]
  7%|▋         | 945/13317 [09:43<2:06:33,  1.63it/s]
                                                     
{'loss': 0.41, 'grad_norm': 0.789139449596405, 'learning_rate': 7.0945945945945946e-06, 'epoch': 0.21}

  7%|▋         | 945/13317 [09:43<2:06:33,  1.63it/s]
  7%|▋         | 946/13317 [09:43<2:07:15,  1.62it/s]
                                                     
{'loss': 0.2672, 'grad_norm': 0.5868982076644897, 'learning_rate': 7.102102102102103e-06, 'epoch': 0.21}

  7%|▋         | 946/13317 [09:43<2:07:15,  1.62it/s]
  7%|▋         | 947/13317 [09:44<2:07:01,  1.62it/s]
                                                     
{'loss': 0.3493, 'grad_norm': 0.622491180896759, 'learning_rate': 7.10960960960961e-06, 'epoch': 0.21}

  7%|▋         | 947/13317 [09:44<2:07:01,  1.62it/s]
  7%|▋         | 948/13317 [09:45<2:06:51,  1.63it/s]
                                                     
{'loss': 0.3444, 'grad_norm': 0.7141217589378357, 'learning_rate': 7.117117117117117e-06, 'epoch': 0.21}

  7%|▋         | 948/13317 [09:45<2:06:51,  1.63it/s]
  7%|▋         | 949/13317 [09:45<2:06:42,  1.63it/s]
                                                     
{'loss': 0.3367, 'grad_norm': 0.7088429927825928, 'learning_rate': 7.124624624624626e-06, 'epoch': 0.21}

  7%|▋         | 949/13317 [09:45<2:06:42,  1.63it/s]
  7%|▋         | 950/13317 [09:46<2:06:40,  1.63it/s]
                                                     
{'loss': 0.3896, 'grad_norm': 0.7505548596382141, 'learning_rate': 7.132132132132132e-06, 'epoch': 0.21}

  7%|▋         | 950/13317 [09:46<2:06:40,  1.63it/s]
  7%|▋         | 951/13317 [09:46<2:06:39,  1.63it/s]
                                                     
{'loss': 0.3121, 'grad_norm': 0.6050063967704773, 'learning_rate': 7.139639639639641e-06, 'epoch': 0.21}

  7%|▋         | 951/13317 [09:47<2:06:39,  1.63it/s]
  7%|▋         | 952/13317 [09:47<2:06:34,  1.63it/s]
                                                     
{'loss': 0.3594, 'grad_norm': 0.6949943900108337, 'learning_rate': 7.147147147147148e-06, 'epoch': 0.21}

  7%|▋         | 952/13317 [09:47<2:06:34,  1.63it/s]
  7%|▋         | 953/13317 [09:48<2:06:32,  1.63it/s]
                                                     
{'loss': 0.3659, 'grad_norm': 0.6135841012001038, 'learning_rate': 7.154654654654654e-06, 'epoch': 0.21}

  7%|▋         | 953/13317 [09:48<2:06:32,  1.63it/s]
  7%|▋         | 954/13317 [09:48<2:06:30,  1.63it/s]
                                                     
{'loss': 0.2756, 'grad_norm': 0.6268079876899719, 'learning_rate': 7.162162162162163e-06, 'epoch': 0.21}

  7%|▋         | 954/13317 [09:48<2:06:30,  1.63it/s]
  7%|▋         | 955/13317 [09:49<2:06:28,  1.63it/s]
                                                     
{'loss': 0.2721, 'grad_norm': 0.548031210899353, 'learning_rate': 7.16966966966967e-06, 'epoch': 0.22}

  7%|▋         | 955/13317 [09:49<2:06:28,  1.63it/s]
  7%|▋         | 956/13317 [09:49<2:06:33,  1.63it/s]
                                                     
{'loss': 0.3492, 'grad_norm': 0.6207965612411499, 'learning_rate': 7.177177177177178e-06, 'epoch': 0.22}

  7%|▋         | 956/13317 [09:50<2:06:33,  1.63it/s]
  7%|▋         | 957/13317 [09:50<2:06:36,  1.63it/s]
                                                     
{'loss': 0.3399, 'grad_norm': 0.6710034608840942, 'learning_rate': 7.1846846846846855e-06, 'epoch': 0.22}

  7%|▋         | 957/13317 [09:50<2:06:36,  1.63it/s]
  7%|▋         | 958/13317 [09:51<2:06:30,  1.63it/s]
                                                     
{'loss': 0.3767, 'grad_norm': 0.6045408248901367, 'learning_rate': 7.1921921921921935e-06, 'epoch': 0.22}

  7%|▋         | 958/13317 [09:51<2:06:30,  1.63it/s]
  7%|▋         | 959/13317 [09:51<2:06:34,  1.63it/s]
                                                     
{'loss': 0.3837, 'grad_norm': 0.7260593175888062, 'learning_rate': 7.199699699699701e-06, 'epoch': 0.22}

  7%|▋         | 959/13317 [09:51<2:06:34,  1.63it/s]
  7%|▋         | 960/13317 [09:52<2:06:31,  1.63it/s]
                                                     
{'loss': 0.3927, 'grad_norm': 0.7404770851135254, 'learning_rate': 7.207207207207208e-06, 'epoch': 0.22}

  7%|▋         | 960/13317 [09:52<2:06:31,  1.63it/s]
  7%|▋         | 961/13317 [09:53<2:06:35,  1.63it/s]
                                                     
{'loss': 0.3742, 'grad_norm': 0.780232846736908, 'learning_rate': 7.214714714714716e-06, 'epoch': 0.22}

  7%|▋         | 961/13317 [09:53<2:06:35,  1.63it/s]
  7%|▋         | 962/13317 [09:53<2:06:29,  1.63it/s]
                                                     
{'loss': 0.3257, 'grad_norm': 0.64482581615448, 'learning_rate': 7.222222222222223e-06, 'epoch': 0.22}

  7%|▋         | 962/13317 [09:53<2:06:29,  1.63it/s]
  7%|▋         | 963/13317 [09:54<2:06:39,  1.63it/s]
                                                     
{'loss': 0.4173, 'grad_norm': 0.7118309140205383, 'learning_rate': 7.229729729729731e-06, 'epoch': 0.22}

  7%|▋         | 963/13317 [09:54<2:06:39,  1.63it/s]
  7%|▋         | 964/13317 [09:54<2:06:36,  1.63it/s]
                                                     
{'loss': 0.2985, 'grad_norm': 0.698716402053833, 'learning_rate': 7.237237237237238e-06, 'epoch': 0.22}

  7%|▋         | 964/13317 [09:54<2:06:36,  1.63it/s]
  7%|▋         | 965/13317 [09:55<2:06:26,  1.63it/s]
                                                     
{'loss': 0.3708, 'grad_norm': 0.8307086825370789, 'learning_rate': 7.244744744744745e-06, 'epoch': 0.22}

  7%|▋         | 965/13317 [09:55<2:06:26,  1.63it/s]
  7%|▋         | 966/13317 [09:56<2:06:36,  1.63it/s]
                                                     
{'loss': 0.2933, 'grad_norm': 0.6014081835746765, 'learning_rate': 7.252252252252253e-06, 'epoch': 0.22}

  7%|▋         | 966/13317 [09:56<2:06:36,  1.63it/s]
  7%|▋         | 967/13317 [09:56<2:06:29,  1.63it/s]
                                                     
{'loss': 0.3393, 'grad_norm': 0.6517188549041748, 'learning_rate': 7.25975975975976e-06, 'epoch': 0.22}

  7%|▋         | 967/13317 [09:56<2:06:29,  1.63it/s]
  7%|▋         | 968/13317 [09:57<2:06:32,  1.63it/s]
                                                     
{'loss': 0.3399, 'grad_norm': 0.6970510482788086, 'learning_rate': 7.267267267267268e-06, 'epoch': 0.22}

  7%|▋         | 968/13317 [09:57<2:06:32,  1.63it/s]
  7%|▋         | 969/13317 [09:57<2:06:27,  1.63it/s]
                                                     
{'loss': 0.2995, 'grad_norm': 0.577782928943634, 'learning_rate': 7.2747747747747755e-06, 'epoch': 0.22}

  7%|▋         | 969/13317 [09:58<2:06:27,  1.63it/s]
  7%|▋         | 970/13317 [09:58<2:06:24,  1.63it/s]
                                                     
{'loss': 0.3237, 'grad_norm': 0.7053235173225403, 'learning_rate': 7.282282282282283e-06, 'epoch': 0.22}

  7%|▋         | 970/13317 [09:58<2:06:24,  1.63it/s]
  7%|▋         | 971/13317 [09:59<2:06:25,  1.63it/s]
                                                     
{'loss': 0.3392, 'grad_norm': 0.6102579236030579, 'learning_rate': 7.289789789789791e-06, 'epoch': 0.22}

  7%|▋         | 971/13317 [09:59<2:06:25,  1.63it/s]
  7%|▋         | 972/13317 [09:59<2:06:21,  1.63it/s]
                                                     
{'loss': 0.3254, 'grad_norm': 0.5741324424743652, 'learning_rate': 7.297297297297298e-06, 'epoch': 0.22}

  7%|▋         | 972/13317 [09:59<2:06:21,  1.63it/s]
  7%|▋         | 973/13317 [10:00<2:06:52,  1.62it/s]
                                                     
{'loss': 0.3542, 'grad_norm': 0.8370537161827087, 'learning_rate': 7.304804804804806e-06, 'epoch': 0.22}

  7%|▋         | 973/13317 [10:00<2:06:52,  1.62it/s]
  7%|▋         | 974/13317 [10:01<2:06:38,  1.62it/s]
                                                     
{'loss': 0.4003, 'grad_norm': 0.7102227210998535, 'learning_rate': 7.312312312312313e-06, 'epoch': 0.22}

  7%|▋         | 974/13317 [10:01<2:06:38,  1.62it/s]
  7%|▋         | 975/13317 [10:01<2:06:29,  1.63it/s]
                                                     
{'loss': 0.3924, 'grad_norm': 0.7195296883583069, 'learning_rate': 7.31981981981982e-06, 'epoch': 0.22}

  7%|▋         | 975/13317 [10:01<2:06:29,  1.63it/s]
  7%|▋         | 976/13317 [10:02<2:06:30,  1.63it/s]
                                                     
{'loss': 0.3173, 'grad_norm': 0.7245159149169922, 'learning_rate': 7.327327327327328e-06, 'epoch': 0.22}

  7%|▋         | 976/13317 [10:02<2:06:30,  1.63it/s]
  7%|▋         | 977/13317 [10:02<2:06:22,  1.63it/s]
                                                     
{'loss': 0.3614, 'grad_norm': 0.6495774388313293, 'learning_rate': 7.334834834834835e-06, 'epoch': 0.22}

  7%|▋         | 977/13317 [10:02<2:06:22,  1.63it/s]
  7%|▋         | 978/13317 [10:03<2:06:35,  1.62it/s]
                                                     
{'loss': 0.399, 'grad_norm': 0.7080403566360474, 'learning_rate': 7.342342342342343e-06, 'epoch': 0.22}

  7%|▋         | 978/13317 [10:03<2:06:35,  1.62it/s]
  7%|▋         | 979/13317 [10:04<2:06:30,  1.63it/s]
                                                     
{'loss': 0.4076, 'grad_norm': 0.9865195155143738, 'learning_rate': 7.34984984984985e-06, 'epoch': 0.22}

  7%|▋         | 979/13317 [10:04<2:06:30,  1.63it/s]
  7%|▋         | 980/13317 [10:04<2:06:21,  1.63it/s]
                                                     
{'loss': 0.2443, 'grad_norm': 0.552013099193573, 'learning_rate': 7.3573573573573575e-06, 'epoch': 0.22}

  7%|▋         | 980/13317 [10:04<2:06:21,  1.63it/s]
  7%|▋         | 981/13317 [10:05<2:06:30,  1.63it/s]
                                                     
{'loss': 0.3308, 'grad_norm': 0.7372044324874878, 'learning_rate': 7.3648648648648655e-06, 'epoch': 0.22}

  7%|▋         | 981/13317 [10:05<2:06:30,  1.63it/s]
  7%|▋         | 982/13317 [10:05<2:06:24,  1.63it/s]
                                                     
{'loss': 0.3863, 'grad_norm': 0.6662688255310059, 'learning_rate': 7.372372372372373e-06, 'epoch': 0.22}

  7%|▋         | 982/13317 [10:06<2:06:24,  1.63it/s]
  7%|▋         | 983/13317 [10:06<2:06:31,  1.62it/s]
                                                     
{'loss': 0.3138, 'grad_norm': 0.6874352097511292, 'learning_rate': 7.379879879879881e-06, 'epoch': 0.22}

  7%|▋         | 983/13317 [10:06<2:06:31,  1.62it/s]
  7%|▋         | 984/13317 [10:07<2:06:24,  1.63it/s]
                                                     
{'loss': 0.2921, 'grad_norm': 0.6792606711387634, 'learning_rate': 7.387387387387388e-06, 'epoch': 0.22}

  7%|▋         | 984/13317 [10:07<2:06:24,  1.63it/s]
  7%|▋         | 985/13317 [10:07<2:13:07,  1.54it/s]
                                                     
{'loss': 0.3423, 'grad_norm': 0.6236268877983093, 'learning_rate': 7.394894894894895e-06, 'epoch': 0.22}

  7%|▋         | 985/13317 [10:08<2:13:07,  1.54it/s]
  7%|▋         | 986/13317 [10:08<2:11:01,  1.57it/s]
                                                     
{'loss': 0.4895, 'grad_norm': 0.7380707263946533, 'learning_rate': 7.402402402402403e-06, 'epoch': 0.22}

  7%|▋         | 986/13317 [10:08<2:11:01,  1.57it/s]
  7%|▋         | 987/13317 [10:09<2:09:36,  1.59it/s]
                                                     
{'loss': 0.3505, 'grad_norm': 0.7391589879989624, 'learning_rate': 7.40990990990991e-06, 'epoch': 0.22}

  7%|▋         | 987/13317 [10:09<2:09:36,  1.59it/s]
  7%|▋         | 988/13317 [10:09<2:08:42,  1.60it/s]
                                                     
{'loss': 0.3998, 'grad_norm': 0.7470443248748779, 'learning_rate': 7.417417417417418e-06, 'epoch': 0.22}

  7%|▋         | 988/13317 [10:09<2:08:42,  1.60it/s]
  7%|▋         | 989/13317 [10:10<2:07:56,  1.61it/s]
                                                     
{'loss': 0.3845, 'grad_norm': 0.6810353994369507, 'learning_rate': 7.424924924924925e-06, 'epoch': 0.22}

  7%|▋         | 989/13317 [10:10<2:07:56,  1.61it/s]
  7%|▋         | 990/13317 [10:10<2:07:20,  1.61it/s]
                                                     
{'loss': 0.3889, 'grad_norm': 0.7144314646720886, 'learning_rate': 7.4324324324324324e-06, 'epoch': 0.22}

  7%|▋         | 990/13317 [10:11<2:07:20,  1.61it/s]
  7%|▋         | 991/13317 [10:11<2:07:01,  1.62it/s]
                                                     
{'loss': 0.3547, 'grad_norm': 0.7039357423782349, 'learning_rate': 7.4399399399399404e-06, 'epoch': 0.22}

  7%|▋         | 991/13317 [10:11<2:07:01,  1.62it/s]
  7%|▋         | 992/13317 [10:12<2:06:46,  1.62it/s]
                                                     
{'loss': 0.3039, 'grad_norm': 0.6466951966285706, 'learning_rate': 7.447447447447448e-06, 'epoch': 0.22}

  7%|▋         | 992/13317 [10:12<2:06:46,  1.62it/s]
  7%|▋         | 993/13317 [10:12<2:06:36,  1.62it/s]
                                                     
{'loss': 0.3194, 'grad_norm': 0.675281822681427, 'learning_rate': 7.4549549549549564e-06, 'epoch': 0.22}

  7%|▋         | 993/13317 [10:12<2:06:36,  1.62it/s]
  7%|▋         | 994/13317 [10:13<2:06:28,  1.62it/s]
                                                     
{'loss': 0.3444, 'grad_norm': 0.7394206523895264, 'learning_rate': 7.462462462462463e-06, 'epoch': 0.22}

  7%|▋         | 994/13317 [10:13<2:06:28,  1.62it/s]
  7%|▋         | 995/13317 [10:14<2:06:20,  1.63it/s]
                                                     
{'loss': 0.4913, 'grad_norm': 0.8425593376159668, 'learning_rate': 7.46996996996997e-06, 'epoch': 0.22}

  7%|▋         | 995/13317 [10:14<2:06:20,  1.63it/s]
  7%|▋         | 996/13317 [10:14<2:06:21,  1.63it/s]
                                                     
{'loss': 0.357, 'grad_norm': 0.6761422753334045, 'learning_rate': 7.477477477477479e-06, 'epoch': 0.22}

  7%|▋         | 996/13317 [10:14<2:06:21,  1.63it/s]
  7%|▋         | 997/13317 [10:15<2:06:16,  1.63it/s]
                                                     
{'loss': 0.336, 'grad_norm': 0.7053333520889282, 'learning_rate': 7.484984984984985e-06, 'epoch': 0.22}

  7%|▋         | 997/13317 [10:15<2:06:16,  1.63it/s]
  7%|▋         | 998/13317 [10:15<2:06:22,  1.62it/s]
                                                     
{'loss': 0.3156, 'grad_norm': 0.6403358578681946, 'learning_rate': 7.492492492492494e-06, 'epoch': 0.22}

  7%|▋         | 998/13317 [10:16<2:06:22,  1.62it/s]
  8%|▊         | 999/13317 [10:16<2:06:16,  1.63it/s]
                                                     
{'loss': 0.3764, 'grad_norm': 0.7442441582679749, 'learning_rate': 7.500000000000001e-06, 'epoch': 0.23}

  8%|▊         | 999/13317 [10:16<2:06:16,  1.63it/s]
  8%|▊         | 1000/13317 [10:17<2:06:13,  1.63it/s]
                                                      
{'loss': 0.3467, 'grad_norm': 0.6786202192306519, 'learning_rate': 7.507507507507507e-06, 'epoch': 0.23}

  8%|▊         | 1000/13317 [10:17<2:06:13,  1.63it/s]

  0%|          | 0/13 [00:00<?, ?it/s]

 15%|█▌        | 2/13 [00:00<00:00, 13.34it/s]

 31%|███       | 4/13 [00:00<00:01,  8.37it/s]

 38%|███▊      | 5/13 [00:00<00:01,  7.78it/s]

 46%|████▌     | 6/13 [00:00<00:00,  7.41it/s]

 54%|█████▍    | 7/13 [00:00<00:00,  7.17it/s]

 62%|██████▏   | 8/13 [00:01<00:00,  7.01it/s]

 69%|██████▉   | 9/13 [00:01<00:00,  6.91it/s]

 77%|███████▋  | 10/13 [00:01<00:00,  6.83it/s]

 85%|████████▍ | 11/13 [00:01<00:00,  6.78it/s]

 92%|█████████▏| 12/13 [00:01<00:00,  6.75it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]
                                                      

                                               
{'eval_loss': 0.8145520091056824, 'eval_runtime': 1.9546, 'eval_samples_per_second': 102.323, 'eval_steps_per_second': 6.651, 'epoch': 0.23}

  8%|▊         | 1000/13317 [10:19<2:06:13,  1.63it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]

                                               
  8%|▊         | 1001/13317 [10:19<4:06:58,  1.20s/it]
                                                      
{'loss': 0.3699, 'grad_norm': 0.6874909996986389, 'learning_rate': 7.515015015015016e-06, 'epoch': 0.23}

  8%|▊         | 1001/13317 [10:19<4:06:58,  1.20s/it]
  8%|▊         | 1002/13317 [10:20<3:30:38,  1.03s/it]
                                                      
{'loss': 0.45, 'grad_norm': 0.7177924513816833, 'learning_rate': 7.5225225225225225e-06, 'epoch': 0.23}

  8%|▊         | 1002/13317 [10:20<3:30:38,  1.03s/it]
  8%|▊         | 1003/13317 [10:20<3:05:18,  1.11it/s]
                                                      
{'loss': 0.3612, 'grad_norm': 0.589919924736023, 'learning_rate': 7.530030030030031e-06, 'epoch': 0.23}

  8%|▊         | 1003/13317 [10:21<3:05:18,  1.11it/s]
  8%|▊         | 1004/13317 [10:21<2:47:33,  1.22it/s]
                                                      
{'loss': 0.4083, 'grad_norm': 0.7460801601409912, 'learning_rate': 7.5375375375375385e-06, 'epoch': 0.23}

  8%|▊         | 1004/13317 [10:21<2:47:33,  1.22it/s]
  8%|▊         | 1005/13317 [10:22<2:35:03,  1.32it/s]
                                                      
{'loss': 0.3157, 'grad_norm': 0.6133865714073181, 'learning_rate': 7.545045045045045e-06, 'epoch': 0.23}

  8%|▊         | 1005/13317 [10:22<2:35:03,  1.32it/s]
  8%|▊         | 1006/13317 [10:22<2:26:16,  1.40it/s]
                                                      
{'loss': 0.2978, 'grad_norm': 0.5691664218902588, 'learning_rate': 7.552552552552554e-06, 'epoch': 0.23}

  8%|▊         | 1006/13317 [10:22<2:26:16,  1.40it/s]
  8%|▊         | 1007/13317 [10:23<2:27:07,  1.39it/s]
                                                      
{'loss': 0.3916, 'grad_norm': 0.6881616115570068, 'learning_rate': 7.560060060060061e-06, 'epoch': 0.23}

  8%|▊         | 1007/13317 [10:23<2:27:07,  1.39it/s]
  8%|▊         | 1008/13317 [10:24<2:20:43,  1.46it/s]
                                                      
{'loss': 0.3206, 'grad_norm': 0.6494827270507812, 'learning_rate': 7.567567567567569e-06, 'epoch': 0.23}

  8%|▊         | 1008/13317 [10:24<2:20:43,  1.46it/s]
  8%|▊         | 1009/13317 [10:24<2:16:16,  1.51it/s]
                                                      
{'loss': 0.3753, 'grad_norm': 0.7153424024581909, 'learning_rate': 7.575075075075076e-06, 'epoch': 0.23}

  8%|▊         | 1009/13317 [10:24<2:16:16,  1.51it/s]
  8%|▊         | 1010/13317 [10:25<2:13:08,  1.54it/s]
                                                      
{'loss': 0.3551, 'grad_norm': 0.6301169395446777, 'learning_rate': 7.582582582582583e-06, 'epoch': 0.23}

  8%|▊         | 1010/13317 [10:25<2:13:08,  1.54it/s]
  8%|▊         | 1011/13317 [10:25<2:11:01,  1.57it/s]
                                                      
{'loss': 0.3655, 'grad_norm': 0.7141810059547424, 'learning_rate': 7.590090090090091e-06, 'epoch': 0.23}

  8%|▊         | 1011/13317 [10:26<2:11:01,  1.57it/s]
  8%|▊         | 1012/13317 [10:26<2:09:27,  1.58it/s]
                                                      
{'loss': 0.3377, 'grad_norm': 0.7470843195915222, 'learning_rate': 7.597597597597598e-06, 'epoch': 0.23}

  8%|▊         | 1012/13317 [10:26<2:09:27,  1.58it/s]
  8%|▊         | 1013/13317 [10:27<2:08:23,  1.60it/s]
                                                      
{'loss': 0.284, 'grad_norm': 0.5900028944015503, 'learning_rate': 7.605105105105106e-06, 'epoch': 0.23}

  8%|▊         | 1013/13317 [10:27<2:08:23,  1.60it/s]
  8%|▊         | 1014/13317 [10:27<2:07:33,  1.61it/s]
                                                      
{'loss': 0.3578, 'grad_norm': 0.7027785778045654, 'learning_rate': 7.612612612612613e-06, 'epoch': 0.23}

  8%|▊         | 1014/13317 [10:27<2:07:33,  1.61it/s]
  8%|▊         | 1015/13317 [10:28<2:06:58,  1.61it/s]
                                                      
{'loss': 0.3115, 'grad_norm': 0.6147240996360779, 'learning_rate': 7.6201201201201205e-06, 'epoch': 0.23}

  8%|▊         | 1015/13317 [10:28<2:06:58,  1.61it/s]
  8%|▊         | 1016/13317 [10:28<2:06:38,  1.62it/s]
                                                      
{'loss': 0.2653, 'grad_norm': 0.5859410166740417, 'learning_rate': 7.6276276276276285e-06, 'epoch': 0.23}

  8%|▊         | 1016/13317 [10:29<2:06:38,  1.62it/s]
  8%|▊         | 1017/13317 [10:29<2:06:20,  1.62it/s]
                                                      
{'loss': 0.3515, 'grad_norm': 0.7382671236991882, 'learning_rate': 7.635135135135135e-06, 'epoch': 0.23}

  8%|▊         | 1017/13317 [10:29<2:06:20,  1.62it/s]
  8%|▊         | 1018/13317 [10:30<2:06:10,  1.62it/s]
                                                      
{'loss': 0.3733, 'grad_norm': 0.6097984910011292, 'learning_rate': 7.642642642642644e-06, 'epoch': 0.23}

  8%|▊         | 1018/13317 [10:30<2:06:10,  1.62it/s]
  8%|▊         | 1019/13317 [10:30<2:06:04,  1.63it/s]
                                                      
{'loss': 0.3377, 'grad_norm': 0.6497223377227783, 'learning_rate': 7.65015015015015e-06, 'epoch': 0.23}

  8%|▊         | 1019/13317 [10:30<2:06:04,  1.63it/s]
  8%|▊         | 1020/13317 [10:31<2:05:55,  1.63it/s]
                                                      
{'loss': 0.3122, 'grad_norm': 0.6848929524421692, 'learning_rate': 7.657657657657658e-06, 'epoch': 0.23}

  8%|▊         | 1020/13317 [10:31<2:05:55,  1.63it/s]
  8%|▊         | 1021/13317 [10:32<2:05:48,  1.63it/s]
                                                      
{'loss': 0.3418, 'grad_norm': 0.7345035672187805, 'learning_rate': 7.665165165165165e-06, 'epoch': 0.23}

  8%|▊         | 1021/13317 [10:32<2:05:48,  1.63it/s]
  8%|▊         | 1022/13317 [10:32<2:05:57,  1.63it/s]
                                                      
{'loss': 0.4111, 'grad_norm': 0.7326303124427795, 'learning_rate': 7.672672672672672e-06, 'epoch': 0.23}

  8%|▊         | 1022/13317 [10:32<2:05:57,  1.63it/s]
  8%|▊         | 1023/13317 [10:33<2:05:47,  1.63it/s]
                                                      
{'loss': 0.3055, 'grad_norm': 0.6431784629821777, 'learning_rate': 7.680180180180181e-06, 'epoch': 0.23}

  8%|▊         | 1023/13317 [10:33<2:05:47,  1.63it/s]
  8%|▊         | 1024/13317 [10:33<2:05:45,  1.63it/s]
                                                      
{'loss': 0.3561, 'grad_norm': 0.6504462957382202, 'learning_rate': 7.687687687687688e-06, 'epoch': 0.23}

  8%|▊         | 1024/13317 [10:34<2:05:45,  1.63it/s]
  8%|▊         | 1025/13317 [10:34<2:05:41,  1.63it/s]
                                                      
{'loss': 0.2907, 'grad_norm': 0.7298970818519592, 'learning_rate': 7.695195195195195e-06, 'epoch': 0.23}

  8%|▊         | 1025/13317 [10:34<2:05:41,  1.63it/s]
  8%|▊         | 1026/13317 [10:35<2:05:39,  1.63it/s]
                                                      
{'loss': 0.3508, 'grad_norm': 0.7310369610786438, 'learning_rate': 7.702702702702704e-06, 'epoch': 0.23}

  8%|▊         | 1026/13317 [10:35<2:05:39,  1.63it/s]
  8%|▊         | 1027/13317 [10:35<2:05:36,  1.63it/s]
                                                      
{'loss': 0.3239, 'grad_norm': 0.6799018383026123, 'learning_rate': 7.71021021021021e-06, 'epoch': 0.23}

  8%|▊         | 1027/13317 [10:35<2:05:36,  1.63it/s]
  8%|▊         | 1028/13317 [10:36<2:05:37,  1.63it/s]
                                                      
{'loss': 0.4024, 'grad_norm': 0.8414976000785828, 'learning_rate': 7.717717717717719e-06, 'epoch': 0.23}

  8%|▊         | 1028/13317 [10:36<2:05:37,  1.63it/s]
  8%|▊         | 1029/13317 [10:36<2:05:34,  1.63it/s]
                                                      
{'loss': 0.279, 'grad_norm': 0.556726336479187, 'learning_rate': 7.725225225225226e-06, 'epoch': 0.23}

  8%|▊         | 1029/13317 [10:37<2:05:34,  1.63it/s]
  8%|▊         | 1030/13317 [10:37<2:05:34,  1.63it/s]
                                                      
{'loss': 0.2526, 'grad_norm': 0.6089081764221191, 'learning_rate': 7.732732732732733e-06, 'epoch': 0.23}

  8%|▊         | 1030/13317 [10:37<2:05:34,  1.63it/s]
  8%|▊         | 1031/13317 [10:38<2:05:34,  1.63it/s]
                                                      
{'loss': 0.2716, 'grad_norm': 0.6249637603759766, 'learning_rate': 7.740240240240242e-06, 'epoch': 0.23}

  8%|▊         | 1031/13317 [10:38<2:05:34,  1.63it/s]
  8%|▊         | 1032/13317 [10:38<2:05:34,  1.63it/s]
                                                      
{'loss': 0.3089, 'grad_norm': 0.6763566732406616, 'learning_rate': 7.747747747747749e-06, 'epoch': 0.23}

  8%|▊         | 1032/13317 [10:38<2:05:34,  1.63it/s]
  8%|▊         | 1033/13317 [10:39<2:05:32,  1.63it/s]
                                                      
{'loss': 0.3948, 'grad_norm': 0.7282575964927673, 'learning_rate': 7.755255255255256e-06, 'epoch': 0.23}

  8%|▊         | 1033/13317 [10:39<2:05:32,  1.63it/s]
  8%|▊         | 1034/13317 [10:40<2:05:35,  1.63it/s]
                                                      
{'loss': 0.3154, 'grad_norm': 0.7060515284538269, 'learning_rate': 7.762762762762763e-06, 'epoch': 0.23}

  8%|▊         | 1034/13317 [10:40<2:05:35,  1.63it/s]
  8%|▊         | 1035/13317 [10:40<2:05:36,  1.63it/s]
                                                      
{'loss': 0.3212, 'grad_norm': 0.6651778817176819, 'learning_rate': 7.77027027027027e-06, 'epoch': 0.23}

  8%|▊         | 1035/13317 [10:40<2:05:36,  1.63it/s]
  8%|▊         | 1036/13317 [10:41<2:05:32,  1.63it/s]
                                                      
{'loss': 0.3407, 'grad_norm': 0.6919489502906799, 'learning_rate': 7.77777777777778e-06, 'epoch': 0.23}

  8%|▊         | 1036/13317 [10:41<2:05:32,  1.63it/s]
  8%|▊         | 1037/13317 [10:41<2:05:31,  1.63it/s]
                                                      
{'loss': 0.3441, 'grad_norm': 0.7015458941459656, 'learning_rate': 7.785285285285286e-06, 'epoch': 0.23}

  8%|▊         | 1037/13317 [10:42<2:05:31,  1.63it/s]
  8%|▊         | 1038/13317 [10:42<2:05:32,  1.63it/s]
                                                      
{'loss': 0.346, 'grad_norm': 0.6370284557342529, 'learning_rate': 7.792792792792793e-06, 'epoch': 0.23}

  8%|▊         | 1038/13317 [10:42<2:05:32,  1.63it/s]
  8%|▊         | 1039/13317 [10:43<2:05:35,  1.63it/s]
                                                      
{'loss': 0.3458, 'grad_norm': 0.7174455523490906, 'learning_rate': 7.8003003003003e-06, 'epoch': 0.23}

  8%|▊         | 1039/13317 [10:43<2:05:35,  1.63it/s]
  8%|▊         | 1040/13317 [10:43<2:05:38,  1.63it/s]
                                                      
{'loss': 0.3879, 'grad_norm': 0.6505283713340759, 'learning_rate': 7.807807807807808e-06, 'epoch': 0.23}

  8%|▊         | 1040/13317 [10:43<2:05:38,  1.63it/s]
  8%|▊         | 1041/13317 [10:44<2:05:41,  1.63it/s]
                                                      
{'loss': 0.4148, 'grad_norm': 0.6890393495559692, 'learning_rate': 7.815315315315317e-06, 'epoch': 0.23}

  8%|▊         | 1041/13317 [10:44<2:05:41,  1.63it/s]
  8%|▊         | 1042/13317 [10:44<2:05:34,  1.63it/s]
                                                      
{'loss': 0.356, 'grad_norm': 0.7155529856681824, 'learning_rate': 7.822822822822824e-06, 'epoch': 0.23}

  8%|▊         | 1042/13317 [10:45<2:05:34,  1.63it/s]
  8%|▊         | 1043/13317 [10:45<2:05:37,  1.63it/s]
                                                      
{'loss': 0.4181, 'grad_norm': 0.6800033450126648, 'learning_rate': 7.830330330330331e-06, 'epoch': 0.23}

  8%|▊         | 1043/13317 [10:45<2:05:37,  1.63it/s]
  8%|▊         | 1044/13317 [10:46<2:05:34,  1.63it/s]
                                                      
{'loss': 0.3806, 'grad_norm': 0.7283975481987, 'learning_rate': 7.837837837837838e-06, 'epoch': 0.24}

  8%|▊         | 1044/13317 [10:46<2:05:34,  1.63it/s]
  8%|▊         | 1045/13317 [10:46<2:05:34,  1.63it/s]
                                                      
{'loss': 0.3416, 'grad_norm': 0.7864100337028503, 'learning_rate': 7.845345345345347e-06, 'epoch': 0.24}

  8%|▊         | 1045/13317 [10:46<2:05:34,  1.63it/s]
  8%|▊         | 1046/13317 [10:47<2:05:32,  1.63it/s]
                                                      
{'loss': 0.3962, 'grad_norm': 0.8433253765106201, 'learning_rate': 7.852852852852854e-06, 'epoch': 0.24}

  8%|▊         | 1046/13317 [10:47<2:05:32,  1.63it/s]
  8%|▊         | 1047/13317 [10:48<2:05:28,  1.63it/s]
                                                      
{'loss': 0.3927, 'grad_norm': 0.7770147323608398, 'learning_rate': 7.860360360360361e-06, 'epoch': 0.24}

  8%|▊         | 1047/13317 [10:48<2:05:28,  1.63it/s]
  8%|▊         | 1048/13317 [10:48<2:05:26,  1.63it/s]
                                                      
{'loss': 0.2942, 'grad_norm': 0.7273305058479309, 'learning_rate': 7.867867867867868e-06, 'epoch': 0.24}

  8%|▊         | 1048/13317 [10:48<2:05:26,  1.63it/s]
  8%|▊         | 1049/13317 [10:49<2:05:37,  1.63it/s]
                                                      
{'loss': 0.3141, 'grad_norm': 0.5855206847190857, 'learning_rate': 7.875375375375376e-06, 'epoch': 0.24}

  8%|▊         | 1049/13317 [10:49<2:05:37,  1.63it/s]
  8%|▊         | 1050/13317 [10:49<2:05:38,  1.63it/s]
                                                      
{'loss': 0.3496, 'grad_norm': 0.7767764925956726, 'learning_rate': 7.882882882882884e-06, 'epoch': 0.24}

  8%|▊         | 1050/13317 [10:49<2:05:38,  1.63it/s]
  8%|▊         | 1051/13317 [10:50<2:05:34,  1.63it/s]
                                                      
{'loss': 0.3769, 'grad_norm': 0.7466822862625122, 'learning_rate': 7.890390390390392e-06, 'epoch': 0.24}

  8%|▊         | 1051/13317 [10:50<2:05:34,  1.63it/s]
  8%|▊         | 1052/13317 [10:51<2:05:30,  1.63it/s]
                                                      
{'loss': 0.3192, 'grad_norm': 0.5365979075431824, 'learning_rate': 7.897897897897899e-06, 'epoch': 0.24}

  8%|▊         | 1052/13317 [10:51<2:05:30,  1.63it/s]
  8%|▊         | 1053/13317 [10:51<2:05:31,  1.63it/s]
                                                      
{'loss': 0.2787, 'grad_norm': 0.6007304787635803, 'learning_rate': 7.905405405405406e-06, 'epoch': 0.24}

  8%|▊         | 1053/13317 [10:51<2:05:31,  1.63it/s]
  8%|▊         | 1054/13317 [10:52<2:05:35,  1.63it/s]
                                                      
{'loss': 0.364, 'grad_norm': 0.8009090423583984, 'learning_rate': 7.912912912912913e-06, 'epoch': 0.24}

  8%|▊         | 1054/13317 [10:52<2:05:35,  1.63it/s]
  8%|▊         | 1055/13317 [10:52<2:05:33,  1.63it/s]
                                                      
{'loss': 0.3454, 'grad_norm': 0.7096115350723267, 'learning_rate': 7.920420420420422e-06, 'epoch': 0.24}

  8%|▊         | 1055/13317 [10:53<2:05:33,  1.63it/s]
  8%|▊         | 1056/13317 [10:53<2:05:30,  1.63it/s]
                                                      
{'loss': 0.3185, 'grad_norm': 0.7269859910011292, 'learning_rate': 7.927927927927929e-06, 'epoch': 0.24}

  8%|▊         | 1056/13317 [10:53<2:05:30,  1.63it/s]
  8%|▊         | 1057/13317 [10:54<2:05:30,  1.63it/s]
                                                      
{'loss': 0.3295, 'grad_norm': 0.695466935634613, 'learning_rate': 7.935435435435436e-06, 'epoch': 0.24}

  8%|▊         | 1057/13317 [10:54<2:05:30,  1.63it/s]
  8%|▊         | 1058/13317 [10:54<2:05:25,  1.63it/s]
                                                      
{'loss': 0.2903, 'grad_norm': 0.6020987033843994, 'learning_rate': 7.942942942942943e-06, 'epoch': 0.24}

  8%|▊         | 1058/13317 [10:54<2:05:25,  1.63it/s]
  8%|▊         | 1059/13317 [10:55<2:05:37,  1.63it/s]
                                                      
{'loss': 0.3774, 'grad_norm': 0.808205783367157, 'learning_rate': 7.95045045045045e-06, 'epoch': 0.24}

  8%|▊         | 1059/13317 [10:55<2:05:37,  1.63it/s]
  8%|▊         | 1060/13317 [10:56<2:05:27,  1.63it/s]
                                                      
{'loss': 0.284, 'grad_norm': 0.7726279497146606, 'learning_rate': 7.95795795795796e-06, 'epoch': 0.24}

  8%|▊         | 1060/13317 [10:56<2:05:27,  1.63it/s]
  8%|▊         | 1061/13317 [10:56<2:05:27,  1.63it/s]
                                                      
{'loss': 0.3083, 'grad_norm': 0.6241421103477478, 'learning_rate': 7.965465465465466e-06, 'epoch': 0.24}

  8%|▊         | 1061/13317 [10:56<2:05:27,  1.63it/s]
  8%|▊         | 1062/13317 [10:57<2:05:22,  1.63it/s]
                                                      
{'loss': 0.319, 'grad_norm': 0.7230749726295471, 'learning_rate': 7.972972972972974e-06, 'epoch': 0.24}

  8%|▊         | 1062/13317 [10:57<2:05:22,  1.63it/s]
  8%|▊         | 1063/13317 [10:57<2:05:42,  1.62it/s]
                                                      
{'loss': 0.3654, 'grad_norm': 0.7685800194740295, 'learning_rate': 7.98048048048048e-06, 'epoch': 0.24}

  8%|▊         | 1063/13317 [10:57<2:05:42,  1.62it/s]
  8%|▊         | 1064/13317 [10:58<2:05:38,  1.63it/s]
                                                      
{'loss': 0.3009, 'grad_norm': 0.6416943669319153, 'learning_rate': 7.987987987987988e-06, 'epoch': 0.24}

  8%|▊         | 1064/13317 [10:58<2:05:38,  1.63it/s]
  8%|▊         | 1065/13317 [10:59<2:05:33,  1.63it/s]
                                                      
{'loss': 0.3485, 'grad_norm': 0.6728408932685852, 'learning_rate': 7.995495495495497e-06, 'epoch': 0.24}

  8%|▊         | 1065/13317 [10:59<2:05:33,  1.63it/s]
  8%|▊         | 1066/13317 [10:59<2:05:26,  1.63it/s]
                                                      
{'loss': 0.335, 'grad_norm': 0.6744813919067383, 'learning_rate': 8.003003003003004e-06, 'epoch': 0.24}

  8%|▊         | 1066/13317 [10:59<2:05:26,  1.63it/s]
  8%|▊         | 1067/13317 [11:00<2:05:19,  1.63it/s]
                                                      
{'loss': 0.2923, 'grad_norm': 0.5424734354019165, 'learning_rate': 8.010510510510511e-06, 'epoch': 0.24}

  8%|▊         | 1067/13317 [11:00<2:05:19,  1.63it/s]
  8%|▊         | 1068/13317 [11:00<2:05:20,  1.63it/s]
                                                      
{'loss': 0.4337, 'grad_norm': 0.7797048091888428, 'learning_rate': 8.018018018018018e-06, 'epoch': 0.24}

  8%|▊         | 1068/13317 [11:01<2:05:20,  1.63it/s]
  8%|▊         | 1069/13317 [11:01<2:05:31,  1.63it/s]
                                                      
{'loss': 0.3151, 'grad_norm': 0.6391661763191223, 'learning_rate': 8.025525525525525e-06, 'epoch': 0.24}

  8%|▊         | 1069/13317 [11:01<2:05:31,  1.63it/s]
  8%|▊         | 1070/13317 [11:02<2:05:21,  1.63it/s]
                                                      
{'loss': 0.3721, 'grad_norm': 0.8033766746520996, 'learning_rate': 8.033033033033034e-06, 'epoch': 0.24}

  8%|▊         | 1070/13317 [11:02<2:05:21,  1.63it/s]
  8%|▊         | 1071/13317 [11:02<2:05:20,  1.63it/s]
                                                      
{'loss': 0.4356, 'grad_norm': 0.7559898495674133, 'learning_rate': 8.040540540540541e-06, 'epoch': 0.24}

  8%|▊         | 1071/13317 [11:02<2:05:20,  1.63it/s]
  8%|▊         | 1072/13317 [11:03<2:05:31,  1.63it/s]
                                                      
{'loss': 0.3402, 'grad_norm': 0.6742945313453674, 'learning_rate': 8.048048048048048e-06, 'epoch': 0.24}

  8%|▊         | 1072/13317 [11:03<2:05:31,  1.63it/s]
  8%|▊         | 1073/13317 [11:03<2:05:13,  1.63it/s]
                                                      
{'loss': 0.4132, 'grad_norm': 0.7832605242729187, 'learning_rate': 8.055555555555557e-06, 'epoch': 0.24}

  8%|▊         | 1073/13317 [11:04<2:05:13,  1.63it/s]
  8%|▊         | 1074/13317 [11:04<2:05:18,  1.63it/s]
                                                      
{'loss': 0.2736, 'grad_norm': 0.6298468708992004, 'learning_rate': 8.063063063063063e-06, 'epoch': 0.24}

  8%|▊         | 1074/13317 [11:04<2:05:18,  1.63it/s]
  8%|▊         | 1075/13317 [11:05<2:05:12,  1.63it/s]
                                                      
{'loss': 0.3345, 'grad_norm': 0.7428615689277649, 'learning_rate': 8.070570570570572e-06, 'epoch': 0.24}

  8%|▊         | 1075/13317 [11:05<2:05:12,  1.63it/s]
  8%|▊         | 1076/13317 [11:05<2:05:11,  1.63it/s]
                                                      
{'loss': 0.3456, 'grad_norm': 0.7073543667793274, 'learning_rate': 8.078078078078079e-06, 'epoch': 0.24}

  8%|▊         | 1076/13317 [11:05<2:05:11,  1.63it/s]
  8%|▊         | 1077/13317 [11:06<2:05:07,  1.63it/s]
                                                      
{'loss': 0.3088, 'grad_norm': 0.6384907364845276, 'learning_rate': 8.085585585585586e-06, 'epoch': 0.24}

  8%|▊         | 1077/13317 [11:06<2:05:07,  1.63it/s]
  8%|▊         | 1078/13317 [11:07<2:05:05,  1.63it/s]
                                                      
{'loss': 0.3697, 'grad_norm': 0.8172017931938171, 'learning_rate': 8.093093093093095e-06, 'epoch': 0.24}

  8%|▊         | 1078/13317 [11:07<2:05:05,  1.63it/s]
  8%|▊         | 1079/13317 [11:07<2:05:13,  1.63it/s]
                                                      
{'loss': 0.3487, 'grad_norm': 0.6245198249816895, 'learning_rate': 8.100600600600602e-06, 'epoch': 0.24}

  8%|▊         | 1079/13317 [11:07<2:05:13,  1.63it/s]
  8%|▊         | 1080/13317 [11:08<2:05:09,  1.63it/s]
                                                      
{'loss': 0.3249, 'grad_norm': 0.6469165086746216, 'learning_rate': 8.108108108108109e-06, 'epoch': 0.24}

  8%|▊         | 1080/13317 [11:08<2:05:09,  1.63it/s]
  8%|▊         | 1081/13317 [11:08<2:05:09,  1.63it/s]
                                                      
{'loss': 0.2593, 'grad_norm': 0.5712411999702454, 'learning_rate': 8.115615615615616e-06, 'epoch': 0.24}

  8%|▊         | 1081/13317 [11:09<2:05:09,  1.63it/s]
  8%|▊         | 1082/13317 [11:09<2:05:10,  1.63it/s]
                                                      
{'loss': 0.3533, 'grad_norm': 0.7163971066474915, 'learning_rate': 8.123123123123123e-06, 'epoch': 0.24}

  8%|▊         | 1082/13317 [11:09<2:05:10,  1.63it/s]
  8%|▊         | 1083/13317 [11:10<2:05:10,  1.63it/s]
                                                      
{'loss': 0.366, 'grad_norm': 0.6952530741691589, 'learning_rate': 8.130630630630632e-06, 'epoch': 0.24}

  8%|▊         | 1083/13317 [11:10<2:05:10,  1.63it/s]
  8%|▊         | 1084/13317 [11:10<2:05:19,  1.63it/s]
                                                      
{'loss': 0.3394, 'grad_norm': 0.7308059334754944, 'learning_rate': 8.13813813813814e-06, 'epoch': 0.24}

  8%|▊         | 1084/13317 [11:10<2:05:19,  1.63it/s]
  8%|▊         | 1085/13317 [11:11<2:05:20,  1.63it/s]
                                                      
{'loss': 0.4107, 'grad_norm': 0.7589032053947449, 'learning_rate': 8.145645645645646e-06, 'epoch': 0.24}

  8%|▊         | 1085/13317 [11:11<2:05:20,  1.63it/s]
  8%|▊         | 1086/13317 [11:11<2:05:15,  1.63it/s]
                                                      
{'loss': 0.3373, 'grad_norm': 0.7607566714286804, 'learning_rate': 8.153153153153154e-06, 'epoch': 0.24}

  8%|▊         | 1086/13317 [11:12<2:05:15,  1.63it/s]
  8%|▊         | 1087/13317 [11:12<2:05:09,  1.63it/s]
                                                      
{'loss': 0.2656, 'grad_norm': 0.6671813726425171, 'learning_rate': 8.16066066066066e-06, 'epoch': 0.24}

  8%|▊         | 1087/13317 [11:12<2:05:09,  1.63it/s]
  8%|▊         | 1088/13317 [11:13<2:05:04,  1.63it/s]
                                                      
{'loss': 0.3772, 'grad_norm': 0.7553513646125793, 'learning_rate': 8.16816816816817e-06, 'epoch': 0.25}

  8%|▊         | 1088/13317 [11:13<2:05:04,  1.63it/s]
  8%|▊         | 1089/13317 [11:13<2:05:18,  1.63it/s]
                                                      
{'loss': 0.3528, 'grad_norm': 0.6369515061378479, 'learning_rate': 8.175675675675677e-06, 'epoch': 0.25}

  8%|▊         | 1089/13317 [11:13<2:05:18,  1.63it/s]
  8%|▊         | 1090/13317 [11:14<2:05:10,  1.63it/s]
                                                      
{'loss': 0.3519, 'grad_norm': 0.783295750617981, 'learning_rate': 8.183183183183184e-06, 'epoch': 0.25}

  8%|▊         | 1090/13317 [11:14<2:05:10,  1.63it/s]
  8%|▊         | 1091/13317 [11:15<2:05:06,  1.63it/s]
                                                      
{'loss': 0.3698, 'grad_norm': 0.7967883944511414, 'learning_rate': 8.190690690690691e-06, 'epoch': 0.25}

  8%|▊         | 1091/13317 [11:15<2:05:06,  1.63it/s]
  8%|▊         | 1092/13317 [11:15<2:05:03,  1.63it/s]
                                                      
{'loss': 0.339, 'grad_norm': 0.6881406307220459, 'learning_rate': 8.198198198198198e-06, 'epoch': 0.25}

  8%|▊         | 1092/13317 [11:15<2:05:03,  1.63it/s]
  8%|▊         | 1093/13317 [11:16<2:04:59,  1.63it/s]
                                                      
{'loss': 0.2494, 'grad_norm': 0.5423270463943481, 'learning_rate': 8.205705705705707e-06, 'epoch': 0.25}

  8%|▊         | 1093/13317 [11:16<2:04:59,  1.63it/s]
  8%|▊         | 1094/13317 [11:16<2:05:06,  1.63it/s]
                                                      
{'loss': 0.3518, 'grad_norm': 0.8276898264884949, 'learning_rate': 8.213213213213214e-06, 'epoch': 0.25}

  8%|▊         | 1094/13317 [11:17<2:05:06,  1.63it/s]
  8%|▊         | 1095/13317 [11:17<2:04:59,  1.63it/s]
                                                      
{'loss': 0.3808, 'grad_norm': 0.7536709308624268, 'learning_rate': 8.220720720720721e-06, 'epoch': 0.25}

  8%|▊         | 1095/13317 [11:17<2:04:59,  1.63it/s]
  8%|▊         | 1096/13317 [11:18<2:05:02,  1.63it/s]
                                                      
{'loss': 0.2721, 'grad_norm': 0.6645088791847229, 'learning_rate': 8.228228228228229e-06, 'epoch': 0.25}

  8%|▊         | 1096/13317 [11:18<2:05:02,  1.63it/s]
  8%|▊         | 1097/13317 [11:18<2:05:05,  1.63it/s]
                                                      
{'loss': 0.3073, 'grad_norm': 0.5745928883552551, 'learning_rate': 8.235735735735736e-06, 'epoch': 0.25}

  8%|▊         | 1097/13317 [11:18<2:05:05,  1.63it/s]
  8%|▊         | 1098/13317 [11:19<2:04:58,  1.63it/s]
                                                      
{'loss': 0.3462, 'grad_norm': 0.8224321603775024, 'learning_rate': 8.243243243243245e-06, 'epoch': 0.25}

  8%|▊         | 1098/13317 [11:19<2:04:58,  1.63it/s]
  8%|▊         | 1099/13317 [11:19<2:05:06,  1.63it/s]
                                                      
{'loss': 0.3603, 'grad_norm': 0.7924473285675049, 'learning_rate': 8.250750750750752e-06, 'epoch': 0.25}

  8%|▊         | 1099/13317 [11:20<2:05:06,  1.63it/s]
  8%|▊         | 1100/13317 [11:20<2:05:24,  1.62it/s]
                                                      
{'loss': 0.365, 'grad_norm': 0.8002750873565674, 'learning_rate': 8.258258258258259e-06, 'epoch': 0.25}

  8%|▊         | 1100/13317 [11:20<2:05:24,  1.62it/s]
  8%|▊         | 1101/13317 [11:21<2:05:14,  1.63it/s]
                                                      
{'loss': 0.3397, 'grad_norm': 0.7008338570594788, 'learning_rate': 8.265765765765766e-06, 'epoch': 0.25}

  8%|▊         | 1101/13317 [11:21<2:05:14,  1.63it/s]
  8%|▊         | 1102/13317 [11:21<2:05:09,  1.63it/s]
                                                      
{'loss': 0.3197, 'grad_norm': 0.5932532548904419, 'learning_rate': 8.273273273273273e-06, 'epoch': 0.25}

  8%|▊         | 1102/13317 [11:21<2:05:09,  1.63it/s]
  8%|▊         | 1103/13317 [11:22<2:05:08,  1.63it/s]
                                                      
{'loss': 0.2953, 'grad_norm': 0.6848196983337402, 'learning_rate': 8.280780780780782e-06, 'epoch': 0.25}

  8%|▊         | 1103/13317 [11:22<2:05:08,  1.63it/s]
  8%|▊         | 1104/13317 [11:23<2:05:14,  1.63it/s]
                                                      
{'loss': 0.3944, 'grad_norm': 0.7408862709999084, 'learning_rate': 8.288288288288289e-06, 'epoch': 0.25}

  8%|▊         | 1104/13317 [11:23<2:05:14,  1.63it/s]
  8%|▊         | 1105/13317 [11:23<2:05:05,  1.63it/s]
                                                      
{'loss': 0.3526, 'grad_norm': 0.735761284828186, 'learning_rate': 8.295795795795796e-06, 'epoch': 0.25}

  8%|▊         | 1105/13317 [11:23<2:05:05,  1.63it/s]
  8%|▊         | 1106/13317 [11:24<2:05:00,  1.63it/s]
                                                      
{'loss': 0.3704, 'grad_norm': 0.7394276857376099, 'learning_rate': 8.303303303303303e-06, 'epoch': 0.25}

  8%|▊         | 1106/13317 [11:24<2:05:00,  1.63it/s]
  8%|▊         | 1107/13317 [11:24<2:04:59,  1.63it/s]
                                                      
{'loss': 0.2988, 'grad_norm': 0.7181752920150757, 'learning_rate': 8.31081081081081e-06, 'epoch': 0.25}

  8%|▊         | 1107/13317 [11:25<2:04:59,  1.63it/s]
  8%|▊         | 1108/13317 [11:25<2:04:56,  1.63it/s]
                                                      
{'loss': 0.3365, 'grad_norm': 0.6857311725616455, 'learning_rate': 8.31831831831832e-06, 'epoch': 0.25}

  8%|▊         | 1108/13317 [11:25<2:04:56,  1.63it/s]
  8%|▊         | 1109/13317 [11:26<2:05:08,  1.63it/s]
                                                      
{'loss': 0.3256, 'grad_norm': 0.6235151290893555, 'learning_rate': 8.325825825825827e-06, 'epoch': 0.25}

  8%|▊         | 1109/13317 [11:26<2:05:08,  1.63it/s]
  8%|▊         | 1110/13317 [11:26<2:05:02,  1.63it/s]
                                                      
{'loss': 0.3618, 'grad_norm': 0.6997745633125305, 'learning_rate': 8.333333333333334e-06, 'epoch': 0.25}

  8%|▊         | 1110/13317 [11:26<2:05:02,  1.63it/s]
  8%|▊         | 1111/13317 [11:27<2:04:56,  1.63it/s]
                                                      
{'loss': 0.4179, 'grad_norm': 0.6884016990661621, 'learning_rate': 8.340840840840841e-06, 'epoch': 0.25}

  8%|▊         | 1111/13317 [11:27<2:04:56,  1.63it/s]
  8%|▊         | 1112/13317 [11:27<2:04:53,  1.63it/s]
                                                      
{'loss': 0.3033, 'grad_norm': 0.6338616609573364, 'learning_rate': 8.348348348348348e-06, 'epoch': 0.25}

  8%|▊         | 1112/13317 [11:28<2:04:53,  1.63it/s]
  8%|▊         | 1113/13317 [11:28<2:04:51,  1.63it/s]
                                                      
{'loss': 0.4339, 'grad_norm': 0.7468676567077637, 'learning_rate': 8.355855855855857e-06, 'epoch': 0.25}

  8%|▊         | 1113/13317 [11:28<2:04:51,  1.63it/s]
  8%|▊         | 1114/13317 [11:29<2:05:00,  1.63it/s]
                                                      
{'loss': 0.3153, 'grad_norm': 0.5926698446273804, 'learning_rate': 8.363363363363364e-06, 'epoch': 0.25}

  8%|▊         | 1114/13317 [11:29<2:05:00,  1.63it/s]
  8%|▊         | 1115/13317 [11:29<2:04:57,  1.63it/s]
                                                      
{'loss': 0.3483, 'grad_norm': 0.7296412587165833, 'learning_rate': 8.370870870870871e-06, 'epoch': 0.25}

  8%|▊         | 1115/13317 [11:29<2:04:57,  1.63it/s]
  8%|▊         | 1116/13317 [11:30<2:05:00,  1.63it/s]
                                                      
{'loss': 0.2966, 'grad_norm': 0.6379625201225281, 'learning_rate': 8.378378378378378e-06, 'epoch': 0.25}

  8%|▊         | 1116/13317 [11:30<2:05:00,  1.63it/s]
  8%|▊         | 1117/13317 [11:31<2:04:53,  1.63it/s]
                                                      
{'loss': 0.3945, 'grad_norm': 0.649151623249054, 'learning_rate': 8.385885885885885e-06, 'epoch': 0.25}

  8%|▊         | 1117/13317 [11:31<2:04:53,  1.63it/s]
  8%|▊         | 1118/13317 [11:31<2:04:50,  1.63it/s]
                                                      
{'loss': 0.2754, 'grad_norm': 0.5853186249732971, 'learning_rate': 8.393393393393394e-06, 'epoch': 0.25}

  8%|▊         | 1118/13317 [11:31<2:04:50,  1.63it/s]
  8%|▊         | 1119/13317 [11:32<2:05:01,  1.63it/s]
                                                      
{'loss': 0.3377, 'grad_norm': 0.6234282851219177, 'learning_rate': 8.400900900900901e-06, 'epoch': 0.25}

  8%|▊         | 1119/13317 [11:32<2:05:01,  1.63it/s]
  8%|▊         | 1120/13317 [11:32<2:04:57,  1.63it/s]
                                                      
{'loss': 0.3548, 'grad_norm': 0.641842782497406, 'learning_rate': 8.408408408408409e-06, 'epoch': 0.25}

  8%|▊         | 1120/13317 [11:32<2:04:57,  1.63it/s]
  8%|▊         | 1121/13317 [11:33<2:05:06,  1.62it/s]
                                                      
{'loss': 0.2618, 'grad_norm': 0.5626994371414185, 'learning_rate': 8.415915915915916e-06, 'epoch': 0.25}

  8%|▊         | 1121/13317 [11:33<2:05:06,  1.62it/s]
  8%|▊         | 1122/13317 [11:34<2:05:02,  1.63it/s]
                                                      
{'loss': 0.3426, 'grad_norm': 0.7685413360595703, 'learning_rate': 8.423423423423423e-06, 'epoch': 0.25}

  8%|▊         | 1122/13317 [11:34<2:05:02,  1.63it/s]
  8%|▊         | 1123/13317 [11:34<2:04:57,  1.63it/s]
                                                      
{'loss': 0.3382, 'grad_norm': 0.6236857175827026, 'learning_rate': 8.430930930930932e-06, 'epoch': 0.25}

  8%|▊         | 1123/13317 [11:34<2:04:57,  1.63it/s]
  8%|▊         | 1124/13317 [11:35<2:05:06,  1.62it/s]
                                                      
{'loss': 0.3201, 'grad_norm': 0.7045652866363525, 'learning_rate': 8.438438438438439e-06, 'epoch': 0.25}

  8%|▊         | 1124/13317 [11:35<2:05:06,  1.62it/s]
  8%|▊         | 1125/13317 [11:35<2:04:59,  1.63it/s]
                                                      
{'loss': 0.3461, 'grad_norm': 0.6700553297996521, 'learning_rate': 8.445945945945948e-06, 'epoch': 0.25}

  8%|▊         | 1125/13317 [11:36<2:04:59,  1.63it/s]
  8%|▊         | 1126/13317 [11:36<2:05:05,  1.62it/s]
                                                      
{'loss': 0.3574, 'grad_norm': 0.6689032912254333, 'learning_rate': 8.453453453453453e-06, 'epoch': 0.25}

  8%|▊         | 1126/13317 [11:36<2:05:05,  1.62it/s]
  8%|▊         | 1127/13317 [11:37<2:04:59,  1.63it/s]
                                                      
{'loss': 0.2567, 'grad_norm': 0.5884860157966614, 'learning_rate': 8.460960960960962e-06, 'epoch': 0.25}

  8%|▊         | 1127/13317 [11:37<2:04:59,  1.63it/s]
  8%|▊         | 1128/13317 [11:37<2:04:53,  1.63it/s]
                                                      
{'loss': 0.3485, 'grad_norm': 0.745843768119812, 'learning_rate': 8.46846846846847e-06, 'epoch': 0.25}

  8%|▊         | 1128/13317 [11:37<2:04:53,  1.63it/s]
  8%|▊         | 1129/13317 [11:38<2:05:31,  1.62it/s]
                                                      
{'loss': 0.2705, 'grad_norm': 0.5884525179862976, 'learning_rate': 8.475975975975976e-06, 'epoch': 0.25}

  8%|▊         | 1129/13317 [11:38<2:05:31,  1.62it/s]
  8%|▊         | 1130/13317 [11:39<2:05:17,  1.62it/s]
                                                      
{'loss': 0.2886, 'grad_norm': 0.6655063033103943, 'learning_rate': 8.483483483483485e-06, 'epoch': 0.25}

  8%|▊         | 1130/13317 [11:39<2:05:17,  1.62it/s]
  8%|▊         | 1131/13317 [11:39<2:05:19,  1.62it/s]
                                                      
{'loss': 0.2246, 'grad_norm': 0.5798375606536865, 'learning_rate': 8.490990990990992e-06, 'epoch': 0.25}

  8%|▊         | 1131/13317 [11:39<2:05:19,  1.62it/s]
  9%|▊         | 1132/13317 [11:40<2:05:09,  1.62it/s]
                                                      
{'loss': 0.2898, 'grad_norm': 0.5954354405403137, 'learning_rate': 8.4984984984985e-06, 'epoch': 0.26}

  9%|▊         | 1132/13317 [11:40<2:05:09,  1.62it/s]
  9%|▊         | 1133/13317 [11:40<2:05:01,  1.62it/s]
                                                      
{'loss': 0.3183, 'grad_norm': 0.6561781764030457, 'learning_rate': 8.506006006006007e-06, 'epoch': 0.26}

  9%|▊         | 1133/13317 [11:41<2:05:01,  1.62it/s]
  9%|▊         | 1134/13317 [11:41<2:05:04,  1.62it/s]
                                                      
{'loss': 0.3906, 'grad_norm': 0.7586193680763245, 'learning_rate': 8.513513513513514e-06, 'epoch': 0.26}

  9%|▊         | 1134/13317 [11:41<2:05:04,  1.62it/s]
  9%|▊         | 1135/13317 [11:42<2:04:51,  1.63it/s]
                                                      
{'loss': 0.4433, 'grad_norm': 0.9533849358558655, 'learning_rate': 8.521021021021023e-06, 'epoch': 0.26}

  9%|▊         | 1135/13317 [11:42<2:04:51,  1.63it/s]
  9%|▊         | 1136/13317 [11:42<2:04:53,  1.63it/s]
                                                      
{'loss': 0.3745, 'grad_norm': 0.6923743486404419, 'learning_rate': 8.52852852852853e-06, 'epoch': 0.26}

  9%|▊         | 1136/13317 [11:42<2:04:53,  1.63it/s]
  9%|▊         | 1137/13317 [11:43<2:04:48,  1.63it/s]
                                                      
{'loss': 0.3604, 'grad_norm': 0.8022654056549072, 'learning_rate': 8.536036036036037e-06, 'epoch': 0.26}

  9%|▊         | 1137/13317 [11:43<2:04:48,  1.63it/s]
  9%|▊         | 1138/13317 [11:43<2:04:40,  1.63it/s]
                                                      
{'loss': 0.3074, 'grad_norm': 0.7071915864944458, 'learning_rate': 8.543543543543544e-06, 'epoch': 0.26}

  9%|▊         | 1138/13317 [11:44<2:04:40,  1.63it/s]
  9%|▊         | 1139/13317 [11:44<2:04:43,  1.63it/s]
                                                      
{'loss': 0.3016, 'grad_norm': 0.5812227129936218, 'learning_rate': 8.551051051051051e-06, 'epoch': 0.26}

  9%|▊         | 1139/13317 [11:44<2:04:43,  1.63it/s]
  9%|▊         | 1140/13317 [11:45<2:04:37,  1.63it/s]
                                                      
{'loss': 0.3012, 'grad_norm': 0.6889265179634094, 'learning_rate': 8.55855855855856e-06, 'epoch': 0.26}

  9%|▊         | 1140/13317 [11:45<2:04:37,  1.63it/s]
  9%|▊         | 1141/13317 [11:45<2:04:42,  1.63it/s]
                                                      
{'loss': 0.3539, 'grad_norm': 0.7447793483734131, 'learning_rate': 8.566066066066067e-06, 'epoch': 0.26}

  9%|▊         | 1141/13317 [11:45<2:04:42,  1.63it/s]
  9%|▊         | 1142/13317 [11:46<2:04:33,  1.63it/s]
                                                      
{'loss': 0.2517, 'grad_norm': 0.5638440251350403, 'learning_rate': 8.573573573573574e-06, 'epoch': 0.26}

  9%|▊         | 1142/13317 [11:46<2:04:33,  1.63it/s]
  9%|▊         | 1143/13317 [11:47<2:04:31,  1.63it/s]
                                                      
{'loss': 0.2968, 'grad_norm': 0.6387895345687866, 'learning_rate': 8.581081081081082e-06, 'epoch': 0.26}

  9%|▊         | 1143/13317 [11:47<2:04:31,  1.63it/s]
  9%|▊         | 1144/13317 [11:47<2:04:44,  1.63it/s]
                                                      
{'loss': 0.3459, 'grad_norm': 0.6395075917243958, 'learning_rate': 8.588588588588589e-06, 'epoch': 0.26}

  9%|▊         | 1144/13317 [11:47<2:04:44,  1.63it/s]
  9%|▊         | 1145/13317 [11:48<2:04:35,  1.63it/s]
                                                      
{'loss': 0.2593, 'grad_norm': 0.5857076644897461, 'learning_rate': 8.596096096096098e-06, 'epoch': 0.26}

  9%|▊         | 1145/13317 [11:48<2:04:35,  1.63it/s]
  9%|▊         | 1146/13317 [11:48<2:04:29,  1.63it/s]
                                                      
{'loss': 0.356, 'grad_norm': 0.7590922713279724, 'learning_rate': 8.603603603603605e-06, 'epoch': 0.26}

  9%|▊         | 1146/13317 [11:48<2:04:29,  1.63it/s]
  9%|▊         | 1147/13317 [11:49<2:04:29,  1.63it/s]
                                                      
{'loss': 0.408, 'grad_norm': 0.7659421563148499, 'learning_rate': 8.611111111111112e-06, 'epoch': 0.26}

  9%|▊         | 1147/13317 [11:49<2:04:29,  1.63it/s]
  9%|▊         | 1148/13317 [11:50<2:04:30,  1.63it/s]
                                                      
{'loss': 0.38, 'grad_norm': 0.7032918930053711, 'learning_rate': 8.618618618618619e-06, 'epoch': 0.26}

  9%|▊         | 1148/13317 [11:50<2:04:30,  1.63it/s]
  9%|▊         | 1149/13317 [11:50<2:04:32,  1.63it/s]
                                                      
{'loss': 0.3493, 'grad_norm': 0.6920309066772461, 'learning_rate': 8.626126126126126e-06, 'epoch': 0.26}

  9%|▊         | 1149/13317 [11:50<2:04:32,  1.63it/s]
  9%|▊         | 1150/13317 [11:51<2:04:32,  1.63it/s]
                                                      
{'loss': 0.3837, 'grad_norm': 0.6600304841995239, 'learning_rate': 8.633633633633635e-06, 'epoch': 0.26}

  9%|▊         | 1150/13317 [11:51<2:04:32,  1.63it/s]
  9%|▊         | 1151/13317 [11:51<2:04:31,  1.63it/s]
                                                      
{'loss': 0.2786, 'grad_norm': 0.6172508001327515, 'learning_rate': 8.641141141141142e-06, 'epoch': 0.26}

  9%|▊         | 1151/13317 [11:52<2:04:31,  1.63it/s]
  9%|▊         | 1152/13317 [11:52<2:04:25,  1.63it/s]
                                                      
{'loss': 0.2865, 'grad_norm': 0.5717065334320068, 'learning_rate': 8.64864864864865e-06, 'epoch': 0.26}

  9%|▊         | 1152/13317 [11:52<2:04:25,  1.63it/s]
  9%|▊         | 1153/13317 [11:53<2:04:22,  1.63it/s]
                                                      
{'loss': 0.3671, 'grad_norm': 0.6850570440292358, 'learning_rate': 8.656156156156156e-06, 'epoch': 0.26}

  9%|▊         | 1153/13317 [11:53<2:04:22,  1.63it/s]
  9%|▊         | 1154/13317 [11:53<2:04:23,  1.63it/s]
                                                      
{'loss': 0.2663, 'grad_norm': 0.5535637140274048, 'learning_rate': 8.663663663663664e-06, 'epoch': 0.26}

  9%|▊         | 1154/13317 [11:53<2:04:23,  1.63it/s]
  9%|▊         | 1155/13317 [11:54<2:04:23,  1.63it/s]
                                                      
{'loss': 0.3499, 'grad_norm': 0.7190561890602112, 'learning_rate': 8.671171171171172e-06, 'epoch': 0.26}

  9%|▊         | 1155/13317 [11:54<2:04:23,  1.63it/s]
  9%|▊         | 1156/13317 [11:54<2:04:20,  1.63it/s]
                                                      
{'loss': 0.3547, 'grad_norm': 0.735019326210022, 'learning_rate': 8.67867867867868e-06, 'epoch': 0.26}

  9%|▊         | 1156/13317 [11:55<2:04:20,  1.63it/s]
  9%|▊         | 1157/13317 [11:55<2:04:20,  1.63it/s]
                                                      
{'loss': 0.3571, 'grad_norm': 0.7271222472190857, 'learning_rate': 8.686186186186187e-06, 'epoch': 0.26}

  9%|▊         | 1157/13317 [11:55<2:04:20,  1.63it/s]
  9%|▊         | 1158/13317 [11:56<2:04:22,  1.63it/s]
                                                      
{'loss': 0.3915, 'grad_norm': 0.7214857935905457, 'learning_rate': 8.693693693693694e-06, 'epoch': 0.26}

  9%|▊         | 1158/13317 [11:56<2:04:22,  1.63it/s]
  9%|▊         | 1159/13317 [11:56<2:04:35,  1.63it/s]
                                                      
{'loss': 0.3605, 'grad_norm': 0.752098798751831, 'learning_rate': 8.701201201201201e-06, 'epoch': 0.26}

  9%|▊         | 1159/13317 [11:56<2:04:35,  1.63it/s]
  9%|▊         | 1160/13317 [11:57<2:04:32,  1.63it/s]
                                                      
{'loss': 0.2575, 'grad_norm': 0.5803020596504211, 'learning_rate': 8.70870870870871e-06, 'epoch': 0.26}

  9%|▊         | 1160/13317 [11:57<2:04:32,  1.63it/s]
  9%|▊         | 1161/13317 [11:58<2:04:30,  1.63it/s]
                                                      
{'loss': 0.3687, 'grad_norm': 0.6590619683265686, 'learning_rate': 8.716216216216217e-06, 'epoch': 0.26}

  9%|▊         | 1161/13317 [11:58<2:04:30,  1.63it/s]
  9%|▊         | 1162/13317 [11:58<2:04:29,  1.63it/s]
                                                      
{'loss': 0.2942, 'grad_norm': 0.7704602479934692, 'learning_rate': 8.723723723723724e-06, 'epoch': 0.26}

  9%|▊         | 1162/13317 [11:58<2:04:29,  1.63it/s]
  9%|▊         | 1163/13317 [11:59<2:04:23,  1.63it/s]
                                                      
{'loss': 0.2819, 'grad_norm': 0.7158790826797485, 'learning_rate': 8.731231231231231e-06, 'epoch': 0.26}

  9%|▊         | 1163/13317 [11:59<2:04:23,  1.63it/s]
  9%|▊         | 1164/13317 [11:59<2:04:17,  1.63it/s]
                                                      
{'loss': 0.3491, 'grad_norm': 0.6899182200431824, 'learning_rate': 8.738738738738739e-06, 'epoch': 0.26}

  9%|▊         | 1164/13317 [12:00<2:04:17,  1.63it/s]
  9%|▊         | 1165/13317 [12:00<2:04:17,  1.63it/s]
                                                      
{'loss': 0.3138, 'grad_norm': 0.6224047541618347, 'learning_rate': 8.746246246246247e-06, 'epoch': 0.26}

  9%|▊         | 1165/13317 [12:00<2:04:17,  1.63it/s]
  9%|▉         | 1166/13317 [12:01<2:04:15,  1.63it/s]
                                                      
{'loss': 0.3201, 'grad_norm': 0.678869903087616, 'learning_rate': 8.753753753753755e-06, 'epoch': 0.26}

  9%|▉         | 1166/13317 [12:01<2:04:15,  1.63it/s]
  9%|▉         | 1167/13317 [12:01<2:04:15,  1.63it/s]
                                                      
{'loss': 0.37, 'grad_norm': 0.803135335445404, 'learning_rate': 8.761261261261262e-06, 'epoch': 0.26}

  9%|▉         | 1167/13317 [12:01<2:04:15,  1.63it/s]
  9%|▉         | 1168/13317 [12:02<2:04:16,  1.63it/s]
                                                      
{'loss': 0.4131, 'grad_norm': 0.8125234246253967, 'learning_rate': 8.768768768768769e-06, 'epoch': 0.26}

  9%|▉         | 1168/13317 [12:02<2:04:16,  1.63it/s]
  9%|▉         | 1169/13317 [12:02<2:04:18,  1.63it/s]
                                                      
{'loss': 0.3522, 'grad_norm': 0.7340235710144043, 'learning_rate': 8.776276276276276e-06, 'epoch': 0.26}

  9%|▉         | 1169/13317 [12:03<2:04:18,  1.63it/s]
  9%|▉         | 1170/13317 [12:03<2:04:14,  1.63it/s]
                                                      
{'loss': 0.3151, 'grad_norm': 0.5663572549819946, 'learning_rate': 8.783783783783785e-06, 'epoch': 0.26}

  9%|▉         | 1170/13317 [12:03<2:04:14,  1.63it/s]
  9%|▉         | 1171/13317 [12:04<2:04:16,  1.63it/s]
                                                      
{'loss': 0.3731, 'grad_norm': 0.7122007012367249, 'learning_rate': 8.791291291291292e-06, 'epoch': 0.26}

  9%|▉         | 1171/13317 [12:04<2:04:16,  1.63it/s]
  9%|▉         | 1172/13317 [12:04<2:04:20,  1.63it/s]
                                                      
{'loss': 0.3925, 'grad_norm': 0.7478320002555847, 'learning_rate': 8.798798798798799e-06, 'epoch': 0.26}

  9%|▉         | 1172/13317 [12:04<2:04:20,  1.63it/s]
  9%|▉         | 1173/13317 [12:05<2:04:16,  1.63it/s]
                                                      
{'loss': 0.3001, 'grad_norm': 0.5502858757972717, 'learning_rate': 8.806306306306306e-06, 'epoch': 0.26}

  9%|▉         | 1173/13317 [12:05<2:04:16,  1.63it/s]
  9%|▉         | 1174/13317 [12:06<2:04:13,  1.63it/s]
                                                      
{'loss': 0.2596, 'grad_norm': 0.585471510887146, 'learning_rate': 8.813813813813813e-06, 'epoch': 0.26}

  9%|▉         | 1174/13317 [12:06<2:04:13,  1.63it/s]
  9%|▉         | 1175/13317 [12:06<2:04:11,  1.63it/s]
                                                      
{'loss': 0.3249, 'grad_norm': 0.7501271367073059, 'learning_rate': 8.821321321321322e-06, 'epoch': 0.26}

  9%|▉         | 1175/13317 [12:06<2:04:11,  1.63it/s]
  9%|▉         | 1176/13317 [12:07<2:04:09,  1.63it/s]
                                                      
{'loss': 0.2428, 'grad_norm': 0.6109883189201355, 'learning_rate': 8.82882882882883e-06, 'epoch': 0.26}

  9%|▉         | 1176/13317 [12:07<2:04:09,  1.63it/s]
  9%|▉         | 1177/13317 [12:07<2:04:08,  1.63it/s]
                                                      
{'loss': 0.2947, 'grad_norm': 0.5550290942192078, 'learning_rate': 8.836336336336338e-06, 'epoch': 0.27}

  9%|▉         | 1177/13317 [12:08<2:04:08,  1.63it/s]
  9%|▉         | 1178/13317 [12:08<2:04:08,  1.63it/s]
                                                      
{'loss': 0.3414, 'grad_norm': 0.6400628685951233, 'learning_rate': 8.843843843843844e-06, 'epoch': 0.27}

  9%|▉         | 1178/13317 [12:08<2:04:08,  1.63it/s]
  9%|▉         | 1179/13317 [12:09<2:04:04,  1.63it/s]
                                                      
{'loss': 0.3216, 'grad_norm': 0.6320927143096924, 'learning_rate': 8.851351351351351e-06, 'epoch': 0.27}

  9%|▉         | 1179/13317 [12:09<2:04:04,  1.63it/s]
  9%|▉         | 1180/13317 [12:09<2:04:09,  1.63it/s]
                                                      
{'loss': 0.3656, 'grad_norm': 0.6920549869537354, 'learning_rate': 8.85885885885886e-06, 'epoch': 0.27}

  9%|▉         | 1180/13317 [12:09<2:04:09,  1.63it/s]
  9%|▉         | 1181/13317 [12:10<2:04:07,  1.63it/s]
                                                      
{'loss': 0.26, 'grad_norm': 0.595150351524353, 'learning_rate': 8.866366366366367e-06, 'epoch': 0.27}

  9%|▉         | 1181/13317 [12:10<2:04:07,  1.63it/s]
  9%|▉         | 1182/13317 [12:10<2:04:04,  1.63it/s]
                                                      
{'loss': 0.3137, 'grad_norm': 0.6257100701332092, 'learning_rate': 8.873873873873876e-06, 'epoch': 0.27}

  9%|▉         | 1182/13317 [12:11<2:04:04,  1.63it/s]
  9%|▉         | 1183/13317 [12:11<2:04:08,  1.63it/s]
                                                      
{'loss': 0.2596, 'grad_norm': 0.6334328055381775, 'learning_rate': 8.881381381381383e-06, 'epoch': 0.27}

  9%|▉         | 1183/13317 [12:11<2:04:08,  1.63it/s]
  9%|▉         | 1184/13317 [12:12<2:04:04,  1.63it/s]
                                                      
{'loss': 0.3411, 'grad_norm': 0.7255639433860779, 'learning_rate': 8.888888888888888e-06, 'epoch': 0.27}

  9%|▉         | 1184/13317 [12:12<2:04:04,  1.63it/s]
  9%|▉         | 1185/13317 [12:12<2:04:02,  1.63it/s]
                                                      
{'loss': 0.3082, 'grad_norm': 0.6193342208862305, 'learning_rate': 8.896396396396397e-06, 'epoch': 0.27}

  9%|▉         | 1185/13317 [12:12<2:04:02,  1.63it/s]
  9%|▉         | 1186/13317 [12:13<2:04:04,  1.63it/s]
                                                      
{'loss': 0.2932, 'grad_norm': 0.6404427886009216, 'learning_rate': 8.903903903903904e-06, 'epoch': 0.27}

  9%|▉         | 1186/13317 [12:13<2:04:04,  1.63it/s]
  9%|▉         | 1187/13317 [12:14<2:04:03,  1.63it/s]
                                                      
{'loss': 0.3502, 'grad_norm': 0.7138048410415649, 'learning_rate': 8.911411411411413e-06, 'epoch': 0.27}

  9%|▉         | 1187/13317 [12:14<2:04:03,  1.63it/s]
  9%|▉         | 1188/13317 [12:14<2:04:04,  1.63it/s]
                                                      
{'loss': 0.3015, 'grad_norm': 0.6489798426628113, 'learning_rate': 8.91891891891892e-06, 'epoch': 0.27}

  9%|▉         | 1188/13317 [12:14<2:04:04,  1.63it/s]
  9%|▉         | 1189/13317 [12:15<2:04:01,  1.63it/s]
                                                      
{'loss': 0.2909, 'grad_norm': 0.5893575549125671, 'learning_rate': 8.926426426426427e-06, 'epoch': 0.27}

  9%|▉         | 1189/13317 [12:15<2:04:01,  1.63it/s]
  9%|▉         | 1190/13317 [12:15<2:04:01,  1.63it/s]
                                                      
{'loss': 0.3327, 'grad_norm': 0.6025270223617554, 'learning_rate': 8.933933933933935e-06, 'epoch': 0.27}

  9%|▉         | 1190/13317 [12:15<2:04:01,  1.63it/s]
  9%|▉         | 1191/13317 [12:16<2:03:57,  1.63it/s]
                                                      
{'loss': 0.3192, 'grad_norm': 0.6831895709037781, 'learning_rate': 8.941441441441442e-06, 'epoch': 0.27}

  9%|▉         | 1191/13317 [12:16<2:03:57,  1.63it/s]
  9%|▉         | 1192/13317 [12:17<2:03:55,  1.63it/s]
                                                      
{'loss': 0.3074, 'grad_norm': 0.6907995343208313, 'learning_rate': 8.94894894894895e-06, 'epoch': 0.27}

  9%|▉         | 1192/13317 [12:17<2:03:55,  1.63it/s]
  9%|▉         | 1193/13317 [12:17<2:03:51,  1.63it/s]
                                                      
{'loss': 0.3172, 'grad_norm': 0.701304018497467, 'learning_rate': 8.956456456456458e-06, 'epoch': 0.27}

  9%|▉         | 1193/13317 [12:17<2:03:51,  1.63it/s]
  9%|▉         | 1194/13317 [12:18<2:03:51,  1.63it/s]
                                                      
{'loss': 0.3353, 'grad_norm': 0.6715788841247559, 'learning_rate': 8.963963963963965e-06, 'epoch': 0.27}

  9%|▉         | 1194/13317 [12:18<2:03:51,  1.63it/s]
  9%|▉         | 1195/13317 [12:18<2:03:48,  1.63it/s]
                                                      
{'loss': 0.3487, 'grad_norm': 0.7472814321517944, 'learning_rate': 8.971471471471472e-06, 'epoch': 0.27}

  9%|▉         | 1195/13317 [12:19<2:03:48,  1.63it/s]
  9%|▉         | 1196/13317 [12:19<2:03:54,  1.63it/s]
                                                      
{'loss': 0.2975, 'grad_norm': 0.6651027202606201, 'learning_rate': 8.97897897897898e-06, 'epoch': 0.27}

  9%|▉         | 1196/13317 [12:19<2:03:54,  1.63it/s]
  9%|▉         | 1197/13317 [12:20<2:03:53,  1.63it/s]
                                                      
{'loss': 0.2993, 'grad_norm': 0.6543668508529663, 'learning_rate': 8.986486486486488e-06, 'epoch': 0.27}

  9%|▉         | 1197/13317 [12:20<2:03:53,  1.63it/s]
  9%|▉         | 1198/13317 [12:20<2:03:53,  1.63it/s]
                                                      
{'loss': 0.3771, 'grad_norm': 0.752967119216919, 'learning_rate': 8.993993993993995e-06, 'epoch': 0.27}

  9%|▉         | 1198/13317 [12:20<2:03:53,  1.63it/s]
  9%|▉         | 1199/13317 [12:21<2:03:53,  1.63it/s]
                                                      
{'loss': 0.4219, 'grad_norm': 0.8792669177055359, 'learning_rate': 9.001501501501502e-06, 'epoch': 0.27}

  9%|▉         | 1199/13317 [12:21<2:03:53,  1.63it/s]
  9%|▉         | 1200/13317 [12:22<2:04:20,  1.62it/s]
                                                      
{'loss': 0.3524, 'grad_norm': 0.6256784796714783, 'learning_rate': 9.00900900900901e-06, 'epoch': 0.27}

  9%|▉         | 1200/13317 [12:22<2:04:20,  1.62it/s]
  9%|▉         | 1201/13317 [12:22<2:04:10,  1.63it/s]
                                                      
{'loss': 0.3568, 'grad_norm': 0.7856670618057251, 'learning_rate': 9.016516516516517e-06, 'epoch': 0.27}

  9%|▉         | 1201/13317 [12:22<2:04:10,  1.63it/s]
  9%|▉         | 1202/13317 [12:23<2:04:03,  1.63it/s]
                                                      
{'loss': 0.2677, 'grad_norm': 0.5992921590805054, 'learning_rate': 9.024024024024025e-06, 'epoch': 0.27}

  9%|▉         | 1202/13317 [12:23<2:04:03,  1.63it/s]
  9%|▉         | 1203/13317 [12:23<2:03:57,  1.63it/s]
                                                      
{'loss': 0.3256, 'grad_norm': 0.6378344297409058, 'learning_rate': 9.031531531531533e-06, 'epoch': 0.27}

  9%|▉         | 1203/13317 [12:23<2:03:57,  1.63it/s]
  9%|▉         | 1204/13317 [12:24<2:03:54,  1.63it/s]
                                                      
{'loss': 0.3256, 'grad_norm': 0.6404006481170654, 'learning_rate': 9.03903903903904e-06, 'epoch': 0.27}

  9%|▉         | 1204/13317 [12:24<2:03:54,  1.63it/s]
  9%|▉         | 1205/13317 [12:25<2:03:55,  1.63it/s]
                                                      
{'loss': 0.2982, 'grad_norm': 0.5925794839859009, 'learning_rate': 9.046546546546547e-06, 'epoch': 0.27}

  9%|▉         | 1205/13317 [12:25<2:03:55,  1.63it/s]
  9%|▉         | 1206/13317 [12:25<2:04:00,  1.63it/s]
                                                      
{'loss': 0.2713, 'grad_norm': 0.5825552940368652, 'learning_rate': 9.054054054054054e-06, 'epoch': 0.27}

  9%|▉         | 1206/13317 [12:25<2:04:00,  1.63it/s]
  9%|▉         | 1207/13317 [12:26<2:03:53,  1.63it/s]
                                                      
{'loss': 0.3472, 'grad_norm': 0.6966301798820496, 'learning_rate': 9.061561561561563e-06, 'epoch': 0.27}

  9%|▉         | 1207/13317 [12:26<2:03:53,  1.63it/s]
  9%|▉         | 1208/13317 [12:26<2:03:49,  1.63it/s]
                                                      
{'loss': 0.2965, 'grad_norm': 0.602179765701294, 'learning_rate': 9.06906906906907e-06, 'epoch': 0.27}

  9%|▉         | 1208/13317 [12:27<2:03:49,  1.63it/s]
  9%|▉         | 1209/13317 [12:27<2:03:53,  1.63it/s]
                                                      
{'loss': 0.3089, 'grad_norm': 0.6330350041389465, 'learning_rate': 9.076576576576577e-06, 'epoch': 0.27}

  9%|▉         | 1209/13317 [12:27<2:03:53,  1.63it/s]
  9%|▉         | 1210/13317 [12:28<2:03:56,  1.63it/s]
                                                      
{'loss': 0.2325, 'grad_norm': 0.5592118501663208, 'learning_rate': 9.084084084084084e-06, 'epoch': 0.27}

  9%|▉         | 1210/13317 [12:28<2:03:56,  1.63it/s]
  9%|▉         | 1211/13317 [12:28<2:04:11,  1.62it/s]
                                                      
{'loss': 0.2737, 'grad_norm': 0.6828854084014893, 'learning_rate': 9.091591591591592e-06, 'epoch': 0.27}

  9%|▉         | 1211/13317 [12:28<2:04:11,  1.62it/s]
  9%|▉         | 1212/13317 [12:29<2:04:06,  1.63it/s]
                                                      
{'loss': 0.3488, 'grad_norm': 0.6612675786018372, 'learning_rate': 9.0990990990991e-06, 'epoch': 0.27}

  9%|▉         | 1212/13317 [12:29<2:04:06,  1.63it/s]
  9%|▉         | 1213/13317 [12:29<2:04:03,  1.63it/s]
                                                      
{'loss': 0.2859, 'grad_norm': 0.7305225133895874, 'learning_rate': 9.106606606606608e-06, 'epoch': 0.27}

  9%|▉         | 1213/13317 [12:30<2:04:03,  1.63it/s]
  9%|▉         | 1214/13317 [12:30<2:04:06,  1.63it/s]
                                                      
{'loss': 0.2782, 'grad_norm': 0.7781265377998352, 'learning_rate': 9.114114114114115e-06, 'epoch': 0.27}

  9%|▉         | 1214/13317 [12:30<2:04:06,  1.63it/s]
  9%|▉         | 1215/13317 [12:31<2:03:56,  1.63it/s]
                                                      
{'loss': 0.4117, 'grad_norm': 0.725769579410553, 'learning_rate': 9.121621621621622e-06, 'epoch': 0.27}

  9%|▉         | 1215/13317 [12:31<2:03:56,  1.63it/s]
  9%|▉         | 1216/13317 [12:31<2:04:05,  1.63it/s]
                                                      
{'loss': 0.3568, 'grad_norm': 0.7210356593132019, 'learning_rate': 9.129129129129129e-06, 'epoch': 0.27}

  9%|▉         | 1216/13317 [12:31<2:04:05,  1.63it/s]
  9%|▉         | 1217/13317 [12:32<2:03:58,  1.63it/s]
                                                      
{'loss': 0.3774, 'grad_norm': 0.7344104647636414, 'learning_rate': 9.136636636636638e-06, 'epoch': 0.27}

  9%|▉         | 1217/13317 [12:32<2:03:58,  1.63it/s]
  9%|▉         | 1218/13317 [12:33<2:03:53,  1.63it/s]
                                                      
{'loss': 0.2506, 'grad_norm': 0.6286000609397888, 'learning_rate': 9.144144144144145e-06, 'epoch': 0.27}

  9%|▉         | 1218/13317 [12:33<2:03:53,  1.63it/s]
  9%|▉         | 1219/13317 [12:33<2:03:59,  1.63it/s]
                                                      
{'loss': 0.2516, 'grad_norm': 0.6589239835739136, 'learning_rate': 9.151651651651652e-06, 'epoch': 0.27}

  9%|▉         | 1219/13317 [12:33<2:03:59,  1.63it/s]
  9%|▉         | 1220/13317 [12:34<2:03:51,  1.63it/s]
                                                      
{'loss': 0.3759, 'grad_norm': 0.771934986114502, 'learning_rate': 9.15915915915916e-06, 'epoch': 0.27}

  9%|▉         | 1220/13317 [12:34<2:03:51,  1.63it/s]
  9%|▉         | 1221/13317 [12:34<2:03:48,  1.63it/s]
                                                      
{'loss': 0.4436, 'grad_norm': 0.6901326179504395, 'learning_rate': 9.166666666666666e-06, 'epoch': 0.28}

  9%|▉         | 1221/13317 [12:35<2:03:48,  1.63it/s]
  9%|▉         | 1222/13317 [12:35<2:03:44,  1.63it/s]
                                                      
{'loss': 0.3386, 'grad_norm': 0.741788923740387, 'learning_rate': 9.174174174174175e-06, 'epoch': 0.28}

  9%|▉         | 1222/13317 [12:35<2:03:44,  1.63it/s]
  9%|▉         | 1223/13317 [12:36<2:03:42,  1.63it/s]
                                                      
{'loss': 0.3712, 'grad_norm': 0.6703755855560303, 'learning_rate': 9.181681681681682e-06, 'epoch': 0.28}

  9%|▉         | 1223/13317 [12:36<2:03:42,  1.63it/s]
  9%|▉         | 1224/13317 [12:36<2:03:52,  1.63it/s]
                                                      
{'loss': 0.3498, 'grad_norm': 0.65214604139328, 'learning_rate': 9.189189189189191e-06, 'epoch': 0.28}

  9%|▉         | 1224/13317 [12:36<2:03:52,  1.63it/s]
  9%|▉         | 1225/13317 [12:37<2:03:44,  1.63it/s]
                                                      
{'loss': 0.2715, 'grad_norm': 0.6254073977470398, 'learning_rate': 9.196696696696697e-06, 'epoch': 0.28}

  9%|▉         | 1225/13317 [12:37<2:03:44,  1.63it/s]
  9%|▉         | 1226/13317 [12:37<2:03:48,  1.63it/s]
                                                      
{'loss': 0.3676, 'grad_norm': 0.6594048738479614, 'learning_rate': 9.204204204204204e-06, 'epoch': 0.28}

  9%|▉         | 1226/13317 [12:38<2:03:48,  1.63it/s]
  9%|▉         | 1227/13317 [12:38<2:03:42,  1.63it/s]
                                                      
{'loss': 0.3704, 'grad_norm': 0.8022565841674805, 'learning_rate': 9.211711711711713e-06, 'epoch': 0.28}

  9%|▉         | 1227/13317 [12:38<2:03:42,  1.63it/s]
  9%|▉         | 1228/13317 [12:39<2:03:37,  1.63it/s]
                                                      
{'loss': 0.3474, 'grad_norm': 0.6630275845527649, 'learning_rate': 9.21921921921922e-06, 'epoch': 0.28}

  9%|▉         | 1228/13317 [12:39<2:03:37,  1.63it/s]
  9%|▉         | 1229/13317 [12:39<2:03:46,  1.63it/s]
                                                      
{'loss': 0.3085, 'grad_norm': 0.7502239346504211, 'learning_rate': 9.226726726726729e-06, 'epoch': 0.28}

  9%|▉         | 1229/13317 [12:39<2:03:46,  1.63it/s]
  9%|▉         | 1230/13317 [12:40<2:03:48,  1.63it/s]
                                                      
{'loss': 0.3219, 'grad_norm': 0.8021007776260376, 'learning_rate': 9.234234234234236e-06, 'epoch': 0.28}

  9%|▉         | 1230/13317 [12:40<2:03:48,  1.63it/s]
  9%|▉         | 1231/13317 [12:41<2:03:37,  1.63it/s]
                                                      
{'loss': 0.3352, 'grad_norm': 0.6818904280662537, 'learning_rate': 9.241741741741741e-06, 'epoch': 0.28}

  9%|▉         | 1231/13317 [12:41<2:03:37,  1.63it/s]
  9%|▉         | 1232/13317 [12:41<2:03:34,  1.63it/s]
                                                      
{'loss': 0.2975, 'grad_norm': 0.6796093583106995, 'learning_rate': 9.24924924924925e-06, 'epoch': 0.28}

  9%|▉         | 1232/13317 [12:41<2:03:34,  1.63it/s]
  9%|▉         | 1233/13317 [12:42<2:03:36,  1.63it/s]
                                                      
{'loss': 0.2977, 'grad_norm': 0.5982251167297363, 'learning_rate': 9.256756756756757e-06, 'epoch': 0.28}

  9%|▉         | 1233/13317 [12:42<2:03:36,  1.63it/s]
  9%|▉         | 1234/13317 [12:42<2:03:41,  1.63it/s]
                                                      
{'loss': 0.307, 'grad_norm': 0.5849611163139343, 'learning_rate': 9.264264264264266e-06, 'epoch': 0.28}

  9%|▉         | 1234/13317 [12:43<2:03:41,  1.63it/s]
  9%|▉         | 1235/13317 [12:43<2:03:37,  1.63it/s]
                                                      
{'loss': 0.2713, 'grad_norm': 0.6457366943359375, 'learning_rate': 9.271771771771773e-06, 'epoch': 0.28}

  9%|▉         | 1235/13317 [12:43<2:03:37,  1.63it/s]
  9%|▉         | 1236/13317 [12:44<2:03:39,  1.63it/s]
                                                      
{'loss': 0.2926, 'grad_norm': 0.631959855556488, 'learning_rate': 9.27927927927928e-06, 'epoch': 0.28}

  9%|▉         | 1236/13317 [12:44<2:03:39,  1.63it/s]
  9%|▉         | 1237/13317 [12:44<2:03:36,  1.63it/s]
                                                      
{'loss': 0.305, 'grad_norm': 0.6948050260543823, 'learning_rate': 9.286786786786788e-06, 'epoch': 0.28}

  9%|▉         | 1237/13317 [12:44<2:03:36,  1.63it/s]
  9%|▉         | 1238/13317 [12:45<2:03:33,  1.63it/s]
                                                      
{'loss': 0.4334, 'grad_norm': 0.7455312013626099, 'learning_rate': 9.294294294294295e-06, 'epoch': 0.28}

  9%|▉         | 1238/13317 [12:45<2:03:33,  1.63it/s]
  9%|▉         | 1239/13317 [12:45<2:03:42,  1.63it/s]
                                                      
{'loss': 0.2508, 'grad_norm': 0.6758842468261719, 'learning_rate': 9.301801801801804e-06, 'epoch': 0.28}

  9%|▉         | 1239/13317 [12:46<2:03:42,  1.63it/s]
  9%|▉         | 1240/13317 [12:46<2:03:35,  1.63it/s]
                                                      
{'loss': 0.3945, 'grad_norm': 0.8015835881233215, 'learning_rate': 9.30930930930931e-06, 'epoch': 0.28}

  9%|▉         | 1240/13317 [12:46<2:03:35,  1.63it/s]
  9%|▉         | 1241/13317 [12:47<2:03:34,  1.63it/s]
                                                      
{'loss': 0.3182, 'grad_norm': 0.5445021986961365, 'learning_rate': 9.316816816816818e-06, 'epoch': 0.28}

  9%|▉         | 1241/13317 [12:47<2:03:34,  1.63it/s]
  9%|▉         | 1242/13317 [12:47<2:03:37,  1.63it/s]
                                                      
{'loss': 0.2717, 'grad_norm': 0.5863715410232544, 'learning_rate': 9.324324324324325e-06, 'epoch': 0.28}

  9%|▉         | 1242/13317 [12:47<2:03:37,  1.63it/s]
  9%|▉         | 1243/13317 [12:48<2:03:35,  1.63it/s]
                                                      
{'loss': 0.3363, 'grad_norm': 0.6645376682281494, 'learning_rate': 9.331831831831832e-06, 'epoch': 0.28}

  9%|▉         | 1243/13317 [12:48<2:03:35,  1.63it/s]
  9%|▉         | 1244/13317 [12:49<2:03:45,  1.63it/s]
                                                      
{'loss': 0.3563, 'grad_norm': 0.5623888969421387, 'learning_rate': 9.339339339339341e-06, 'epoch': 0.28}

  9%|▉         | 1244/13317 [12:49<2:03:45,  1.63it/s]
  9%|▉         | 1245/13317 [12:49<2:03:42,  1.63it/s]
                                                      
{'loss': 0.3176, 'grad_norm': 0.5997814536094666, 'learning_rate': 9.346846846846848e-06, 'epoch': 0.28}

  9%|▉         | 1245/13317 [12:49<2:03:42,  1.63it/s]
  9%|▉         | 1246/13317 [12:50<2:03:35,  1.63it/s]
                                                      
{'loss': 0.274, 'grad_norm': 0.6043806672096252, 'learning_rate': 9.354354354354355e-06, 'epoch': 0.28}

  9%|▉         | 1246/13317 [12:50<2:03:35,  1.63it/s]
  9%|▉         | 1247/13317 [12:50<2:03:33,  1.63it/s]
                                                      
{'loss': 0.3799, 'grad_norm': 0.7212200164794922, 'learning_rate': 9.361861861861863e-06, 'epoch': 0.28}

  9%|▉         | 1247/13317 [12:50<2:03:33,  1.63it/s]
  9%|▉         | 1248/13317 [12:51<2:03:33,  1.63it/s]
                                                      
{'loss': 0.3493, 'grad_norm': 0.8611450791358948, 'learning_rate': 9.36936936936937e-06, 'epoch': 0.28}

  9%|▉         | 1248/13317 [12:51<2:03:33,  1.63it/s]
  9%|▉         | 1249/13317 [12:52<2:03:34,  1.63it/s]
                                                      
{'loss': 0.3973, 'grad_norm': 0.8325867652893066, 'learning_rate': 9.376876876876879e-06, 'epoch': 0.28}

  9%|▉         | 1249/13317 [12:52<2:03:34,  1.63it/s]
  9%|▉         | 1250/13317 [12:52<2:03:28,  1.63it/s]
                                                      
{'loss': 0.3166, 'grad_norm': 0.6236060261726379, 'learning_rate': 9.384384384384386e-06, 'epoch': 0.28}

  9%|▉         | 1250/13317 [12:52<2:03:28,  1.63it/s]
  9%|▉         | 1251/13317 [12:53<2:03:25,  1.63it/s]
                                                      
{'loss': 0.2597, 'grad_norm': 0.7103241086006165, 'learning_rate': 9.391891891891893e-06, 'epoch': 0.28}

  9%|▉         | 1251/13317 [12:53<2:03:25,  1.63it/s]
  9%|▉         | 1252/13317 [12:53<2:03:25,  1.63it/s]
                                                      
{'loss': 0.3058, 'grad_norm': 0.6358507871627808, 'learning_rate': 9.3993993993994e-06, 'epoch': 0.28}

  9%|▉         | 1252/13317 [12:54<2:03:25,  1.63it/s]
  9%|▉         | 1253/13317 [12:54<2:03:21,  1.63it/s]
                                                      
{'loss': 0.3093, 'grad_norm': 0.6396396160125732, 'learning_rate': 9.406906906906907e-06, 'epoch': 0.28}

  9%|▉         | 1253/13317 [12:54<2:03:21,  1.63it/s]
  9%|▉         | 1254/13317 [12:55<2:03:35,  1.63it/s]
                                                      
{'loss': 0.2701, 'grad_norm': 0.6948420405387878, 'learning_rate': 9.414414414414416e-06, 'epoch': 0.28}

  9%|▉         | 1254/13317 [12:55<2:03:35,  1.63it/s]
  9%|▉         | 1255/13317 [12:55<2:03:25,  1.63it/s]
                                                      
{'loss': 0.3524, 'grad_norm': 0.6439563632011414, 'learning_rate': 9.421921921921923e-06, 'epoch': 0.28}

  9%|▉         | 1255/13317 [12:55<2:03:25,  1.63it/s]
  9%|▉         | 1256/13317 [12:56<2:03:24,  1.63it/s]
                                                      
{'loss': 0.3067, 'grad_norm': 0.6722915768623352, 'learning_rate': 9.42942942942943e-06, 'epoch': 0.28}

  9%|▉         | 1256/13317 [12:56<2:03:24,  1.63it/s]
  9%|▉         | 1257/13317 [12:57<2:03:21,  1.63it/s]
                                                      
{'loss': 0.27, 'grad_norm': 0.6308885216712952, 'learning_rate': 9.436936936936937e-06, 'epoch': 0.28}

  9%|▉         | 1257/13317 [12:57<2:03:21,  1.63it/s]
  9%|▉         | 1258/13317 [12:57<2:03:19,  1.63it/s]
                                                      
{'loss': 0.3303, 'grad_norm': 0.7535562515258789, 'learning_rate': 9.444444444444445e-06, 'epoch': 0.28}

  9%|▉         | 1258/13317 [12:57<2:03:19,  1.63it/s]
  9%|▉         | 1259/13317 [12:58<2:03:24,  1.63it/s]
                                                      
{'loss': 0.2708, 'grad_norm': 0.5859531760215759, 'learning_rate': 9.451951951951953e-06, 'epoch': 0.28}

  9%|▉         | 1259/13317 [12:58<2:03:24,  1.63it/s]
  9%|▉         | 1260/13317 [12:58<2:03:22,  1.63it/s]
                                                      
{'loss': 0.3495, 'grad_norm': 0.6788567304611206, 'learning_rate': 9.45945945945946e-06, 'epoch': 0.28}

  9%|▉         | 1260/13317 [12:58<2:03:22,  1.63it/s]
  9%|▉         | 1261/13317 [12:59<2:03:54,  1.62it/s]
                                                      
{'loss': 0.4563, 'grad_norm': 0.7237330675125122, 'learning_rate': 9.466966966966968e-06, 'epoch': 0.28}

  9%|▉         | 1261/13317 [12:59<2:03:54,  1.62it/s]
  9%|▉         | 1262/13317 [13:00<2:03:41,  1.62it/s]
                                                      
{'loss': 0.3175, 'grad_norm': 0.6669389605522156, 'learning_rate': 9.474474474474475e-06, 'epoch': 0.28}

  9%|▉         | 1262/13317 [13:00<2:03:41,  1.62it/s]
  9%|▉         | 1263/13317 [13:00<2:03:33,  1.63it/s]
                                                      
{'loss': 0.2841, 'grad_norm': 0.7487065196037292, 'learning_rate': 9.481981981981982e-06, 'epoch': 0.28}

  9%|▉         | 1263/13317 [13:00<2:03:33,  1.63it/s]
  9%|▉         | 1264/13317 [13:01<2:03:35,  1.63it/s]
                                                      
{'loss': 0.4552, 'grad_norm': 0.7319648861885071, 'learning_rate': 9.489489489489491e-06, 'epoch': 0.28}

  9%|▉         | 1264/13317 [13:01<2:03:35,  1.63it/s]
  9%|▉         | 1265/13317 [13:01<2:03:27,  1.63it/s]
                                                      
{'loss': 0.3732, 'grad_norm': 0.8088781237602234, 'learning_rate': 9.496996996996998e-06, 'epoch': 0.28}

  9%|▉         | 1265/13317 [13:02<2:03:27,  1.63it/s]
 10%|▉         | 1266/13317 [13:02<2:03:18,  1.63it/s]
                                                      
{'loss': 0.3179, 'grad_norm': 0.6775524616241455, 'learning_rate': 9.504504504504505e-06, 'epoch': 0.29}

 10%|▉         | 1266/13317 [13:02<2:03:18,  1.63it/s]
 10%|▉         | 1267/13317 [13:03<2:03:17,  1.63it/s]
                                                      
{'loss': 0.3379, 'grad_norm': 0.6807026267051697, 'learning_rate': 9.512012012012012e-06, 'epoch': 0.29}

 10%|▉         | 1267/13317 [13:03<2:03:17,  1.63it/s]
 10%|▉         | 1268/13317 [13:03<2:03:13,  1.63it/s]
                                                      
{'loss': 0.3836, 'grad_norm': 0.7659010291099548, 'learning_rate': 9.51951951951952e-06, 'epoch': 0.29}

 10%|▉         | 1268/13317 [13:03<2:03:13,  1.63it/s]
 10%|▉         | 1269/13317 [13:04<2:03:25,  1.63it/s]
                                                      
{'loss': 0.3493, 'grad_norm': 0.729824423789978, 'learning_rate': 9.527027027027028e-06, 'epoch': 0.29}

 10%|▉         | 1269/13317 [13:04<2:03:25,  1.63it/s]
 10%|▉         | 1270/13317 [13:04<2:03:24,  1.63it/s]
                                                      
{'loss': 0.3191, 'grad_norm': 0.6416417956352234, 'learning_rate': 9.534534534534535e-06, 'epoch': 0.29}

 10%|▉         | 1270/13317 [13:05<2:03:24,  1.63it/s]
 10%|▉         | 1271/13317 [13:05<2:03:17,  1.63it/s]
                                                      
{'loss': 0.3638, 'grad_norm': 0.7341480851173401, 'learning_rate': 9.542042042042043e-06, 'epoch': 0.29}

 10%|▉         | 1271/13317 [13:05<2:03:17,  1.63it/s]
 10%|▉         | 1272/13317 [13:06<2:03:12,  1.63it/s]
                                                      
{'loss': 0.2754, 'grad_norm': 0.6575518250465393, 'learning_rate': 9.54954954954955e-06, 'epoch': 0.29}

 10%|▉         | 1272/13317 [13:06<2:03:12,  1.63it/s]
 10%|▉         | 1273/13317 [13:06<2:03:13,  1.63it/s]
                                                      
{'loss': 0.2944, 'grad_norm': 0.5514397025108337, 'learning_rate': 9.557057057057057e-06, 'epoch': 0.29}

 10%|▉         | 1273/13317 [13:06<2:03:13,  1.63it/s]
 10%|▉         | 1274/13317 [13:07<2:03:17,  1.63it/s]
                                                      
{'loss': 0.3026, 'grad_norm': 0.7235797643661499, 'learning_rate': 9.564564564564566e-06, 'epoch': 0.29}

 10%|▉         | 1274/13317 [13:07<2:03:17,  1.63it/s]
 10%|▉         | 1275/13317 [13:08<2:03:12,  1.63it/s]
                                                      
{'loss': 0.2892, 'grad_norm': 0.5703788995742798, 'learning_rate': 9.572072072072073e-06, 'epoch': 0.29}

 10%|▉         | 1275/13317 [13:08<2:03:12,  1.63it/s]
 10%|▉         | 1276/13317 [13:08<2:03:09,  1.63it/s]
                                                      
{'loss': 0.2839, 'grad_norm': 0.6485397815704346, 'learning_rate': 9.57957957957958e-06, 'epoch': 0.29}

 10%|▉         | 1276/13317 [13:08<2:03:09,  1.63it/s]
 10%|▉         | 1277/13317 [13:09<2:03:06,  1.63it/s]
                                                      
{'loss': 0.316, 'grad_norm': 0.7397987246513367, 'learning_rate': 9.587087087087087e-06, 'epoch': 0.29}

 10%|▉         | 1277/13317 [13:09<2:03:06,  1.63it/s]
 10%|▉         | 1278/13317 [13:09<2:03:01,  1.63it/s]
                                                      
{'loss': 0.282, 'grad_norm': 0.5860146880149841, 'learning_rate': 9.594594594594594e-06, 'epoch': 0.29}

 10%|▉         | 1278/13317 [13:10<2:03:01,  1.63it/s]
 10%|▉         | 1279/13317 [13:10<2:03:15,  1.63it/s]
                                                      
{'loss': 0.2725, 'grad_norm': 0.7169305086135864, 'learning_rate': 9.602102102102103e-06, 'epoch': 0.29}

 10%|▉         | 1279/13317 [13:10<2:03:15,  1.63it/s]
 10%|▉         | 1280/13317 [13:11<2:03:08,  1.63it/s]
                                                      
{'loss': 0.3863, 'grad_norm': 0.7372720241546631, 'learning_rate': 9.60960960960961e-06, 'epoch': 0.29}

 10%|▉         | 1280/13317 [13:11<2:03:08,  1.63it/s]
 10%|▉         | 1281/13317 [13:11<2:03:04,  1.63it/s]
                                                      
{'loss': 0.3231, 'grad_norm': 0.6847164630889893, 'learning_rate': 9.617117117117117e-06, 'epoch': 0.29}

 10%|▉         | 1281/13317 [13:11<2:03:04,  1.63it/s]
 10%|▉         | 1282/13317 [13:12<2:03:04,  1.63it/s]
                                                      
{'loss': 0.3389, 'grad_norm': 0.7193471789360046, 'learning_rate': 9.624624624624626e-06, 'epoch': 0.29}

 10%|▉         | 1282/13317 [13:12<2:03:04,  1.63it/s]
 10%|▉         | 1283/13317 [13:12<2:03:04,  1.63it/s]
                                                      
{'loss': 0.2761, 'grad_norm': 0.7160437703132629, 'learning_rate': 9.632132132132132e-06, 'epoch': 0.29}

 10%|▉         | 1283/13317 [13:13<2:03:04,  1.63it/s]
 10%|▉         | 1284/13317 [13:13<2:03:34,  1.62it/s]
                                                      
{'loss': 0.3498, 'grad_norm': 0.5738153457641602, 'learning_rate': 9.63963963963964e-06, 'epoch': 0.29}

 10%|▉         | 1284/13317 [13:13<2:03:34,  1.62it/s]
 10%|▉         | 1285/13317 [13:14<2:03:22,  1.63it/s]
                                                      
{'loss': 0.3137, 'grad_norm': 0.6597110629081726, 'learning_rate': 9.647147147147148e-06, 'epoch': 0.29}

 10%|▉         | 1285/13317 [13:14<2:03:22,  1.63it/s]
 10%|▉         | 1286/13317 [13:14<2:03:18,  1.63it/s]
                                                      
{'loss': 0.3403, 'grad_norm': 0.7840790748596191, 'learning_rate': 9.654654654654655e-06, 'epoch': 0.29}

 10%|▉         | 1286/13317 [13:14<2:03:18,  1.63it/s]
 10%|▉         | 1287/13317 [13:15<2:03:11,  1.63it/s]
                                                      
{'loss': 0.3006, 'grad_norm': 0.6983162760734558, 'learning_rate': 9.662162162162164e-06, 'epoch': 0.29}

 10%|▉         | 1287/13317 [13:15<2:03:11,  1.63it/s]
 10%|▉         | 1288/13317 [13:16<2:03:08,  1.63it/s]
                                                      
{'loss': 0.3293, 'grad_norm': 0.8246876001358032, 'learning_rate': 9.669669669669671e-06, 'epoch': 0.29}

 10%|▉         | 1288/13317 [13:16<2:03:08,  1.63it/s]
 10%|▉         | 1289/13317 [13:16<2:03:11,  1.63it/s]
                                                      
{'loss': 0.3113, 'grad_norm': 0.6906337738037109, 'learning_rate': 9.677177177177178e-06, 'epoch': 0.29}

 10%|▉         | 1289/13317 [13:16<2:03:11,  1.63it/s]
 10%|▉         | 1290/13317 [13:17<2:03:05,  1.63it/s]
                                                      
{'loss': 0.3476, 'grad_norm': 0.694891095161438, 'learning_rate': 9.684684684684685e-06, 'epoch': 0.29}

 10%|▉         | 1290/13317 [13:17<2:03:05,  1.63it/s]
 10%|▉         | 1291/13317 [13:17<2:03:05,  1.63it/s]
                                                      
{'loss': 0.2721, 'grad_norm': 0.7423439025878906, 'learning_rate': 9.692192192192194e-06, 'epoch': 0.29}

 10%|▉         | 1291/13317 [13:18<2:03:05,  1.63it/s]
 10%|▉         | 1292/13317 [13:18<2:03:02,  1.63it/s]
                                                      
{'loss': 0.3686, 'grad_norm': 0.8409532308578491, 'learning_rate': 9.699699699699701e-06, 'epoch': 0.29}

 10%|▉         | 1292/13317 [13:18<2:03:02,  1.63it/s]
 10%|▉         | 1293/13317 [13:19<2:03:02,  1.63it/s]
                                                      
{'loss': 0.3388, 'grad_norm': 0.708159863948822, 'learning_rate': 9.707207207207208e-06, 'epoch': 0.29}

 10%|▉         | 1293/13317 [13:19<2:03:02,  1.63it/s]
 10%|▉         | 1294/13317 [13:19<2:03:07,  1.63it/s]
                                                      
{'loss': 0.3123, 'grad_norm': 0.6437242031097412, 'learning_rate': 9.714714714714716e-06, 'epoch': 0.29}

 10%|▉         | 1294/13317 [13:19<2:03:07,  1.63it/s]
 10%|▉         | 1295/13317 [13:20<2:03:04,  1.63it/s]
                                                      
{'loss': 0.4013, 'grad_norm': 0.8258215188980103, 'learning_rate': 9.722222222222223e-06, 'epoch': 0.29}

 10%|▉         | 1295/13317 [13:20<2:03:04,  1.63it/s]
 10%|▉         | 1296/13317 [13:20<2:03:02,  1.63it/s]
                                                      
{'loss': 0.3287, 'grad_norm': 0.7083560228347778, 'learning_rate': 9.729729729729732e-06, 'epoch': 0.29}

 10%|▉         | 1296/13317 [13:21<2:03:02,  1.63it/s]
 10%|▉         | 1297/13317 [13:21<2:02:59,  1.63it/s]
                                                      
{'loss': 0.3302, 'grad_norm': 0.5983224511146545, 'learning_rate': 9.737237237237239e-06, 'epoch': 0.29}

 10%|▉         | 1297/13317 [13:21<2:02:59,  1.63it/s]
 10%|▉         | 1298/13317 [13:22<2:02:59,  1.63it/s]
                                                      
{'loss': 0.3047, 'grad_norm': 0.7723128199577332, 'learning_rate': 9.744744744744746e-06, 'epoch': 0.29}

 10%|▉         | 1298/13317 [13:22<2:02:59,  1.63it/s]
 10%|▉         | 1299/13317 [13:22<2:03:03,  1.63it/s]
                                                      
{'loss': 0.3004, 'grad_norm': 0.6936661005020142, 'learning_rate': 9.752252252252253e-06, 'epoch': 0.29}

 10%|▉         | 1299/13317 [13:22<2:03:03,  1.63it/s]
 10%|▉         | 1300/13317 [13:23<2:03:04,  1.63it/s]
                                                      
{'loss': 0.2268, 'grad_norm': 0.5858381390571594, 'learning_rate': 9.75975975975976e-06, 'epoch': 0.29}

 10%|▉         | 1300/13317 [13:23<2:03:04,  1.63it/s]
 10%|▉         | 1301/13317 [13:24<2:02:58,  1.63it/s]
                                                      
{'loss': 0.2857, 'grad_norm': 0.5878675580024719, 'learning_rate': 9.767267267267269e-06, 'epoch': 0.29}

 10%|▉         | 1301/13317 [13:24<2:02:58,  1.63it/s]
 10%|▉         | 1302/13317 [13:24<2:02:53,  1.63it/s]
                                                      
{'loss': 0.3109, 'grad_norm': 0.5702260136604309, 'learning_rate': 9.774774774774776e-06, 'epoch': 0.29}

 10%|▉         | 1302/13317 [13:24<2:02:53,  1.63it/s]
 10%|▉         | 1303/13317 [13:25<2:02:51,  1.63it/s]
                                                      
{'loss': 0.3293, 'grad_norm': 0.6797556281089783, 'learning_rate': 9.782282282282283e-06, 'epoch': 0.29}

 10%|▉         | 1303/13317 [13:25<2:02:51,  1.63it/s]
 10%|▉         | 1304/13317 [13:25<2:02:56,  1.63it/s]
                                                      
{'loss': 0.3527, 'grad_norm': 0.7031927704811096, 'learning_rate': 9.78978978978979e-06, 'epoch': 0.29}

 10%|▉         | 1304/13317 [13:26<2:02:56,  1.63it/s]
 10%|▉         | 1305/13317 [13:26<2:02:53,  1.63it/s]
                                                      
{'loss': 0.3164, 'grad_norm': 0.676983654499054, 'learning_rate': 9.797297297297298e-06, 'epoch': 0.29}

 10%|▉         | 1305/13317 [13:26<2:02:53,  1.63it/s]
 10%|▉         | 1306/13317 [13:27<2:02:47,  1.63it/s]
                                                      
{'loss': 0.3372, 'grad_norm': 0.6821779608726501, 'learning_rate': 9.804804804804806e-06, 'epoch': 0.29}

 10%|▉         | 1306/13317 [13:27<2:02:47,  1.63it/s]
 10%|▉         | 1307/13317 [13:27<2:02:50,  1.63it/s]
                                                      
{'loss': 0.2866, 'grad_norm': 0.6401535868644714, 'learning_rate': 9.812312312312314e-06, 'epoch': 0.29}

 10%|▉         | 1307/13317 [13:27<2:02:50,  1.63it/s]
 10%|▉         | 1308/13317 [13:28<2:02:47,  1.63it/s]
                                                      
{'loss': 0.2913, 'grad_norm': 0.621741771697998, 'learning_rate': 9.81981981981982e-06, 'epoch': 0.29}

 10%|▉         | 1308/13317 [13:28<2:02:47,  1.63it/s]
 10%|▉         | 1309/13317 [13:28<2:03:00,  1.63it/s]
                                                      
{'loss': 0.2249, 'grad_norm': 0.5369464159011841, 'learning_rate': 9.827327327327328e-06, 'epoch': 0.29}

 10%|▉         | 1309/13317 [13:29<2:03:00,  1.63it/s]
 10%|▉         | 1310/13317 [13:29<2:03:00,  1.63it/s]
                                                      
{'loss': 0.3565, 'grad_norm': 0.6252511143684387, 'learning_rate': 9.834834834834835e-06, 'epoch': 0.3}

 10%|▉         | 1310/13317 [13:29<2:03:00,  1.63it/s]
 10%|▉         | 1311/13317 [13:30<2:02:53,  1.63it/s]
                                                      
{'loss': 0.3119, 'grad_norm': 0.649142324924469, 'learning_rate': 9.842342342342344e-06, 'epoch': 0.3}

 10%|▉         | 1311/13317 [13:30<2:02:53,  1.63it/s]
 10%|▉         | 1312/13317 [13:30<2:02:52,  1.63it/s]
                                                      
{'loss': 0.3863, 'grad_norm': 0.9470739960670471, 'learning_rate': 9.849849849849851e-06, 'epoch': 0.3}

 10%|▉         | 1312/13317 [13:30<2:02:52,  1.63it/s]
 10%|▉         | 1313/13317 [13:31<2:02:52,  1.63it/s]
                                                      
{'loss': 0.3062, 'grad_norm': 0.7411321997642517, 'learning_rate': 9.857357357357358e-06, 'epoch': 0.3}

 10%|▉         | 1313/13317 [13:31<2:02:52,  1.63it/s]
 10%|▉         | 1314/13317 [13:32<2:02:53,  1.63it/s]
                                                      
{'loss': 0.2882, 'grad_norm': 0.6262528300285339, 'learning_rate': 9.864864864864865e-06, 'epoch': 0.3}

 10%|▉         | 1314/13317 [13:32<2:02:53,  1.63it/s]
 10%|▉         | 1315/13317 [13:32<2:02:52,  1.63it/s]
                                                      
{'loss': 0.3534, 'grad_norm': 0.7302519083023071, 'learning_rate': 9.872372372372372e-06, 'epoch': 0.3}

 10%|▉         | 1315/13317 [13:32<2:02:52,  1.63it/s]
 10%|▉         | 1316/13317 [13:33<2:02:50,  1.63it/s]
                                                      
{'loss': 0.3298, 'grad_norm': 0.6734743118286133, 'learning_rate': 9.879879879879881e-06, 'epoch': 0.3}

 10%|▉         | 1316/13317 [13:33<2:02:50,  1.63it/s]
 10%|▉         | 1317/13317 [13:33<2:02:48,  1.63it/s]
                                                      
{'loss': 0.3278, 'grad_norm': 0.6313687562942505, 'learning_rate': 9.887387387387388e-06, 'epoch': 0.3}

 10%|▉         | 1317/13317 [13:33<2:02:48,  1.63it/s]
 10%|▉         | 1318/13317 [13:34<2:02:45,  1.63it/s]
                                                      
{'loss': 0.2918, 'grad_norm': 0.6062349677085876, 'learning_rate': 9.894894894894896e-06, 'epoch': 0.3}

 10%|▉         | 1318/13317 [13:34<2:02:45,  1.63it/s]
 10%|▉         | 1319/13317 [13:35<2:02:45,  1.63it/s]
                                                      
{'loss': 0.2691, 'grad_norm': 0.6356523633003235, 'learning_rate': 9.902402402402403e-06, 'epoch': 0.3}

 10%|▉         | 1319/13317 [13:35<2:02:45,  1.63it/s]
 10%|▉         | 1320/13317 [13:35<2:02:40,  1.63it/s]
                                                      
{'loss': 0.3038, 'grad_norm': 0.6323164105415344, 'learning_rate': 9.90990990990991e-06, 'epoch': 0.3}

 10%|▉         | 1320/13317 [13:35<2:02:40,  1.63it/s]
 10%|▉         | 1321/13317 [13:36<2:02:37,  1.63it/s]
                                                      
{'loss': 0.3141, 'grad_norm': 0.6173000931739807, 'learning_rate': 9.917417417417419e-06, 'epoch': 0.3}

 10%|▉         | 1321/13317 [13:36<2:02:37,  1.63it/s]
 10%|▉         | 1322/13317 [13:36<2:02:37,  1.63it/s]
                                                      
{'loss': 0.423, 'grad_norm': 0.6600233912467957, 'learning_rate': 9.924924924924926e-06, 'epoch': 0.3}

 10%|▉         | 1322/13317 [13:37<2:02:37,  1.63it/s]
 10%|▉         | 1323/13317 [13:37<2:02:38,  1.63it/s]
                                                      
{'loss': 0.2633, 'grad_norm': 0.6221367716789246, 'learning_rate': 9.932432432432433e-06, 'epoch': 0.3}

 10%|▉         | 1323/13317 [13:37<2:02:38,  1.63it/s]
 10%|▉         | 1324/13317 [13:38<2:02:34,  1.63it/s]
                                                      
{'loss': 0.2971, 'grad_norm': 0.6465907692909241, 'learning_rate': 9.93993993993994e-06, 'epoch': 0.3}

 10%|▉         | 1324/13317 [13:38<2:02:34,  1.63it/s]
 10%|▉         | 1325/13317 [13:38<2:02:37,  1.63it/s]
                                                      
{'loss': 0.2664, 'grad_norm': 0.6048935055732727, 'learning_rate': 9.947447447447447e-06, 'epoch': 0.3}

 10%|▉         | 1325/13317 [13:38<2:02:37,  1.63it/s]
 10%|▉         | 1326/13317 [13:39<2:02:36,  1.63it/s]
                                                      
{'loss': 0.2642, 'grad_norm': 0.6517884135246277, 'learning_rate': 9.954954954954956e-06, 'epoch': 0.3}

 10%|▉         | 1326/13317 [13:39<2:02:36,  1.63it/s]
 10%|▉         | 1327/13317 [13:39<2:02:33,  1.63it/s]
                                                      
{'loss': 0.4062, 'grad_norm': 0.7850205302238464, 'learning_rate': 9.962462462462463e-06, 'epoch': 0.3}

 10%|▉         | 1327/13317 [13:40<2:02:33,  1.63it/s]
 10%|▉         | 1328/13317 [13:40<2:02:33,  1.63it/s]
                                                      
{'loss': 0.4018, 'grad_norm': 0.8096587061882019, 'learning_rate': 9.96996996996997e-06, 'epoch': 0.3}

 10%|▉         | 1328/13317 [13:40<2:02:33,  1.63it/s]
 10%|▉         | 1329/13317 [13:41<2:02:34,  1.63it/s]
                                                      
{'loss': 0.2809, 'grad_norm': 0.5801488757133484, 'learning_rate': 9.97747747747748e-06, 'epoch': 0.3}

 10%|▉         | 1329/13317 [13:41<2:02:34,  1.63it/s]
 10%|▉         | 1330/13317 [13:41<2:02:34,  1.63it/s]
                                                      
{'loss': 0.3299, 'grad_norm': 0.7098427414894104, 'learning_rate': 9.984984984984985e-06, 'epoch': 0.3}

 10%|▉         | 1330/13317 [13:41<2:02:34,  1.63it/s]
 10%|▉         | 1331/13317 [13:42<2:02:33,  1.63it/s]
                                                      
{'loss': 0.3013, 'grad_norm': 0.8151565790176392, 'learning_rate': 9.992492492492494e-06, 'epoch': 0.3}

 10%|▉         | 1331/13317 [13:42<2:02:33,  1.63it/s]
 10%|█         | 1332/13317 [13:43<2:02:31,  1.63it/s]
                                                      
{'loss': 0.3167, 'grad_norm': 0.6404622197151184, 'learning_rate': 1e-05, 'epoch': 0.3}

 10%|█         | 1332/13317 [13:43<2:02:31,  1.63it/s]
 10%|█         | 1333/13317 [13:43<2:02:27,  1.63it/s]
                                                      
{'loss': 0.2596, 'grad_norm': 0.6925714612007141, 'learning_rate': 9.99999982822353e-06, 'epoch': 0.3}

 10%|█         | 1333/13317 [13:43<2:02:27,  1.63it/s]
 10%|█         | 1334/13317 [13:44<2:02:28,  1.63it/s]
                                                      
{'loss': 0.2353, 'grad_norm': 0.5570275783538818, 'learning_rate': 9.999999312894131e-06, 'epoch': 0.3}

 10%|█         | 1334/13317 [13:44<2:02:28,  1.63it/s]
 10%|█         | 1335/13317 [13:44<2:02:34,  1.63it/s]
                                                      
{'loss': 0.4126, 'grad_norm': 0.8844746947288513, 'learning_rate': 9.999998454011837e-06, 'epoch': 0.3}

 10%|█         | 1335/13317 [13:45<2:02:34,  1.63it/s]
 10%|█         | 1336/13317 [13:45<2:02:32,  1.63it/s]
                                                      
{'loss': 0.3414, 'grad_norm': 0.7237457633018494, 'learning_rate': 9.99999725157671e-06, 'epoch': 0.3}

 10%|█         | 1336/13317 [13:45<2:02:32,  1.63it/s]
 10%|█         | 1337/13317 [13:46<2:02:31,  1.63it/s]
                                                      
{'loss': 0.3099, 'grad_norm': 0.6600624918937683, 'learning_rate': 9.99999570558883e-06, 'epoch': 0.3}

 10%|█         | 1337/13317 [13:46<2:02:31,  1.63it/s]
 10%|█         | 1338/13317 [13:46<2:02:28,  1.63it/s]
                                                      
{'loss': 0.2733, 'grad_norm': 0.5687398314476013, 'learning_rate': 9.999993816048306e-06, 'epoch': 0.3}

 10%|█         | 1338/13317 [13:46<2:02:28,  1.63it/s]
 10%|█         | 1339/13317 [13:47<2:02:27,  1.63it/s]
                                                      
{'loss': 0.3382, 'grad_norm': 0.7302866578102112, 'learning_rate': 9.999991582955264e-06, 'epoch': 0.3}

 10%|█         | 1339/13317 [13:47<2:02:27,  1.63it/s]
 10%|█         | 1340/13317 [13:47<2:02:31,  1.63it/s]
                                                      
{'loss': 0.3101, 'grad_norm': 0.6116707921028137, 'learning_rate': 9.99998900630986e-06, 'epoch': 0.3}

 10%|█         | 1340/13317 [13:48<2:02:31,  1.63it/s]
 10%|█         | 1341/13317 [13:48<2:02:35,  1.63it/s]
                                                      
{'loss': 0.3139, 'grad_norm': 0.6210039258003235, 'learning_rate': 9.99998608611227e-06, 'epoch': 0.3}

 10%|█         | 1341/13317 [13:48<2:02:35,  1.63it/s]
 10%|█         | 1342/13317 [13:49<2:02:35,  1.63it/s]
                                                      
{'loss': 0.3333, 'grad_norm': 0.6442962884902954, 'learning_rate': 9.999982822362696e-06, 'epoch': 0.3}

 10%|█         | 1342/13317 [13:49<2:02:35,  1.63it/s]
 10%|█         | 1343/13317 [13:49<2:02:32,  1.63it/s]
                                                      
{'loss': 0.3162, 'grad_norm': 0.6589913368225098, 'learning_rate': 9.999979215061362e-06, 'epoch': 0.3}

 10%|█         | 1343/13317 [13:49<2:02:32,  1.63it/s]
 10%|█         | 1344/13317 [13:50<2:02:32,  1.63it/s]
                                                      
{'loss': 0.3231, 'grad_norm': 0.6462382078170776, 'learning_rate': 9.999975264208513e-06, 'epoch': 0.3}

 10%|█         | 1344/13317 [13:50<2:02:32,  1.63it/s]
 10%|█         | 1345/13317 [13:51<2:02:27,  1.63it/s]
                                                      
{'loss': 0.3312, 'grad_norm': 0.6308842897415161, 'learning_rate': 9.999970969804424e-06, 'epoch': 0.3}

 10%|█         | 1345/13317 [13:51<2:02:27,  1.63it/s]
 10%|█         | 1346/13317 [13:51<2:02:23,  1.63it/s]
                                                      
{'loss': 0.2932, 'grad_norm': 0.6171318888664246, 'learning_rate': 9.99996633184939e-06, 'epoch': 0.3}

 10%|█         | 1346/13317 [13:51<2:02:23,  1.63it/s]
 10%|█         | 1347/13317 [13:52<2:02:23,  1.63it/s]
                                                      
{'loss': 0.2811, 'grad_norm': 0.6225621104240417, 'learning_rate': 9.999961350343727e-06, 'epoch': 0.3}

 10%|█         | 1347/13317 [13:52<2:02:23,  1.63it/s]
 10%|█         | 1348/13317 [13:52<2:02:21,  1.63it/s]
                                                      
{'loss': 0.3156, 'grad_norm': 0.7729416489601135, 'learning_rate': 9.999956025287781e-06, 'epoch': 0.3}

 10%|█         | 1348/13317 [13:53<2:02:21,  1.63it/s]
 10%|█         | 1349/13317 [13:53<2:02:24,  1.63it/s]
                                                      
{'loss': 0.2695, 'grad_norm': 0.5477426052093506, 'learning_rate': 9.999950356681914e-06, 'epoch': 0.3}

 10%|█         | 1349/13317 [13:53<2:02:24,  1.63it/s]
 10%|█         | 1350/13317 [13:54<2:02:21,  1.63it/s]
                                                      
{'loss': 0.2946, 'grad_norm': 0.6637920141220093, 'learning_rate': 9.999944344526515e-06, 'epoch': 0.3}

 10%|█         | 1350/13317 [13:54<2:02:21,  1.63it/s]
 10%|█         | 1351/13317 [13:54<2:02:16,  1.63it/s]
                                                      
{'loss': 0.3282, 'grad_norm': 0.6120142936706543, 'learning_rate': 9.999937988822002e-06, 'epoch': 0.3}

 10%|█         | 1351/13317 [13:54<2:02:16,  1.63it/s]
 10%|█         | 1352/13317 [13:55<2:02:18,  1.63it/s]
                                                      
{'loss': 0.2719, 'grad_norm': 0.5910621881484985, 'learning_rate': 9.99993128956881e-06, 'epoch': 0.3}

 10%|█         | 1352/13317 [13:55<2:02:18,  1.63it/s]
 10%|█         | 1353/13317 [13:55<2:02:18,  1.63it/s]
                                                      
{'loss': 0.2275, 'grad_norm': 0.5807482004165649, 'learning_rate': 9.999924246767396e-06, 'epoch': 0.3}

 10%|█         | 1353/13317 [13:56<2:02:18,  1.63it/s]
 10%|█         | 1354/13317 [13:56<2:02:19,  1.63it/s]
                                                      
{'loss': 0.3295, 'grad_norm': 0.6461781859397888, 'learning_rate': 9.999916860418246e-06, 'epoch': 0.31}

 10%|█         | 1354/13317 [13:56<2:02:19,  1.63it/s]
 10%|█         | 1355/13317 [13:57<2:02:19,  1.63it/s]
                                                      
{'loss': 0.3284, 'grad_norm': 0.6314394474029541, 'learning_rate': 9.99990913052187e-06, 'epoch': 0.31}

 10%|█         | 1355/13317 [13:57<2:02:19,  1.63it/s]
 10%|█         | 1356/13317 [13:57<2:02:20,  1.63it/s]
                                                      
{'loss': 0.3953, 'grad_norm': 0.7046138048171997, 'learning_rate': 9.999901057078796e-06, 'epoch': 0.31}

 10%|█         | 1356/13317 [13:57<2:02:20,  1.63it/s]
 10%|█         | 1357/13317 [13:58<2:02:17,  1.63it/s]
                                                      
{'loss': 0.3351, 'grad_norm': 0.7907109260559082, 'learning_rate': 9.999892640089578e-06, 'epoch': 0.31}

 10%|█         | 1357/13317 [13:58<2:02:17,  1.63it/s]
 10%|█         | 1358/13317 [13:59<2:02:13,  1.63it/s]
                                                      
{'loss': 0.3609, 'grad_norm': 0.6969332098960876, 'learning_rate': 9.999883879554797e-06, 'epoch': 0.31}

 10%|█         | 1358/13317 [13:59<2:02:13,  1.63it/s]
 10%|█         | 1359/13317 [13:59<2:02:16,  1.63it/s]
                                                      
{'loss': 0.37, 'grad_norm': 0.6634169816970825, 'learning_rate': 9.999874775475053e-06, 'epoch': 0.31}

 10%|█         | 1359/13317 [13:59<2:02:16,  1.63it/s]
 10%|█         | 1360/13317 [14:00<2:02:15,  1.63it/s]
                                                      
{'loss': 0.3089, 'grad_norm': 0.6534088850021362, 'learning_rate': 9.999865327850975e-06, 'epoch': 0.31}

 10%|█         | 1360/13317 [14:00<2:02:15,  1.63it/s]
 10%|█         | 1361/13317 [14:00<2:02:13,  1.63it/s]
                                                      
{'loss': 0.2613, 'grad_norm': 0.5772269368171692, 'learning_rate': 9.999855536683207e-06, 'epoch': 0.31}

 10%|█         | 1361/13317 [14:00<2:02:13,  1.63it/s]
 10%|█         | 1362/13317 [14:01<2:02:13,  1.63it/s]
                                                      
{'loss': 0.256, 'grad_norm': 0.5818325281143188, 'learning_rate': 9.999845401972426e-06, 'epoch': 0.31}

 10%|█         | 1362/13317 [14:01<2:02:13,  1.63it/s]
 10%|█         | 1363/13317 [14:02<2:02:11,  1.63it/s]
                                                      
{'loss': 0.3284, 'grad_norm': 0.6786175966262817, 'learning_rate': 9.999834923719324e-06, 'epoch': 0.31}

 10%|█         | 1363/13317 [14:02<2:02:11,  1.63it/s]
 10%|█         | 1364/13317 [14:02<2:02:11,  1.63it/s]
                                                      
{'loss': 0.325, 'grad_norm': 0.7166447639465332, 'learning_rate': 9.999824101924628e-06, 'epoch': 0.31}

 10%|█         | 1364/13317 [14:02<2:02:11,  1.63it/s]
 10%|█         | 1365/13317 [14:03<2:02:10,  1.63it/s]
                                                      
{'loss': 0.2733, 'grad_norm': 0.7198257446289062, 'learning_rate': 9.999812936589073e-06, 'epoch': 0.31}

 10%|█         | 1365/13317 [14:03<2:02:10,  1.63it/s]
 10%|█         | 1366/13317 [14:03<2:02:09,  1.63it/s]
                                                      
{'loss': 0.2956, 'grad_norm': 0.625848650932312, 'learning_rate': 9.999801427713432e-06, 'epoch': 0.31}

 10%|█         | 1366/13317 [14:04<2:02:09,  1.63it/s]
 10%|█         | 1367/13317 [14:04<2:02:11,  1.63it/s]
                                                      
{'loss': 0.2564, 'grad_norm': 0.6776215434074402, 'learning_rate': 9.999789575298495e-06, 'epoch': 0.31}

 10%|█         | 1367/13317 [14:04<2:02:11,  1.63it/s]
 10%|█         | 1368/13317 [14:05<2:02:08,  1.63it/s]
                                                      
{'loss': 0.3401, 'grad_norm': 0.6745787262916565, 'learning_rate': 9.999777379345075e-06, 'epoch': 0.31}

 10%|█         | 1368/13317 [14:05<2:02:08,  1.63it/s]
 10%|█         | 1369/13317 [14:05<2:02:10,  1.63it/s]
                                                      
{'loss': 0.3349, 'grad_norm': 0.6506442427635193, 'learning_rate': 9.99976483985401e-06, 'epoch': 0.31}

 10%|█         | 1369/13317 [14:05<2:02:10,  1.63it/s]
 10%|█         | 1370/13317 [14:06<2:02:07,  1.63it/s]
                                                      
{'loss': 0.3811, 'grad_norm': 0.7130613923072815, 'learning_rate': 9.999751956826162e-06, 'epoch': 0.31}

 10%|█         | 1370/13317 [14:06<2:02:07,  1.63it/s]
 10%|█         | 1371/13317 [14:06<2:02:05,  1.63it/s]
                                                      
{'loss': 0.231, 'grad_norm': 0.6159073114395142, 'learning_rate': 9.999738730262417e-06, 'epoch': 0.31}

 10%|█         | 1371/13317 [14:07<2:02:05,  1.63it/s]
 10%|█         | 1372/13317 [14:07<2:02:05,  1.63it/s]
                                                      
{'loss': 0.2805, 'grad_norm': 0.5981456637382507, 'learning_rate': 9.999725160163683e-06, 'epoch': 0.31}

 10%|█         | 1372/13317 [14:07<2:02:05,  1.63it/s]
 10%|█         | 1373/13317 [14:08<2:02:04,  1.63it/s]
                                                      
{'loss': 0.2953, 'grad_norm': 0.7038057446479797, 'learning_rate': 9.999711246530893e-06, 'epoch': 0.31}

 10%|█         | 1373/13317 [14:08<2:02:04,  1.63it/s]
 10%|█         | 1374/13317 [14:08<2:02:05,  1.63it/s]
                                                      
{'loss': 0.366, 'grad_norm': 0.7469061613082886, 'learning_rate': 9.999696989365002e-06, 'epoch': 0.31}

 10%|█         | 1374/13317 [14:08<2:02:05,  1.63it/s]
 10%|█         | 1375/13317 [14:09<2:02:02,  1.63it/s]
                                                      
{'loss': 0.2555, 'grad_norm': 0.5987260341644287, 'learning_rate': 9.99968238866699e-06, 'epoch': 0.31}

 10%|█         | 1375/13317 [14:09<2:02:02,  1.63it/s]
 10%|█         | 1376/13317 [14:10<2:02:03,  1.63it/s]
                                                      
{'loss': 0.3317, 'grad_norm': 0.6385735273361206, 'learning_rate': 9.99966744443786e-06, 'epoch': 0.31}

 10%|█         | 1376/13317 [14:10<2:02:03,  1.63it/s]
 10%|█         | 1377/13317 [14:10<2:02:01,  1.63it/s]
                                                      
{'loss': 0.3583, 'grad_norm': 0.6880871653556824, 'learning_rate': 9.999652156678641e-06, 'epoch': 0.31}

 10%|█         | 1377/13317 [14:10<2:02:01,  1.63it/s]
 10%|█         | 1378/13317 [14:11<2:02:07,  1.63it/s]
                                                      
{'loss': 0.4055, 'grad_norm': 0.7010153532028198, 'learning_rate': 9.99963652539038e-06, 'epoch': 0.31}

 10%|█         | 1378/13317 [14:11<2:02:07,  1.63it/s]
 10%|█         | 1379/13317 [14:11<2:02:07,  1.63it/s]
                                                      
{'loss': 0.3819, 'grad_norm': 0.7465724945068359, 'learning_rate': 9.999620550574155e-06, 'epoch': 0.31}

 10%|█         | 1379/13317 [14:12<2:02:07,  1.63it/s]
 10%|█         | 1380/13317 [14:12<2:02:11,  1.63it/s]
                                                      
{'loss': 0.3403, 'grad_norm': 0.723497211933136, 'learning_rate': 9.999604232231058e-06, 'epoch': 0.31}

 10%|█         | 1380/13317 [14:12<2:02:11,  1.63it/s]
 10%|█         | 1381/13317 [14:13<2:02:07,  1.63it/s]
                                                      
{'loss': 0.4083, 'grad_norm': 0.7631259560585022, 'learning_rate': 9.999587570362216e-06, 'epoch': 0.31}

 10%|█         | 1381/13317 [14:13<2:02:07,  1.63it/s]
 10%|█         | 1382/13317 [14:13<2:02:07,  1.63it/s]
                                                      
{'loss': 0.2765, 'grad_norm': 0.6012881398200989, 'learning_rate': 9.99957056496877e-06, 'epoch': 0.31}

 10%|█         | 1382/13317 [14:13<2:02:07,  1.63it/s]
 10%|█         | 1383/13317 [14:14<2:02:08,  1.63it/s]
                                                      
{'loss': 0.3806, 'grad_norm': 0.6239575147628784, 'learning_rate': 9.999553216051892e-06, 'epoch': 0.31}

 10%|█         | 1383/13317 [14:14<2:02:08,  1.63it/s]
 10%|█         | 1384/13317 [14:14<2:02:08,  1.63it/s]
                                                      
{'loss': 0.3154, 'grad_norm': 0.6220383048057556, 'learning_rate': 9.999535523612768e-06, 'epoch': 0.31}

 10%|█         | 1384/13317 [14:15<2:02:08,  1.63it/s]
 10%|█         | 1385/13317 [14:15<2:02:07,  1.63it/s]
                                                      
{'loss': 0.2715, 'grad_norm': 0.6285386085510254, 'learning_rate': 9.999517487652622e-06, 'epoch': 0.31}

 10%|█         | 1385/13317 [14:15<2:02:07,  1.63it/s]
 10%|█         | 1386/13317 [14:16<2:02:03,  1.63it/s]
                                                      
{'loss': 0.3514, 'grad_norm': 0.708305835723877, 'learning_rate': 9.999499108172687e-06, 'epoch': 0.31}

 10%|█         | 1386/13317 [14:16<2:02:03,  1.63it/s]
 10%|█         | 1387/13317 [14:16<2:02:01,  1.63it/s]
                                                      
{'loss': 0.374, 'grad_norm': 0.7247787714004517, 'learning_rate': 9.999480385174228e-06, 'epoch': 0.31}

 10%|█         | 1387/13317 [14:16<2:02:01,  1.63it/s]
 10%|█         | 1388/13317 [14:17<2:01:59,  1.63it/s]
                                                      
{'loss': 0.3292, 'grad_norm': 0.7473137974739075, 'learning_rate': 9.99946131865853e-06, 'epoch': 0.31}

 10%|█         | 1388/13317 [14:17<2:01:59,  1.63it/s]
 10%|█         | 1389/13317 [14:18<2:01:59,  1.63it/s]
                                                      
{'loss': 0.2954, 'grad_norm': 0.5819560289382935, 'learning_rate': 9.999441908626906e-06, 'epoch': 0.31}

 10%|█         | 1389/13317 [14:18<2:01:59,  1.63it/s]
 10%|█         | 1390/13317 [14:18<2:02:02,  1.63it/s]
                                                      
{'loss': 0.2775, 'grad_norm': 0.5817522406578064, 'learning_rate': 9.999422155080687e-06, 'epoch': 0.31}

 10%|█         | 1390/13317 [14:18<2:02:02,  1.63it/s]
 10%|█         | 1391/13317 [14:19<2:02:26,  1.62it/s]
                                                      
{'loss': 0.2918, 'grad_norm': 0.5886651873588562, 'learning_rate': 9.99940205802123e-06, 'epoch': 0.31}

 10%|█         | 1391/13317 [14:19<2:02:26,  1.62it/s]
 10%|█         | 1392/13317 [14:19<2:02:16,  1.63it/s]
                                                      
{'loss': 0.2566, 'grad_norm': 0.5706602334976196, 'learning_rate': 9.99938161744992e-06, 'epoch': 0.31}

 10%|█         | 1392/13317 [14:20<2:02:16,  1.63it/s]
 10%|█         | 1393/13317 [14:20<2:02:09,  1.63it/s]
                                                      
{'loss': 0.3031, 'grad_norm': 0.6992654204368591, 'learning_rate': 9.999360833368156e-06, 'epoch': 0.31}

 10%|█         | 1393/13317 [14:20<2:02:09,  1.63it/s]
 10%|█         | 1394/13317 [14:21<2:02:04,  1.63it/s]
                                                      
{'loss': 0.2976, 'grad_norm': 0.5781307220458984, 'learning_rate': 9.999339705777368e-06, 'epoch': 0.31}

 10%|█         | 1394/13317 [14:21<2:02:04,  1.63it/s]
 10%|█         | 1395/13317 [14:21<2:02:02,  1.63it/s]
                                                      
{'loss': 0.2635, 'grad_norm': 0.610206127166748, 'learning_rate': 9.999318234679011e-06, 'epoch': 0.31}

 10%|█         | 1395/13317 [14:21<2:02:02,  1.63it/s]
 10%|█         | 1396/13317 [14:22<2:01:59,  1.63it/s]
                                                      
{'loss': 0.2321, 'grad_norm': 0.5491091012954712, 'learning_rate': 9.999296420074559e-06, 'epoch': 0.31}

 10%|█         | 1396/13317 [14:22<2:01:59,  1.63it/s]
 10%|█         | 1397/13317 [14:22<2:01:56,  1.63it/s]
                                                      
{'loss': 0.28, 'grad_norm': 0.6310189962387085, 'learning_rate': 9.999274261965506e-06, 'epoch': 0.31}

 10%|█         | 1397/13317 [14:23<2:01:56,  1.63it/s]
 10%|█         | 1398/13317 [14:23<2:01:54,  1.63it/s]
                                                      
{'loss': 0.3536, 'grad_norm': 0.6551162600517273, 'learning_rate': 9.99925176035338e-06, 'epoch': 0.31}

 10%|█         | 1398/13317 [14:23<2:01:54,  1.63it/s]
 11%|█         | 1399/13317 [14:24<2:01:59,  1.63it/s]
                                                      
{'loss': 0.3086, 'grad_norm': 0.7214124798774719, 'learning_rate': 9.999228915239722e-06, 'epoch': 0.32}

 11%|█         | 1399/13317 [14:24<2:01:59,  1.63it/s]
 11%|█         | 1400/13317 [14:24<2:01:56,  1.63it/s]
                                                      
{'loss': 0.2696, 'grad_norm': 0.5803145170211792, 'learning_rate': 9.999205726626109e-06, 'epoch': 0.32}

 11%|█         | 1400/13317 [14:24<2:01:56,  1.63it/s]
 11%|█         | 1401/13317 [14:25<2:01:58,  1.63it/s]
                                                      
{'loss': 0.3188, 'grad_norm': 0.6992409825325012, 'learning_rate': 9.999182194514128e-06, 'epoch': 0.32}

 11%|█         | 1401/13317 [14:25<2:01:58,  1.63it/s]
 11%|█         | 1402/13317 [14:26<2:01:58,  1.63it/s]
                                                      
{'loss': 0.3029, 'grad_norm': 0.6191926598548889, 'learning_rate': 9.999158318905398e-06, 'epoch': 0.32}

 11%|█         | 1402/13317 [14:26<2:01:58,  1.63it/s]
 11%|█         | 1403/13317 [14:26<2:01:54,  1.63it/s]
                                                      
{'loss': 0.2737, 'grad_norm': 0.6033204793930054, 'learning_rate': 9.999134099801559e-06, 'epoch': 0.32}

 11%|█         | 1403/13317 [14:26<2:01:54,  1.63it/s]
 11%|█         | 1404/13317 [14:27<2:01:57,  1.63it/s]
                                                      
{'loss': 0.2714, 'grad_norm': 0.6543962359428406, 'learning_rate': 9.999109537204277e-06, 'epoch': 0.32}

 11%|█         | 1404/13317 [14:27<2:01:57,  1.63it/s]
 11%|█         | 1405/13317 [14:27<2:01:50,  1.63it/s]
                                                      
{'loss': 0.3597, 'grad_norm': 0.6664211750030518, 'learning_rate': 9.999084631115238e-06, 'epoch': 0.32}

 11%|█         | 1405/13317 [14:27<2:01:50,  1.63it/s]
 11%|█         | 1406/13317 [14:28<2:01:51,  1.63it/s]
                                                      
{'loss': 0.3245, 'grad_norm': 0.661435604095459, 'learning_rate': 9.999059381536154e-06, 'epoch': 0.32}

 11%|█         | 1406/13317 [14:28<2:01:51,  1.63it/s]
 11%|█         | 1407/13317 [14:29<2:01:50,  1.63it/s]
                                                      
{'loss': 0.269, 'grad_norm': 0.5409224629402161, 'learning_rate': 9.99903378846876e-06, 'epoch': 0.32}

 11%|█         | 1407/13317 [14:29<2:01:50,  1.63it/s]
 11%|█         | 1408/13317 [14:29<2:01:49,  1.63it/s]
                                                      
{'loss': 0.2635, 'grad_norm': 0.5890946388244629, 'learning_rate': 9.999007851914813e-06, 'epoch': 0.32}

 11%|█         | 1408/13317 [14:29<2:01:49,  1.63it/s]
 11%|█         | 1409/13317 [14:30<2:01:46,  1.63it/s]
                                                      
{'loss': 0.2653, 'grad_norm': 0.5464028716087341, 'learning_rate': 9.998981571876097e-06, 'epoch': 0.32}

 11%|█         | 1409/13317 [14:30<2:01:46,  1.63it/s]
 11%|█         | 1410/13317 [14:30<2:01:46,  1.63it/s]
                                                      
{'loss': 0.345, 'grad_norm': 0.7220472693443298, 'learning_rate': 9.998954948354416e-06, 'epoch': 0.32}

 11%|█         | 1410/13317 [14:31<2:01:46,  1.63it/s]
 11%|█         | 1411/13317 [14:31<2:01:47,  1.63it/s]
                                                      
{'loss': 0.3424, 'grad_norm': 0.6431899666786194, 'learning_rate': 9.9989279813516e-06, 'epoch': 0.32}

 11%|█         | 1411/13317 [14:31<2:01:47,  1.63it/s]
 11%|█         | 1412/13317 [14:32<2:01:43,  1.63it/s]
                                                      
{'loss': 0.3396, 'grad_norm': 0.7436617016792297, 'learning_rate': 9.998900670869505e-06, 'epoch': 0.32}

 11%|█         | 1412/13317 [14:32<2:01:43,  1.63it/s]
 11%|█         | 1413/13317 [14:32<2:01:45,  1.63it/s]
                                                      
{'loss': 0.3342, 'grad_norm': 0.7094032168388367, 'learning_rate': 9.998873016910003e-06, 'epoch': 0.32}

 11%|█         | 1413/13317 [14:32<2:01:45,  1.63it/s]
 11%|█         | 1414/13317 [14:33<2:01:47,  1.63it/s]
                                                      
{'loss': 0.3585, 'grad_norm': 0.687957227230072, 'learning_rate': 9.998845019474995e-06, 'epoch': 0.32}

 11%|█         | 1414/13317 [14:33<2:01:47,  1.63it/s]
 11%|█         | 1415/13317 [14:34<2:01:45,  1.63it/s]
                                                      
{'loss': 0.2883, 'grad_norm': 0.6159951686859131, 'learning_rate': 9.998816678566405e-06, 'epoch': 0.32}

 11%|█         | 1415/13317 [14:34<2:01:45,  1.63it/s]
 11%|█         | 1416/13317 [14:34<2:01:53,  1.63it/s]
                                                      
{'loss': 0.3631, 'grad_norm': 0.6724090576171875, 'learning_rate': 9.998787994186183e-06, 'epoch': 0.32}

 11%|█         | 1416/13317 [14:34<2:01:53,  1.63it/s]
 11%|█         | 1417/13317 [14:35<2:01:52,  1.63it/s]
                                                      
{'loss': 0.3353, 'grad_norm': 0.6510727405548096, 'learning_rate': 9.998758966336296e-06, 'epoch': 0.32}

 11%|█         | 1417/13317 [14:35<2:01:52,  1.63it/s]
 11%|█         | 1418/13317 [14:35<2:01:52,  1.63it/s]
                                                      
{'loss': 0.3381, 'grad_norm': 0.7782011032104492, 'learning_rate': 9.998729595018742e-06, 'epoch': 0.32}

 11%|█         | 1418/13317 [14:35<2:01:52,  1.63it/s]
 11%|█         | 1419/13317 [14:36<2:01:43,  1.63it/s]
                                                      
{'loss': 0.2974, 'grad_norm': 0.6187508702278137, 'learning_rate': 9.998699880235536e-06, 'epoch': 0.32}

 11%|█         | 1419/13317 [14:36<2:01:43,  1.63it/s]
 11%|█         | 1420/13317 [14:37<2:01:46,  1.63it/s]
                                                      
{'loss': 0.254, 'grad_norm': 0.5767797827720642, 'learning_rate': 9.99866982198872e-06, 'epoch': 0.32}

 11%|█         | 1420/13317 [14:37<2:01:46,  1.63it/s]
 11%|█         | 1421/13317 [14:37<2:01:47,  1.63it/s]
                                                      
{'loss': 0.3713, 'grad_norm': 0.6201473474502563, 'learning_rate': 9.998639420280363e-06, 'epoch': 0.32}

 11%|█         | 1421/13317 [14:37<2:01:47,  1.63it/s]
 11%|█         | 1422/13317 [14:38<2:01:41,  1.63it/s]
                                                      
{'loss': 0.3034, 'grad_norm': 0.6196393966674805, 'learning_rate': 9.998608675112552e-06, 'epoch': 0.32}

 11%|█         | 1422/13317 [14:38<2:01:41,  1.63it/s]
 11%|█         | 1423/13317 [14:38<2:01:41,  1.63it/s]
                                                      
{'loss': 0.3217, 'grad_norm': 0.6751210689544678, 'learning_rate': 9.998577586487396e-06, 'epoch': 0.32}

 11%|█         | 1423/13317 [14:39<2:01:41,  1.63it/s]
 11%|█         | 1424/13317 [14:39<2:01:52,  1.63it/s]
                                                      
{'loss': 0.2724, 'grad_norm': 0.6401410698890686, 'learning_rate': 9.998546154407035e-06, 'epoch': 0.32}

 11%|█         | 1424/13317 [14:39<2:01:52,  1.63it/s]
 11%|█         | 1425/13317 [14:40<2:01:50,  1.63it/s]
                                                      
{'loss': 0.3589, 'grad_norm': 0.8948376178741455, 'learning_rate': 9.99851437887363e-06, 'epoch': 0.32}

 11%|█         | 1425/13317 [14:40<2:01:50,  1.63it/s]
 11%|█         | 1426/13317 [14:40<2:01:46,  1.63it/s]
                                                      
{'loss': 0.3297, 'grad_norm': 0.6426302790641785, 'learning_rate': 9.99848225988936e-06, 'epoch': 0.32}

 11%|█         | 1426/13317 [14:40<2:01:46,  1.63it/s]
 11%|█         | 1427/13317 [14:41<2:01:45,  1.63it/s]
                                                      
{'loss': 0.3156, 'grad_norm': 0.6760461330413818, 'learning_rate': 9.998449797456434e-06, 'epoch': 0.32}

 11%|█         | 1427/13317 [14:41<2:01:45,  1.63it/s]
 11%|█         | 1428/13317 [14:41<2:01:44,  1.63it/s]
                                                      
{'loss': 0.3568, 'grad_norm': 0.6659609079360962, 'learning_rate': 9.998416991577083e-06, 'epoch': 0.32}

 11%|█         | 1428/13317 [14:42<2:01:44,  1.63it/s]
 11%|█         | 1429/13317 [14:42<2:01:47,  1.63it/s]
                                                      
{'loss': 0.2914, 'grad_norm': 0.633936882019043, 'learning_rate': 9.998383842253558e-06, 'epoch': 0.32}

 11%|█         | 1429/13317 [14:42<2:01:47,  1.63it/s]
 11%|█         | 1430/13317 [14:43<2:01:44,  1.63it/s]
                                                      
{'loss': 0.2559, 'grad_norm': 0.6044177412986755, 'learning_rate': 9.998350349488143e-06, 'epoch': 0.32}

 11%|█         | 1430/13317 [14:43<2:01:44,  1.63it/s]
 11%|█         | 1431/13317 [14:43<2:01:53,  1.63it/s]
                                                      
{'loss': 0.2681, 'grad_norm': 0.5776234865188599, 'learning_rate': 9.998316513283132e-06, 'epoch': 0.32}

 11%|█         | 1431/13317 [14:43<2:01:53,  1.63it/s]
 11%|█         | 1432/13317 [14:44<2:01:48,  1.63it/s]
                                                      
{'loss': 0.2932, 'grad_norm': 0.6354990601539612, 'learning_rate': 9.998282333640856e-06, 'epoch': 0.32}

 11%|█         | 1432/13317 [14:44<2:01:48,  1.63it/s]
 11%|█         | 1433/13317 [14:45<2:01:45,  1.63it/s]
                                                      
{'loss': 0.2977, 'grad_norm': 0.7354450225830078, 'learning_rate': 9.998247810563661e-06, 'epoch': 0.32}

 11%|█         | 1433/13317 [14:45<2:01:45,  1.63it/s]
 11%|█         | 1434/13317 [14:45<2:01:51,  1.63it/s]
                                                      
{'loss': 0.2659, 'grad_norm': 0.5636535882949829, 'learning_rate': 9.99821294405392e-06, 'epoch': 0.32}

 11%|█         | 1434/13317 [14:45<2:01:51,  1.63it/s]
 11%|█         | 1435/13317 [14:46<2:01:46,  1.63it/s]
                                                      
{'loss': 0.2688, 'grad_norm': 0.6051608324050903, 'learning_rate': 9.998177734114027e-06, 'epoch': 0.32}

 11%|█         | 1435/13317 [14:46<2:01:46,  1.63it/s]
 11%|█         | 1436/13317 [14:46<2:01:43,  1.63it/s]
                                                      
{'loss': 0.2887, 'grad_norm': 0.6052917242050171, 'learning_rate': 9.9981421807464e-06, 'epoch': 0.32}

 11%|█         | 1436/13317 [14:47<2:01:43,  1.63it/s]
 11%|█         | 1437/13317 [14:47<2:01:38,  1.63it/s]
                                                      
{'loss': 0.3391, 'grad_norm': 0.5930180549621582, 'learning_rate': 9.998106283953486e-06, 'epoch': 0.32}

 11%|█         | 1437/13317 [14:47<2:01:38,  1.63it/s]
 11%|█         | 1438/13317 [14:48<2:01:36,  1.63it/s]
                                                      
{'loss': 0.3169, 'grad_norm': 0.5718937516212463, 'learning_rate': 9.99807004373775e-06, 'epoch': 0.32}

 11%|█         | 1438/13317 [14:48<2:01:36,  1.63it/s]
 11%|█         | 1439/13317 [14:48<2:02:12,  1.62it/s]
                                                      
{'loss': 0.4487, 'grad_norm': 0.6339423656463623, 'learning_rate': 9.99803346010168e-06, 'epoch': 0.32}

 11%|█         | 1439/13317 [14:48<2:02:12,  1.62it/s]
 11%|█         | 1440/13317 [14:49<2:01:57,  1.62it/s]
                                                      
{'loss': 0.2877, 'grad_norm': 0.6666238903999329, 'learning_rate': 9.997996533047792e-06, 'epoch': 0.32}

 11%|█         | 1440/13317 [14:49<2:01:57,  1.62it/s]
 11%|█         | 1441/13317 [14:49<2:01:46,  1.63it/s]
                                                      
{'loss': 0.3495, 'grad_norm': 0.7341766357421875, 'learning_rate': 9.997959262578622e-06, 'epoch': 0.32}

 11%|█         | 1441/13317 [14:50<2:01:46,  1.63it/s]
 11%|█         | 1442/13317 [14:50<2:01:40,  1.63it/s]
                                                      
{'loss': 0.3048, 'grad_norm': 0.630551278591156, 'learning_rate': 9.997921648696734e-06, 'epoch': 0.32}

 11%|█         | 1442/13317 [14:50<2:01:40,  1.63it/s]
 11%|█         | 1443/13317 [14:51<2:01:37,  1.63it/s]
                                                      
{'loss': 0.2838, 'grad_norm': 0.5845223665237427, 'learning_rate': 9.997883691404707e-06, 'epoch': 0.33}

 11%|█         | 1443/13317 [14:51<2:01:37,  1.63it/s]
 11%|█         | 1444/13317 [14:51<2:01:39,  1.63it/s]
                                                      
{'loss': 0.2776, 'grad_norm': 0.5855510234832764, 'learning_rate': 9.997845390705152e-06, 'epoch': 0.33}

 11%|█         | 1444/13317 [14:51<2:01:39,  1.63it/s]
 11%|█         | 1445/13317 [14:52<2:01:34,  1.63it/s]
                                                      
{'loss': 0.3138, 'grad_norm': 0.6643127799034119, 'learning_rate': 9.997806746600702e-06, 'epoch': 0.33}

 11%|█         | 1445/13317 [14:52<2:01:34,  1.63it/s]
 11%|█         | 1446/13317 [14:53<2:01:30,  1.63it/s]
                                                      
{'loss': 0.2509, 'grad_norm': 0.5796387791633606, 'learning_rate': 9.99776775909401e-06, 'epoch': 0.33}

 11%|█         | 1446/13317 [14:53<2:01:30,  1.63it/s]
 11%|█         | 1447/13317 [14:53<2:01:26,  1.63it/s]
                                                      
{'loss': 0.304, 'grad_norm': 0.6029394268989563, 'learning_rate': 9.997728428187756e-06, 'epoch': 0.33}

 11%|█         | 1447/13317 [14:53<2:01:26,  1.63it/s]
 11%|█         | 1448/13317 [14:54<2:01:29,  1.63it/s]
                                                      
{'loss': 0.2924, 'grad_norm': 0.6056888699531555, 'learning_rate': 9.997688753884644e-06, 'epoch': 0.33}

 11%|█         | 1448/13317 [14:54<2:01:29,  1.63it/s]
 11%|█         | 1449/13317 [14:54<2:01:36,  1.63it/s]
                                                      
{'loss': 0.3587, 'grad_norm': 0.6965287923812866, 'learning_rate': 9.997648736187395e-06, 'epoch': 0.33}

 11%|█         | 1449/13317 [14:55<2:01:36,  1.63it/s]
 11%|█         | 1450/13317 [14:55<2:01:32,  1.63it/s]
                                                      
{'loss': 0.2711, 'grad_norm': 0.5699384212493896, 'learning_rate': 9.997608375098763e-06, 'epoch': 0.33}

 11%|█         | 1450/13317 [14:55<2:01:32,  1.63it/s]
 11%|█         | 1451/13317 [14:56<2:01:29,  1.63it/s]
                                                      
{'loss': 0.2796, 'grad_norm': 0.6135610938072205, 'learning_rate': 9.997567670621521e-06, 'epoch': 0.33}

 11%|█         | 1451/13317 [14:56<2:01:29,  1.63it/s]
 11%|█         | 1452/13317 [14:56<2:01:23,  1.63it/s]
                                                      
{'loss': 0.2925, 'grad_norm': 0.5946162343025208, 'learning_rate': 9.997526622758464e-06, 'epoch': 0.33}

 11%|█         | 1452/13317 [14:56<2:01:23,  1.63it/s]
 11%|█         | 1453/13317 [14:57<2:01:22,  1.63it/s]
                                                      
{'loss': 0.2974, 'grad_norm': 0.5505696535110474, 'learning_rate': 9.997485231512415e-06, 'epoch': 0.33}

 11%|█         | 1453/13317 [14:57<2:01:22,  1.63it/s]
 11%|█         | 1454/13317 [14:57<2:01:25,  1.63it/s]
                                                      
{'loss': 0.3554, 'grad_norm': 0.7110334038734436, 'learning_rate': 9.997443496886215e-06, 'epoch': 0.33}

 11%|█         | 1454/13317 [14:58<2:01:25,  1.63it/s]
 11%|█         | 1455/13317 [14:58<2:01:20,  1.63it/s]
                                                      
{'loss': 0.311, 'grad_norm': 0.6965509653091431, 'learning_rate': 9.997401418882732e-06, 'epoch': 0.33}

 11%|█         | 1455/13317 [14:58<2:01:20,  1.63it/s]
 11%|█         | 1456/13317 [14:59<2:01:24,  1.63it/s]
                                                      
{'loss': 0.2086, 'grad_norm': 0.5215455293655396, 'learning_rate': 9.997358997504861e-06, 'epoch': 0.33}

 11%|█         | 1456/13317 [14:59<2:01:24,  1.63it/s]
 11%|█         | 1457/13317 [14:59<2:01:21,  1.63it/s]
                                                      
{'loss': 0.3422, 'grad_norm': 0.7012002468109131, 'learning_rate': 9.997316232755511e-06, 'epoch': 0.33}

 11%|█         | 1457/13317 [14:59<2:01:21,  1.63it/s]
 11%|█         | 1458/13317 [15:00<2:01:20,  1.63it/s]
                                                      
{'loss': 0.3118, 'grad_norm': 0.6375614404678345, 'learning_rate': 9.997273124637625e-06, 'epoch': 0.33}

 11%|█         | 1458/13317 [15:00<2:01:20,  1.63it/s]
 11%|█         | 1459/13317 [15:01<2:01:28,  1.63it/s]
                                                      
{'loss': 0.2434, 'grad_norm': 0.5985931754112244, 'learning_rate': 9.997229673154164e-06, 'epoch': 0.33}

 11%|█         | 1459/13317 [15:01<2:01:28,  1.63it/s]
 11%|█         | 1460/13317 [15:01<2:01:25,  1.63it/s]
                                                      
{'loss': 0.2763, 'grad_norm': 0.7120979428291321, 'learning_rate': 9.997185878308111e-06, 'epoch': 0.33}

 11%|█         | 1460/13317 [15:01<2:01:25,  1.63it/s]
 11%|█         | 1461/13317 [15:02<2:01:28,  1.63it/s]
                                                      
{'loss': 0.3263, 'grad_norm': 0.7470468282699585, 'learning_rate': 9.997141740102478e-06, 'epoch': 0.33}

 11%|█         | 1461/13317 [15:02<2:01:28,  1.63it/s]
 11%|█         | 1462/13317 [15:02<2:01:20,  1.63it/s]
                                                      
{'loss': 0.3233, 'grad_norm': 0.6848887801170349, 'learning_rate': 9.997097258540298e-06, 'epoch': 0.33}

 11%|█         | 1462/13317 [15:03<2:01:20,  1.63it/s]
 11%|█         | 1463/13317 [15:03<2:01:19,  1.63it/s]
                                                      
{'loss': 0.3304, 'grad_norm': 0.6101332902908325, 'learning_rate': 9.997052433624626e-06, 'epoch': 0.33}

 11%|█         | 1463/13317 [15:03<2:01:19,  1.63it/s]
 11%|█         | 1464/13317 [15:04<2:01:32,  1.63it/s]
                                                      
{'loss': 0.3438, 'grad_norm': 0.7322390079498291, 'learning_rate': 9.997007265358541e-06, 'epoch': 0.33}

 11%|█         | 1464/13317 [15:04<2:01:32,  1.63it/s]
 11%|█         | 1465/13317 [15:04<2:01:26,  1.63it/s]
                                                      
{'loss': 0.306, 'grad_norm': 0.6345277428627014, 'learning_rate': 9.996961753745147e-06, 'epoch': 0.33}

 11%|█         | 1465/13317 [15:04<2:01:26,  1.63it/s]
 11%|█         | 1466/13317 [15:05<2:01:31,  1.63it/s]
                                                      
{'loss': 0.4392, 'grad_norm': 0.6649457812309265, 'learning_rate': 9.996915898787574e-06, 'epoch': 0.33}

 11%|█         | 1466/13317 [15:05<2:01:31,  1.63it/s]
 11%|█         | 1467/13317 [15:05<2:01:26,  1.63it/s]
                                                      
{'loss': 0.3436, 'grad_norm': 0.672316312789917, 'learning_rate': 9.996869700488969e-06, 'epoch': 0.33}

 11%|█         | 1467/13317 [15:06<2:01:26,  1.63it/s]
 11%|█         | 1468/13317 [15:06<2:01:24,  1.63it/s]
                                                      
{'loss': 0.3319, 'grad_norm': 0.6946281790733337, 'learning_rate': 9.996823158852508e-06, 'epoch': 0.33}

 11%|█         | 1468/13317 [15:06<2:01:24,  1.63it/s]
 11%|█         | 1469/13317 [15:07<2:01:32,  1.62it/s]
                                                      
{'loss': 0.3407, 'grad_norm': 0.6540680527687073, 'learning_rate': 9.99677627388139e-06, 'epoch': 0.33}

 11%|█         | 1469/13317 [15:07<2:01:32,  1.62it/s]
 11%|█         | 1470/13317 [15:07<2:01:26,  1.63it/s]
                                                      
{'loss': 0.355, 'grad_norm': 0.7385424375534058, 'learning_rate': 9.996729045578834e-06, 'epoch': 0.33}

 11%|█         | 1470/13317 [15:07<2:01:26,  1.63it/s]
 11%|█         | 1471/13317 [15:08<2:01:22,  1.63it/s]
                                                      
{'loss': 0.2548, 'grad_norm': 0.5571065545082092, 'learning_rate': 9.996681473948087e-06, 'epoch': 0.33}

 11%|█         | 1471/13317 [15:08<2:01:22,  1.63it/s]
 11%|█         | 1472/13317 [15:09<2:01:15,  1.63it/s]
                                                      
{'loss': 0.3077, 'grad_norm': 0.6397600173950195, 'learning_rate': 9.996633558992416e-06, 'epoch': 0.33}

 11%|█         | 1472/13317 [15:09<2:01:15,  1.63it/s]
 11%|█         | 1473/13317 [15:09<2:01:15,  1.63it/s]
                                                      
{'loss': 0.2719, 'grad_norm': 0.5393206477165222, 'learning_rate': 9.996585300715117e-06, 'epoch': 0.33}

 11%|█         | 1473/13317 [15:09<2:01:15,  1.63it/s]
 11%|█         | 1474/13317 [15:10<2:01:19,  1.63it/s]
                                                      
{'loss': 0.2507, 'grad_norm': 0.5624769926071167, 'learning_rate': 9.996536699119499e-06, 'epoch': 0.33}

 11%|█         | 1474/13317 [15:10<2:01:19,  1.63it/s]
 11%|█         | 1475/13317 [15:10<2:01:15,  1.63it/s]
                                                      
{'loss': 0.4013, 'grad_norm': 0.644000232219696, 'learning_rate': 9.996487754208908e-06, 'epoch': 0.33}

 11%|█         | 1475/13317 [15:11<2:01:15,  1.63it/s]
 11%|█         | 1476/13317 [15:11<2:01:48,  1.62it/s]
                                                      
{'loss': 0.3303, 'grad_norm': 0.6414066553115845, 'learning_rate': 9.996438465986704e-06, 'epoch': 0.33}

 11%|█         | 1476/13317 [15:11<2:01:48,  1.62it/s]
 11%|█         | 1477/13317 [15:12<2:01:32,  1.62it/s]
                                                      
{'loss': 0.3278, 'grad_norm': 0.5606587529182434, 'learning_rate': 9.996388834456275e-06, 'epoch': 0.33}

 11%|█         | 1477/13317 [15:12<2:01:32,  1.62it/s]
 11%|█         | 1478/13317 [15:12<2:01:31,  1.62it/s]
                                                      
{'loss': 0.2767, 'grad_norm': 0.7410769462585449, 'learning_rate': 9.996338859621031e-06, 'epoch': 0.33}

 11%|█         | 1478/13317 [15:12<2:01:31,  1.62it/s]
 11%|█         | 1479/13317 [15:13<2:01:33,  1.62it/s]
                                                      
{'loss': 0.278, 'grad_norm': 0.7049342393875122, 'learning_rate': 9.996288541484404e-06, 'epoch': 0.33}

 11%|█         | 1479/13317 [15:13<2:01:33,  1.62it/s]
 11%|█         | 1480/13317 [15:13<2:01:20,  1.63it/s]
                                                      
{'loss': 0.2841, 'grad_norm': 0.6411069631576538, 'learning_rate': 9.996237880049853e-06, 'epoch': 0.33}

 11%|█         | 1480/13317 [15:14<2:01:20,  1.63it/s]
 11%|█         | 1481/13317 [15:14<2:01:21,  1.63it/s]
                                                      
{'loss': 0.3509, 'grad_norm': 0.8276961445808411, 'learning_rate': 9.996186875320859e-06, 'epoch': 0.33}

 11%|█         | 1481/13317 [15:14<2:01:21,  1.63it/s]
 11%|█         | 1482/13317 [15:15<2:01:13,  1.63it/s]
                                                      
{'loss': 0.3477, 'grad_norm': 0.7002006769180298, 'learning_rate': 9.996135527300925e-06, 'epoch': 0.33}

 11%|█         | 1482/13317 [15:15<2:01:13,  1.63it/s]
 11%|█         | 1483/13317 [15:15<2:01:04,  1.63it/s]
                                                      
{'loss': 0.216, 'grad_norm': 0.5119858980178833, 'learning_rate': 9.996083835993581e-06, 'epoch': 0.33}

 11%|█         | 1483/13317 [15:15<2:01:04,  1.63it/s]
 11%|█         | 1484/13317 [15:16<2:01:38,  1.62it/s]
                                                      
{'loss': 0.3784, 'grad_norm': 0.6631624698638916, 'learning_rate': 9.99603180140238e-06, 'epoch': 0.33}

 11%|█         | 1484/13317 [15:16<2:01:38,  1.62it/s]
 11%|█         | 1485/13317 [15:17<2:01:28,  1.62it/s]
                                                      
{'loss': 0.2563, 'grad_norm': 0.5848702192306519, 'learning_rate': 9.995979423530893e-06, 'epoch': 0.33}

 11%|█         | 1485/13317 [15:17<2:01:28,  1.62it/s]
 11%|█         | 1486/13317 [15:17<2:01:19,  1.63it/s]
                                                      
{'loss': 0.3804, 'grad_norm': 0.7974822521209717, 'learning_rate': 9.995926702382722e-06, 'epoch': 0.33}

 11%|█         | 1486/13317 [15:17<2:01:19,  1.63it/s]
 11%|█         | 1487/13317 [15:18<2:01:09,  1.63it/s]
                                                      
{'loss': 0.3091, 'grad_norm': 0.6263759732246399, 'learning_rate': 9.995873637961489e-06, 'epoch': 0.33}

 11%|█         | 1487/13317 [15:18<2:01:09,  1.63it/s]
 11%|█         | 1488/13317 [15:18<2:01:04,  1.63it/s]
                                                      
{'loss': 0.2697, 'grad_norm': 0.7397312521934509, 'learning_rate': 9.99582023027084e-06, 'epoch': 0.34}

 11%|█         | 1488/13317 [15:19<2:01:04,  1.63it/s]
 11%|█         | 1489/13317 [15:19<2:01:38,  1.62it/s]
                                                      
{'loss': 0.2338, 'grad_norm': 0.5309924483299255, 'learning_rate': 9.995766479314444e-06, 'epoch': 0.34}

 11%|█         | 1489/13317 [15:19<2:01:38,  1.62it/s]
 11%|█         | 1490/13317 [15:20<2:01:24,  1.62it/s]
                                                      
{'loss': 0.2598, 'grad_norm': 0.6475595235824585, 'learning_rate': 9.995712385095995e-06, 'epoch': 0.34}

 11%|█         | 1490/13317 [15:20<2:01:24,  1.62it/s]
 11%|█         | 1491/13317 [15:20<2:01:16,  1.63it/s]
                                                      
{'loss': 0.3059, 'grad_norm': 0.6221978664398193, 'learning_rate': 9.99565794761921e-06, 'epoch': 0.34}

 11%|█         | 1491/13317 [15:20<2:01:16,  1.63it/s]
 11%|█         | 1492/13317 [15:21<2:01:12,  1.63it/s]
                                                      
{'loss': 0.2597, 'grad_norm': 0.600888192653656, 'learning_rate': 9.995603166887828e-06, 'epoch': 0.34}

 11%|█         | 1492/13317 [15:21<2:01:12,  1.63it/s]
 11%|█         | 1493/13317 [15:21<2:01:01,  1.63it/s]
                                                      
{'loss': 0.4289, 'grad_norm': 0.7508870959281921, 'learning_rate': 9.995548042905616e-06, 'epoch': 0.34}

 11%|█         | 1493/13317 [15:22<2:01:01,  1.63it/s]
 11%|█         | 1494/13317 [15:22<2:01:11,  1.63it/s]
                                                      
{'loss': 0.2938, 'grad_norm': 0.5698884725570679, 'learning_rate': 9.99549257567636e-06, 'epoch': 0.34}

 11%|█         | 1494/13317 [15:22<2:01:11,  1.63it/s]
 11%|█         | 1495/13317 [15:23<2:01:05,  1.63it/s]
                                                      
{'loss': 0.2873, 'grad_norm': 0.6235120892524719, 'learning_rate': 9.99543676520387e-06, 'epoch': 0.34}

 11%|█         | 1495/13317 [15:23<2:01:05,  1.63it/s]
 11%|█         | 1496/13317 [15:23<2:01:30,  1.62it/s]
                                                      
{'loss': 0.249, 'grad_norm': 0.5433816909790039, 'learning_rate': 9.99538061149198e-06, 'epoch': 0.34}

 11%|█         | 1496/13317 [15:23<2:01:30,  1.62it/s]
 11%|█         | 1497/13317 [15:24<2:01:18,  1.62it/s]
                                                      
{'loss': 0.3314, 'grad_norm': 0.683914840221405, 'learning_rate': 9.995324114544553e-06, 'epoch': 0.34}

 11%|█         | 1497/13317 [15:24<2:01:18,  1.62it/s]
 11%|█         | 1498/13317 [15:25<2:01:12,  1.63it/s]
                                                      
{'loss': 0.2652, 'grad_norm': 0.7190077900886536, 'learning_rate': 9.995267274365466e-06, 'epoch': 0.34}

 11%|█         | 1498/13317 [15:25<2:01:12,  1.63it/s]
 11%|█▏        | 1499/13317 [15:25<2:01:21,  1.62it/s]
                                                      
{'loss': 0.2683, 'grad_norm': 0.6031792163848877, 'learning_rate': 9.995210090958627e-06, 'epoch': 0.34}

 11%|█▏        | 1499/13317 [15:25<2:01:21,  1.62it/s]
 11%|█▏        | 1500/13317 [15:26<2:01:12,  1.62it/s]
                                                      
{'loss': 0.2948, 'grad_norm': 0.6769421696662903, 'learning_rate': 9.995152564327965e-06, 'epoch': 0.34}

 11%|█▏        | 1500/13317 [15:26<2:01:12,  1.62it/s]

  0%|          | 0/13 [00:00<?, ?it/s]

 15%|█▌        | 2/13 [00:00<00:00, 13.11it/s]

 31%|███       | 4/13 [00:00<00:01,  8.36it/s]

 38%|███▊      | 5/13 [00:00<00:01,  7.78it/s]

 46%|████▌     | 6/13 [00:00<00:00,  7.40it/s]

 54%|█████▍    | 7/13 [00:00<00:00,  7.17it/s]

 62%|██████▏   | 8/13 [00:01<00:00,  7.01it/s]

 69%|██████▉   | 9/13 [00:01<00:00,  6.91it/s]

 77%|███████▋  | 10/13 [00:01<00:00,  6.83it/s]

 85%|████████▍ | 11/13 [00:01<00:00,  6.78it/s]

 92%|█████████▏| 12/13 [00:01<00:00,  6.75it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]
                                                      

                                               
{'eval_loss': 0.7640513777732849, 'eval_runtime': 1.9553, 'eval_samples_per_second': 102.288, 'eval_steps_per_second': 6.649, 'epoch': 0.34}

 11%|█▏        | 1500/13317 [15:28<2:01:12,  1.62it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]

                                               
 11%|█▏        | 1501/13317 [15:28<3:57:01,  1.20s/it]
                                                      
{'loss': 0.3264, 'grad_norm': 0.6102054715156555, 'learning_rate': 9.995094694477432e-06, 'epoch': 0.34}

 11%|█▏        | 1501/13317 [15:28<3:57:01,  1.20s/it]
 11%|█▏        | 1502/13317 [15:29<3:22:09,  1.03s/it]
                                                      
{'loss': 0.2348, 'grad_norm': 0.548909068107605, 'learning_rate': 9.995036481411005e-06, 'epoch': 0.34}

 11%|█▏        | 1502/13317 [15:29<3:22:09,  1.03s/it]
 11%|█▏        | 1503/13317 [15:30<2:57:44,  1.11it/s]
                                                      
{'loss': 0.2535, 'grad_norm': 0.5559327006340027, 'learning_rate': 9.994977925132682e-06, 'epoch': 0.34}

 11%|█▏        | 1503/13317 [15:30<2:57:44,  1.11it/s]
 11%|█▏        | 1504/13317 [15:30<2:40:50,  1.22it/s]
                                                      
{'loss': 0.3242, 'grad_norm': 0.7409523725509644, 'learning_rate': 9.994919025646489e-06, 'epoch': 0.34}

 11%|█▏        | 1504/13317 [15:30<2:40:50,  1.22it/s]
 11%|█▏        | 1505/13317 [15:31<2:28:48,  1.32it/s]
                                                      
{'loss': 0.3323, 'grad_norm': 0.6529146432876587, 'learning_rate': 9.994859782956473e-06, 'epoch': 0.34}

 11%|█▏        | 1505/13317 [15:31<2:28:48,  1.32it/s]
 11%|█▏        | 1506/13317 [15:31<2:20:24,  1.40it/s]
                                                      
{'loss': 0.315, 'grad_norm': 0.609963595867157, 'learning_rate': 9.9948001970667e-06, 'epoch': 0.34}

 11%|█▏        | 1506/13317 [15:32<2:20:24,  1.40it/s]
 11%|█▏        | 1507/13317 [15:32<2:14:34,  1.46it/s]
                                                      
{'loss': 0.3073, 'grad_norm': 0.6901487708091736, 'learning_rate': 9.99474026798127e-06, 'epoch': 0.34}

 11%|█▏        | 1507/13317 [15:32<2:14:34,  1.46it/s]
 11%|█▏        | 1508/13317 [15:33<2:10:25,  1.51it/s]
                                                      
{'loss': 0.2801, 'grad_norm': 0.626196026802063, 'learning_rate': 9.9946799957043e-06, 'epoch': 0.34}

 11%|█▏        | 1508/13317 [15:33<2:10:25,  1.51it/s]
 11%|█▏        | 1509/13317 [15:33<2:07:38,  1.54it/s]
                                                      
{'loss': 0.2918, 'grad_norm': 0.7749810218811035, 'learning_rate': 9.994619380239927e-06, 'epoch': 0.34}

 11%|█▏        | 1509/13317 [15:33<2:07:38,  1.54it/s]
 11%|█▏        | 1510/13317 [15:34<2:05:40,  1.57it/s]
                                                      
{'loss': 0.2563, 'grad_norm': 0.6258547902107239, 'learning_rate': 9.99455842159232e-06, 'epoch': 0.34}

 11%|█▏        | 1510/13317 [15:34<2:05:40,  1.57it/s]
 11%|█▏        | 1511/13317 [15:34<2:04:16,  1.58it/s]
                                                      
{'loss': 0.2789, 'grad_norm': 0.5880772471427917, 'learning_rate': 9.994497119765667e-06, 'epoch': 0.34}

 11%|█▏        | 1511/13317 [15:35<2:04:16,  1.58it/s]
 11%|█▏        | 1512/13317 [15:35<2:03:09,  1.60it/s]
                                                      
{'loss': 0.3892, 'grad_norm': 0.7518519759178162, 'learning_rate': 9.994435474764179e-06, 'epoch': 0.34}

 11%|█▏        | 1512/13317 [15:35<2:03:09,  1.60it/s]
 11%|█▏        | 1513/13317 [15:36<2:02:22,  1.61it/s]
                                                      
{'loss': 0.2824, 'grad_norm': 0.6467341780662537, 'learning_rate': 9.99437348659209e-06, 'epoch': 0.34}

 11%|█▏        | 1513/13317 [15:36<2:02:22,  1.61it/s]
 11%|█▏        | 1514/13317 [15:36<2:02:05,  1.61it/s]
                                                      
{'loss': 0.3352, 'grad_norm': 0.6568017601966858, 'learning_rate': 9.994311155253662e-06, 'epoch': 0.34}

 11%|█▏        | 1514/13317 [15:36<2:02:05,  1.61it/s]
 11%|█▏        | 1515/13317 [15:37<2:01:39,  1.62it/s]
                                                      
{'loss': 0.3522, 'grad_norm': 0.7289713025093079, 'learning_rate': 9.994248480753178e-06, 'epoch': 0.34}

 11%|█▏        | 1515/13317 [15:37<2:01:39,  1.62it/s]
 11%|█▏        | 1516/13317 [15:38<2:01:21,  1.62it/s]
                                                      
{'loss': 0.2454, 'grad_norm': 0.6278430223464966, 'learning_rate': 9.994185463094943e-06, 'epoch': 0.34}

 11%|█▏        | 1516/13317 [15:38<2:01:21,  1.62it/s]
 11%|█▏        | 1517/13317 [15:38<2:01:09,  1.62it/s]
                                                      
{'loss': 0.2272, 'grad_norm': 0.6283231973648071, 'learning_rate': 9.994122102283287e-06, 'epoch': 0.34}

 11%|█▏        | 1517/13317 [15:38<2:01:09,  1.62it/s]
 11%|█▏        | 1518/13317 [15:39<2:01:00,  1.63it/s]
                                                      
{'loss': 0.2817, 'grad_norm': 0.6913866996765137, 'learning_rate': 9.994058398322564e-06, 'epoch': 0.34}

 11%|█▏        | 1518/13317 [15:39<2:01:00,  1.63it/s]
 11%|█▏        | 1519/13317 [15:39<2:01:04,  1.62it/s]
                                                      
{'loss': 0.2839, 'grad_norm': 0.6609846949577332, 'learning_rate': 9.993994351217153e-06, 'epoch': 0.34}

 11%|█▏        | 1519/13317 [15:40<2:01:04,  1.62it/s]
 11%|█▏        | 1520/13317 [15:40<2:00:54,  1.63it/s]
                                                      
{'loss': 0.3092, 'grad_norm': 0.6113572120666504, 'learning_rate': 9.99392996097145e-06, 'epoch': 0.34}

 11%|█▏        | 1520/13317 [15:40<2:00:54,  1.63it/s]
 11%|█▏        | 1521/13317 [15:41<2:00:47,  1.63it/s]
                                                      
{'loss': 0.253, 'grad_norm': 0.5329825282096863, 'learning_rate': 9.993865227589884e-06, 'epoch': 0.34}

 11%|█▏        | 1521/13317 [15:41<2:00:47,  1.63it/s]
 11%|█▏        | 1522/13317 [15:41<2:00:43,  1.63it/s]
                                                      
{'loss': 0.3114, 'grad_norm': 0.6287066340446472, 'learning_rate': 9.9938001510769e-06, 'epoch': 0.34}

 11%|█▏        | 1522/13317 [15:41<2:00:43,  1.63it/s]
 11%|█▏        | 1523/13317 [15:42<2:00:40,  1.63it/s]
                                                      
{'loss': 0.3556, 'grad_norm': 0.5819591283798218, 'learning_rate': 9.99373473143697e-06, 'epoch': 0.34}

 11%|█▏        | 1523/13317 [15:42<2:00:40,  1.63it/s]
 11%|█▏        | 1524/13317 [15:42<2:00:44,  1.63it/s]
                                                      
{'loss': 0.2872, 'grad_norm': 0.6612334251403809, 'learning_rate': 9.99366896867459e-06, 'epoch': 0.34}

 11%|█▏        | 1524/13317 [15:43<2:00:44,  1.63it/s]
 11%|█▏        | 1525/13317 [15:43<2:00:45,  1.63it/s]
                                                      
{'loss': 0.4155, 'grad_norm': 0.7622774243354797, 'learning_rate': 9.99360286279428e-06, 'epoch': 0.34}

 11%|█▏        | 1525/13317 [15:43<2:00:45,  1.63it/s]
 11%|█▏        | 1526/13317 [15:44<2:00:40,  1.63it/s]
                                                      
{'loss': 0.2984, 'grad_norm': 0.6436214447021484, 'learning_rate': 9.993536413800577e-06, 'epoch': 0.34}

 11%|█▏        | 1526/13317 [15:44<2:00:40,  1.63it/s]
 11%|█▏        | 1527/13317 [15:44<2:00:38,  1.63it/s]
                                                      
{'loss': 0.3542, 'grad_norm': 0.7113814353942871, 'learning_rate': 9.993469621698052e-06, 'epoch': 0.34}

 11%|█▏        | 1527/13317 [15:44<2:00:38,  1.63it/s]
 11%|█▏        | 1528/13317 [15:45<2:00:34,  1.63it/s]
                                                      
{'loss': 0.248, 'grad_norm': 0.5704458951950073, 'learning_rate': 9.993402486491292e-06, 'epoch': 0.34}

 11%|█▏        | 1528/13317 [15:45<2:00:34,  1.63it/s]
 11%|█▏        | 1529/13317 [15:46<2:00:46,  1.63it/s]
                                                      
{'loss': 0.2726, 'grad_norm': 0.6315699815750122, 'learning_rate': 9.99333500818491e-06, 'epoch': 0.34}

 11%|█▏        | 1529/13317 [15:46<2:00:46,  1.63it/s]
 11%|█▏        | 1530/13317 [15:46<2:00:39,  1.63it/s]
                                                      
{'loss': 0.2781, 'grad_norm': 0.6749526858329773, 'learning_rate': 9.993267186783543e-06, 'epoch': 0.34}

 11%|█▏        | 1530/13317 [15:46<2:00:39,  1.63it/s]
 11%|█▏        | 1531/13317 [15:47<2:00:35,  1.63it/s]
                                                      
{'loss': 0.3017, 'grad_norm': 0.6096208691596985, 'learning_rate': 9.993199022291849e-06, 'epoch': 0.34}

 11%|█▏        | 1531/13317 [15:47<2:00:35,  1.63it/s]
 12%|█▏        | 1532/13317 [15:47<2:00:35,  1.63it/s]
                                                      
{'loss': 0.2706, 'grad_norm': 0.5406699776649475, 'learning_rate': 9.993130514714515e-06, 'epoch': 0.35}

 12%|█▏        | 1532/13317 [15:48<2:00:35,  1.63it/s]
 12%|█▏        | 1533/13317 [15:48<2:00:31,  1.63it/s]
                                                      
{'loss': 0.2562, 'grad_norm': 0.5260266065597534, 'learning_rate': 9.993061664056245e-06, 'epoch': 0.35}

 12%|█▏        | 1533/13317 [15:48<2:00:31,  1.63it/s]
 12%|█▏        | 1534/13317 [15:49<2:00:38,  1.63it/s]
                                                      
{'loss': 0.3179, 'grad_norm': 0.6310276985168457, 'learning_rate': 9.992992470321773e-06, 'epoch': 0.35}

 12%|█▏        | 1534/13317 [15:49<2:00:38,  1.63it/s]
 12%|█▏        | 1535/13317 [15:49<2:00:35,  1.63it/s]
                                                      
{'loss': 0.4044, 'grad_norm': 0.7050585746765137, 'learning_rate': 9.992922933515852e-06, 'epoch': 0.35}

 12%|█▏        | 1535/13317 [15:49<2:00:35,  1.63it/s]
 12%|█▏        | 1536/13317 [15:50<2:00:28,  1.63it/s]
                                                      
{'loss': 0.3293, 'grad_norm': 0.7053653597831726, 'learning_rate': 9.992853053643257e-06, 'epoch': 0.35}

 12%|█▏        | 1536/13317 [15:50<2:00:28,  1.63it/s]
 12%|█▏        | 1537/13317 [15:50<2:00:28,  1.63it/s]
                                                      
{'loss': 0.2885, 'grad_norm': 0.6171911954879761, 'learning_rate': 9.992782830708794e-06, 'epoch': 0.35}

 12%|█▏        | 1537/13317 [15:51<2:00:28,  1.63it/s]
 12%|█▏        | 1538/13317 [15:51<2:00:29,  1.63it/s]
                                                      
{'loss': 0.2791, 'grad_norm': 0.6153753995895386, 'learning_rate': 9.992712264717286e-06, 'epoch': 0.35}

 12%|█▏        | 1538/13317 [15:51<2:00:29,  1.63it/s]
 12%|█▏        | 1539/13317 [15:52<2:00:39,  1.63it/s]
                                                      
{'loss': 0.296, 'grad_norm': 0.6289207339286804, 'learning_rate': 9.992641355673582e-06, 'epoch': 0.35}

 12%|█▏        | 1539/13317 [15:52<2:00:39,  1.63it/s]
 12%|█▏        | 1540/13317 [15:52<2:00:34,  1.63it/s]
                                                      
{'loss': 0.3393, 'grad_norm': 0.6292438507080078, 'learning_rate': 9.992570103582552e-06, 'epoch': 0.35}

 12%|█▏        | 1540/13317 [15:52<2:00:34,  1.63it/s]
 12%|█▏        | 1541/13317 [15:53<2:00:30,  1.63it/s]
                                                      
{'loss': 0.2689, 'grad_norm': 0.6320315599441528, 'learning_rate': 9.992498508449095e-06, 'epoch': 0.35}

 12%|█▏        | 1541/13317 [15:53<2:00:30,  1.63it/s]
 12%|█▏        | 1542/13317 [15:54<2:00:28,  1.63it/s]
                                                      
{'loss': 0.228, 'grad_norm': 0.7045512199401855, 'learning_rate': 9.992426570278129e-06, 'epoch': 0.35}

 12%|█▏        | 1542/13317 [15:54<2:00:28,  1.63it/s]
 12%|█▏        | 1543/13317 [15:54<2:00:26,  1.63it/s]
                                                      
{'loss': 0.2911, 'grad_norm': 0.5935370922088623, 'learning_rate': 9.992354289074596e-06, 'epoch': 0.35}

 12%|█▏        | 1543/13317 [15:54<2:00:26,  1.63it/s]
 12%|█▏        | 1544/13317 [15:55<2:00:33,  1.63it/s]
                                                      
{'loss': 0.2294, 'grad_norm': 0.5735299587249756, 'learning_rate': 9.992281664843466e-06, 'epoch': 0.35}

 12%|█▏        | 1544/13317 [15:55<2:00:33,  1.63it/s]
 12%|█▏        | 1545/13317 [15:55<2:00:30,  1.63it/s]
                                                      
{'loss': 0.321, 'grad_norm': 0.67902010679245, 'learning_rate': 9.992208697589724e-06, 'epoch': 0.35}

 12%|█▏        | 1545/13317 [15:55<2:00:30,  1.63it/s]
 12%|█▏        | 1546/13317 [15:56<2:00:35,  1.63it/s]
                                                      
{'loss': 0.3352, 'grad_norm': 0.6157074570655823, 'learning_rate': 9.99213538731839e-06, 'epoch': 0.35}

 12%|█▏        | 1546/13317 [15:56<2:00:35,  1.63it/s]
 12%|█▏        | 1547/13317 [15:57<2:00:27,  1.63it/s]
                                                      
{'loss': 0.3021, 'grad_norm': 0.6876479387283325, 'learning_rate': 9.992061734034492e-06, 'epoch': 0.35}

 12%|█▏        | 1547/13317 [15:57<2:00:27,  1.63it/s]
 12%|█▏        | 1548/13317 [15:57<2:00:20,  1.63it/s]
                                                      
{'loss': 0.2961, 'grad_norm': 0.6495775580406189, 'learning_rate': 9.9919877377431e-06, 'epoch': 0.35}

 12%|█▏        | 1548/13317 [15:57<2:00:20,  1.63it/s]
 12%|█▏        | 1549/13317 [15:58<2:00:31,  1.63it/s]
                                                      
{'loss': 0.2804, 'grad_norm': 0.5913134217262268, 'learning_rate': 9.991913398449291e-06, 'epoch': 0.35}

 12%|█▏        | 1549/13317 [15:58<2:00:31,  1.63it/s]
 12%|█▏        | 1550/13317 [15:58<2:00:26,  1.63it/s]
                                                      
{'loss': 0.424, 'grad_norm': 0.8371443748474121, 'learning_rate': 9.99183871615818e-06, 'epoch': 0.35}

 12%|█▏        | 1550/13317 [15:59<2:00:26,  1.63it/s]
 12%|█▏        | 1551/13317 [15:59<2:00:32,  1.63it/s]
                                                      
{'loss': 0.3331, 'grad_norm': 0.7010399699211121, 'learning_rate': 9.991763690874894e-06, 'epoch': 0.35}

 12%|█▏        | 1551/13317 [15:59<2:00:32,  1.63it/s]
 12%|█▏        | 1552/13317 [16:00<2:00:30,  1.63it/s]
                                                      
{'loss': 0.3455, 'grad_norm': 0.6997034549713135, 'learning_rate': 9.991688322604587e-06, 'epoch': 0.35}

 12%|█▏        | 1552/13317 [16:00<2:00:30,  1.63it/s]
 12%|█▏        | 1553/13317 [16:00<2:00:37,  1.63it/s]
                                                      
{'loss': 0.2438, 'grad_norm': 0.5586917400360107, 'learning_rate': 9.991612611352439e-06, 'epoch': 0.35}

 12%|█▏        | 1553/13317 [16:00<2:00:37,  1.63it/s]
 12%|█▏        | 1554/13317 [16:01<2:00:38,  1.63it/s]
                                                      
{'loss': 0.2556, 'grad_norm': 0.6144219040870667, 'learning_rate': 9.991536557123655e-06, 'epoch': 0.35}

 12%|█▏        | 1554/13317 [16:01<2:00:38,  1.63it/s]
 12%|█▏        | 1555/13317 [16:02<2:00:29,  1.63it/s]
                                                      
{'loss': 0.3809, 'grad_norm': 0.6330212950706482, 'learning_rate': 9.991460159923456e-06, 'epoch': 0.35}

 12%|█▏        | 1555/13317 [16:02<2:00:29,  1.63it/s]
 12%|█▏        | 1556/13317 [16:02<2:00:37,  1.62it/s]
                                                      
{'loss': 0.2682, 'grad_norm': 0.6260191798210144, 'learning_rate': 9.991383419757095e-06, 'epoch': 0.35}

 12%|█▏        | 1556/13317 [16:02<2:00:37,  1.62it/s]
 12%|█▏        | 1557/13317 [16:03<2:00:29,  1.63it/s]
                                                      
{'loss': 0.2951, 'grad_norm': 0.6419384479522705, 'learning_rate': 9.991306336629843e-06, 'epoch': 0.35}

 12%|█▏        | 1557/13317 [16:03<2:00:29,  1.63it/s]
 12%|█▏        | 1558/13317 [16:03<2:00:23,  1.63it/s]
                                                      
{'loss': 0.3831, 'grad_norm': 0.7697676420211792, 'learning_rate': 9.991228910546998e-06, 'epoch': 0.35}

 12%|█▏        | 1558/13317 [16:03<2:00:23,  1.63it/s]
 12%|█▏        | 1559/13317 [16:04<2:00:24,  1.63it/s]
                                                      
{'loss': 0.2056, 'grad_norm': 0.5046180486679077, 'learning_rate': 9.991151141513877e-06, 'epoch': 0.35}

 12%|█▏        | 1559/13317 [16:04<2:00:24,  1.63it/s]
 12%|█▏        | 1560/13317 [16:05<2:00:21,  1.63it/s]
                                                      
{'loss': 0.388, 'grad_norm': 0.66161048412323, 'learning_rate': 9.991073029535827e-06, 'epoch': 0.35}

 12%|█▏        | 1560/13317 [16:05<2:00:21,  1.63it/s]
 12%|█▏        | 1561/13317 [16:05<2:00:20,  1.63it/s]
                                                      
{'loss': 0.3029, 'grad_norm': 0.6662453413009644, 'learning_rate': 9.990994574618212e-06, 'epoch': 0.35}

 12%|█▏        | 1561/13317 [16:05<2:00:20,  1.63it/s]
 12%|█▏        | 1562/13317 [16:06<2:00:14,  1.63it/s]
                                                      
{'loss': 0.3214, 'grad_norm': 0.5721290111541748, 'learning_rate': 9.990915776766425e-06, 'epoch': 0.35}

 12%|█▏        | 1562/13317 [16:06<2:00:14,  1.63it/s]
 12%|█▏        | 1563/13317 [16:06<2:00:10,  1.63it/s]
                                                      
{'loss': 0.2874, 'grad_norm': 0.5623414516448975, 'learning_rate': 9.990836635985879e-06, 'epoch': 0.35}

 12%|█▏        | 1563/13317 [16:07<2:00:10,  1.63it/s]
 12%|█▏        | 1564/13317 [16:07<2:00:20,  1.63it/s]
                                                      
{'loss': 0.3039, 'grad_norm': 0.6481313109397888, 'learning_rate': 9.990757152282014e-06, 'epoch': 0.35}

 12%|█▏        | 1564/13317 [16:07<2:00:20,  1.63it/s]
 12%|█▏        | 1565/13317 [16:08<2:00:17,  1.63it/s]
                                                      
{'loss': 0.2432, 'grad_norm': 0.5227460861206055, 'learning_rate': 9.990677325660288e-06, 'epoch': 0.35}

 12%|█▏        | 1565/13317 [16:08<2:00:17,  1.63it/s]
 12%|█▏        | 1566/13317 [16:08<2:00:23,  1.63it/s]
                                                      
{'loss': 0.3447, 'grad_norm': 0.6086606383323669, 'learning_rate': 9.990597156126188e-06, 'epoch': 0.35}

 12%|█▏        | 1566/13317 [16:08<2:00:23,  1.63it/s]
 12%|█▏        | 1567/13317 [16:09<2:00:15,  1.63it/s]
                                                      
{'loss': 0.3003, 'grad_norm': 0.6240401864051819, 'learning_rate': 9.990516643685222e-06, 'epoch': 0.35}

 12%|█▏        | 1567/13317 [16:09<2:00:15,  1.63it/s]
 12%|█▏        | 1568/13317 [16:09<2:00:12,  1.63it/s]
                                                      
{'loss': 0.2777, 'grad_norm': 0.5624856948852539, 'learning_rate': 9.990435788342921e-06, 'epoch': 0.35}

 12%|█▏        | 1568/13317 [16:10<2:00:12,  1.63it/s]
 12%|█▏        | 1569/13317 [16:10<2:00:15,  1.63it/s]
                                                      
{'loss': 0.3486, 'grad_norm': 0.703545093536377, 'learning_rate': 9.990354590104843e-06, 'epoch': 0.35}

 12%|█▏        | 1569/13317 [16:10<2:00:15,  1.63it/s]
 12%|█▏        | 1570/13317 [16:11<2:00:13,  1.63it/s]
                                                      
{'loss': 0.2794, 'grad_norm': 0.5896840691566467, 'learning_rate': 9.990273048976566e-06, 'epoch': 0.35}

 12%|█▏        | 1570/13317 [16:11<2:00:13,  1.63it/s]
 12%|█▏        | 1571/13317 [16:11<2:00:07,  1.63it/s]
                                                      
{'loss': 0.28, 'grad_norm': 0.6038385629653931, 'learning_rate': 9.990191164963693e-06, 'epoch': 0.35}

 12%|█▏        | 1571/13317 [16:11<2:00:07,  1.63it/s]
 12%|█▏        | 1572/13317 [16:12<2:00:08,  1.63it/s]
                                                      
{'loss': 0.2525, 'grad_norm': 0.5647400617599487, 'learning_rate': 9.99010893807185e-06, 'epoch': 0.35}

 12%|█▏        | 1572/13317 [16:12<2:00:08,  1.63it/s]
 12%|█▏        | 1573/13317 [16:13<2:00:06,  1.63it/s]
                                                      
{'loss': 0.2571, 'grad_norm': 0.7163064479827881, 'learning_rate': 9.990026368306685e-06, 'epoch': 0.35}

 12%|█▏        | 1573/13317 [16:13<2:00:06,  1.63it/s]
 12%|█▏        | 1574/13317 [16:13<2:00:14,  1.63it/s]
                                                      
{'loss': 0.2961, 'grad_norm': 0.8108536601066589, 'learning_rate': 9.989943455673876e-06, 'epoch': 0.35}

 12%|█▏        | 1574/13317 [16:13<2:00:14,  1.63it/s]
 12%|█▏        | 1575/13317 [16:14<2:00:11,  1.63it/s]
                                                      
{'loss': 0.269, 'grad_norm': 0.5227934718132019, 'learning_rate': 9.989860200179116e-06, 'epoch': 0.35}

 12%|█▏        | 1575/13317 [16:14<2:00:11,  1.63it/s]
 12%|█▏        | 1576/13317 [16:14<2:00:09,  1.63it/s]
                                                      
{'loss': 0.4574, 'grad_norm': 0.7918188571929932, 'learning_rate': 9.989776601828127e-06, 'epoch': 0.36}

 12%|█▏        | 1576/13317 [16:15<2:00:09,  1.63it/s]
 12%|█▏        | 1577/13317 [16:15<2:00:06,  1.63it/s]
                                                      
{'loss': 0.3034, 'grad_norm': 0.6621447801589966, 'learning_rate': 9.98969266062665e-06, 'epoch': 0.36}

 12%|█▏        | 1577/13317 [16:15<2:00:06,  1.63it/s]
 12%|█▏        | 1578/13317 [16:16<2:00:04,  1.63it/s]
                                                      
{'loss': 0.2711, 'grad_norm': 0.5872403979301453, 'learning_rate': 9.989608376580459e-06, 'epoch': 0.36}

 12%|█▏        | 1578/13317 [16:16<2:00:04,  1.63it/s]
 12%|█▏        | 1579/13317 [16:16<2:00:12,  1.63it/s]
                                                      
{'loss': 0.3459, 'grad_norm': 0.6084349751472473, 'learning_rate': 9.98952374969534e-06, 'epoch': 0.36}

 12%|█▏        | 1579/13317 [16:16<2:00:12,  1.63it/s]
 12%|█▏        | 1580/13317 [16:17<2:00:04,  1.63it/s]
                                                      
{'loss': 0.3283, 'grad_norm': 0.7621054649353027, 'learning_rate': 9.989438779977112e-06, 'epoch': 0.36}

 12%|█▏        | 1580/13317 [16:17<2:00:04,  1.63it/s]
 12%|█▏        | 1581/13317 [16:17<2:00:05,  1.63it/s]
                                                      
{'loss': 0.3227, 'grad_norm': 0.6871142387390137, 'learning_rate': 9.989353467431608e-06, 'epoch': 0.36}

 12%|█▏        | 1581/13317 [16:18<2:00:05,  1.63it/s]
 12%|█▏        | 1582/13317 [16:18<2:00:00,  1.63it/s]
                                                      
{'loss': 0.3341, 'grad_norm': 0.7874228954315186, 'learning_rate': 9.989267812064694e-06, 'epoch': 0.36}

 12%|█▏        | 1582/13317 [16:18<2:00:00,  1.63it/s]
 12%|█▏        | 1583/13317 [16:19<1:59:58,  1.63it/s]
                                                      
{'loss': 0.2521, 'grad_norm': 0.6040547490119934, 'learning_rate': 9.989181813882253e-06, 'epoch': 0.36}

 12%|█▏        | 1583/13317 [16:19<1:59:58,  1.63it/s]
 12%|█▏        | 1584/13317 [16:19<2:00:03,  1.63it/s]
                                                      
{'loss': 0.2974, 'grad_norm': 0.6873297095298767, 'learning_rate': 9.989095472890197e-06, 'epoch': 0.36}

 12%|█▏        | 1584/13317 [16:19<2:00:03,  1.63it/s]
 12%|█▏        | 1585/13317 [16:20<2:00:03,  1.63it/s]
                                                      
{'loss': 0.3576, 'grad_norm': 0.6518840789794922, 'learning_rate': 9.989008789094454e-06, 'epoch': 0.36}

 12%|█▏        | 1585/13317 [16:20<2:00:03,  1.63it/s]
 12%|█▏        | 1586/13317 [16:21<2:00:00,  1.63it/s]
                                                      
{'loss': 0.2879, 'grad_norm': 0.613768458366394, 'learning_rate': 9.988921762500985e-06, 'epoch': 0.36}

 12%|█▏        | 1586/13317 [16:21<2:00:00,  1.63it/s]
 12%|█▏        | 1587/13317 [16:21<1:59:56,  1.63it/s]
                                                      
{'loss': 0.3859, 'grad_norm': 0.6537529826164246, 'learning_rate': 9.988834393115768e-06, 'epoch': 0.36}

 12%|█▏        | 1587/13317 [16:21<1:59:56,  1.63it/s]
 12%|█▏        | 1588/13317 [16:22<1:59:54,  1.63it/s]
                                                      
{'loss': 0.3039, 'grad_norm': 0.7612906694412231, 'learning_rate': 9.988746680944803e-06, 'epoch': 0.36}

 12%|█▏        | 1588/13317 [16:22<1:59:54,  1.63it/s]
 12%|█▏        | 1589/13317 [16:22<2:00:00,  1.63it/s]
                                                      
{'loss': 0.341, 'grad_norm': 0.663769543170929, 'learning_rate': 9.98865862599412e-06, 'epoch': 0.36}

 12%|█▏        | 1589/13317 [16:23<2:00:00,  1.63it/s]
 12%|█▏        | 1590/13317 [16:23<1:59:54,  1.63it/s]
                                                      
{'loss': 0.3028, 'grad_norm': 0.6051310896873474, 'learning_rate': 9.98857022826977e-06, 'epoch': 0.36}

 12%|█▏        | 1590/13317 [16:23<1:59:54,  1.63it/s]
 12%|█▏        | 1591/13317 [16:24<1:59:51,  1.63it/s]
                                                      
{'loss': 0.2417, 'grad_norm': 0.6683542132377625, 'learning_rate': 9.988481487777823e-06, 'epoch': 0.36}

 12%|█▏        | 1591/13317 [16:24<1:59:51,  1.63it/s]
 12%|█▏        | 1592/13317 [16:24<1:59:49,  1.63it/s]
                                                      
{'loss': 0.267, 'grad_norm': 0.6020950675010681, 'learning_rate': 9.988392404524381e-06, 'epoch': 0.36}

 12%|█▏        | 1592/13317 [16:24<1:59:49,  1.63it/s]
 12%|█▏        | 1593/13317 [16:25<1:59:54,  1.63it/s]
                                                      
{'loss': 0.2644, 'grad_norm': 0.5997943878173828, 'learning_rate': 9.988302978515562e-06, 'epoch': 0.36}

 12%|█▏        | 1593/13317 [16:25<1:59:54,  1.63it/s]
 12%|█▏        | 1594/13317 [16:25<2:00:25,  1.62it/s]
                                                      
{'loss': 0.3166, 'grad_norm': 0.6629067063331604, 'learning_rate': 9.988213209757512e-06, 'epoch': 0.36}

 12%|█▏        | 1594/13317 [16:26<2:00:25,  1.62it/s]
 12%|█▏        | 1595/13317 [16:26<2:00:17,  1.62it/s]
                                                      
{'loss': 0.3526, 'grad_norm': 0.6415657997131348, 'learning_rate': 9.988123098256398e-06, 'epoch': 0.36}

 12%|█▏        | 1595/13317 [16:26<2:00:17,  1.62it/s]
 12%|█▏        | 1596/13317 [16:27<2:00:16,  1.62it/s]
                                                      
{'loss': 0.276, 'grad_norm': 0.6707572340965271, 'learning_rate': 9.988032644018412e-06, 'epoch': 0.36}

 12%|█▏        | 1596/13317 [16:27<2:00:16,  1.62it/s]
 12%|█▏        | 1597/13317 [16:27<2:00:07,  1.63it/s]
                                                      
{'loss': 0.2945, 'grad_norm': 0.6415923237800598, 'learning_rate': 9.98794184704977e-06, 'epoch': 0.36}

 12%|█▏        | 1597/13317 [16:27<2:00:07,  1.63it/s]
 12%|█▏        | 1598/13317 [16:28<2:00:01,  1.63it/s]
                                                      
{'loss': 0.2766, 'grad_norm': 0.6745001673698425, 'learning_rate': 9.987850707356709e-06, 'epoch': 0.36}

 12%|█▏        | 1598/13317 [16:28<2:00:01,  1.63it/s]
 12%|█▏        | 1599/13317 [16:29<2:00:06,  1.63it/s]
                                                      
{'loss': 0.4038, 'grad_norm': 0.7221595644950867, 'learning_rate': 9.987759224945492e-06, 'epoch': 0.36}

 12%|█▏        | 1599/13317 [16:29<2:00:06,  1.63it/s]
 12%|█▏        | 1600/13317 [16:29<2:00:01,  1.63it/s]
                                                      
{'loss': 0.3666, 'grad_norm': 0.6084919571876526, 'learning_rate': 9.987667399822406e-06, 'epoch': 0.36}

 12%|█▏        | 1600/13317 [16:29<2:00:01,  1.63it/s]
 12%|█▏        | 1601/13317 [16:30<1:59:58,  1.63it/s]
                                                      
{'loss': 0.2844, 'grad_norm': 0.5940876603126526, 'learning_rate': 9.98757523199376e-06, 'epoch': 0.36}

 12%|█▏        | 1601/13317 [16:30<1:59:58,  1.63it/s]
 12%|█▏        | 1602/13317 [16:30<1:59:52,  1.63it/s]
                                                      
{'loss': 0.3491, 'grad_norm': 0.6506302952766418, 'learning_rate': 9.987482721465884e-06, 'epoch': 0.36}

 12%|█▏        | 1602/13317 [16:31<1:59:52,  1.63it/s]
 12%|█▏        | 1603/13317 [16:31<1:59:49,  1.63it/s]
                                                      
{'loss': 0.3561, 'grad_norm': 0.873542070388794, 'learning_rate': 9.98738986824514e-06, 'epoch': 0.36}

 12%|█▏        | 1603/13317 [16:31<1:59:49,  1.63it/s]
 12%|█▏        | 1604/13317 [16:32<1:59:58,  1.63it/s]
                                                      
{'loss': 0.3388, 'grad_norm': 0.5724976658821106, 'learning_rate': 9.9872966723379e-06, 'epoch': 0.36}

 12%|█▏        | 1604/13317 [16:32<1:59:58,  1.63it/s]
 12%|█▏        | 1605/13317 [16:32<1:59:53,  1.63it/s]
                                                      
{'loss': 0.3253, 'grad_norm': 0.6535359025001526, 'learning_rate': 9.987203133750576e-06, 'epoch': 0.36}

 12%|█▏        | 1605/13317 [16:32<1:59:53,  1.63it/s]
 12%|█▏        | 1606/13317 [16:33<1:59:56,  1.63it/s]
                                                      
{'loss': 0.3363, 'grad_norm': 0.61922687292099, 'learning_rate': 9.987109252489589e-06, 'epoch': 0.36}

 12%|█▏        | 1606/13317 [16:33<1:59:56,  1.63it/s]
 12%|█▏        | 1607/13317 [16:33<1:59:50,  1.63it/s]
                                                      
{'loss': 0.229, 'grad_norm': 0.5566198229789734, 'learning_rate': 9.987015028561392e-06, 'epoch': 0.36}

 12%|█▏        | 1607/13317 [16:34<1:59:50,  1.63it/s]
 12%|█▏        | 1608/13317 [16:34<1:59:47,  1.63it/s]
                                                      
{'loss': 0.283, 'grad_norm': 0.6740355491638184, 'learning_rate': 9.986920461972463e-06, 'epoch': 0.36}

 12%|█▏        | 1608/13317 [16:34<1:59:47,  1.63it/s]
 12%|█▏        | 1609/13317 [16:35<1:59:54,  1.63it/s]
                                                      
{'loss': 0.4075, 'grad_norm': 0.7930310368537903, 'learning_rate': 9.986825552729292e-06, 'epoch': 0.36}

 12%|█▏        | 1609/13317 [16:35<1:59:54,  1.63it/s]
 12%|█▏        | 1610/13317 [16:35<1:59:49,  1.63it/s]
                                                      
{'loss': 0.3744, 'grad_norm': 0.8671566247940063, 'learning_rate': 9.986730300838406e-06, 'epoch': 0.36}

 12%|█▏        | 1610/13317 [16:35<1:59:49,  1.63it/s]
 12%|█▏        | 1611/13317 [16:36<1:59:43,  1.63it/s]
                                                      
{'loss': 0.233, 'grad_norm': 0.5187944173812866, 'learning_rate': 9.986634706306348e-06, 'epoch': 0.36}

 12%|█▏        | 1611/13317 [16:36<1:59:43,  1.63it/s]
 12%|█▏        | 1612/13317 [16:37<1:59:42,  1.63it/s]
                                                      
{'loss': 0.2545, 'grad_norm': 0.5974021553993225, 'learning_rate': 9.986538769139685e-06, 'epoch': 0.36}

 12%|█▏        | 1612/13317 [16:37<1:59:42,  1.63it/s]
 12%|█▏        | 1613/13317 [16:37<1:59:42,  1.63it/s]
                                                      
{'loss': 0.3339, 'grad_norm': 0.7477859854698181, 'learning_rate': 9.986442489345011e-06, 'epoch': 0.36}

 12%|█▏        | 1613/13317 [16:37<1:59:42,  1.63it/s]
 12%|█▏        | 1614/13317 [16:38<1:59:48,  1.63it/s]
                                                      
{'loss': 0.323, 'grad_norm': 0.6506509184837341, 'learning_rate': 9.98634586692894e-06, 'epoch': 0.36}

 12%|█▏        | 1614/13317 [16:38<1:59:48,  1.63it/s]
 12%|█▏        | 1615/13317 [16:38<1:59:44,  1.63it/s]
                                                      
{'loss': 0.2862, 'grad_norm': 0.5881049633026123, 'learning_rate': 9.986248901898113e-06, 'epoch': 0.36}

 12%|█▏        | 1615/13317 [16:38<1:59:44,  1.63it/s]
 12%|█▏        | 1616/13317 [16:39<1:59:42,  1.63it/s]
                                                      
{'loss': 0.3955, 'grad_norm': 0.8114415407180786, 'learning_rate': 9.98615159425919e-06, 'epoch': 0.36}

 12%|█▏        | 1616/13317 [16:39<1:59:42,  1.63it/s]
 12%|█▏        | 1617/13317 [16:40<1:59:38,  1.63it/s]
                                                      
{'loss': 0.27, 'grad_norm': 0.6909857392311096, 'learning_rate': 9.986053944018859e-06, 'epoch': 0.36}

 12%|█▏        | 1617/13317 [16:40<1:59:38,  1.63it/s]
 12%|█▏        | 1618/13317 [16:40<1:59:34,  1.63it/s]
                                                      
{'loss': 0.3339, 'grad_norm': 0.6472762823104858, 'learning_rate': 9.98595595118383e-06, 'epoch': 0.36}

 12%|█▏        | 1618/13317 [16:40<1:59:34,  1.63it/s]
 12%|█▏        | 1619/13317 [16:41<1:59:40,  1.63it/s]
                                                      
{'loss': 0.2839, 'grad_norm': 0.6418958306312561, 'learning_rate': 9.985857615760833e-06, 'epoch': 0.36}

 12%|█▏        | 1619/13317 [16:41<1:59:40,  1.63it/s]
 12%|█▏        | 1620/13317 [16:41<1:59:36,  1.63it/s]
                                                      
{'loss': 0.3463, 'grad_norm': 0.6016213297843933, 'learning_rate': 9.985758937756627e-06, 'epoch': 0.36}

 12%|█▏        | 1620/13317 [16:42<1:59:36,  1.63it/s]
 12%|█▏        | 1621/13317 [16:42<1:59:36,  1.63it/s]
                                                      
{'loss': 0.3433, 'grad_norm': 0.6214809417724609, 'learning_rate': 9.985659917177991e-06, 'epoch': 0.37}

 12%|█▏        | 1621/13317 [16:42<1:59:36,  1.63it/s]
 12%|█▏        | 1622/13317 [16:43<1:59:37,  1.63it/s]
                                                      
{'loss': 0.3787, 'grad_norm': 0.6828439831733704, 'learning_rate': 9.985560554031732e-06, 'epoch': 0.37}

 12%|█▏        | 1622/13317 [16:43<1:59:37,  1.63it/s]
 12%|█▏        | 1623/13317 [16:43<1:59:31,  1.63it/s]
                                                      
{'loss': 0.2459, 'grad_norm': 0.5995084047317505, 'learning_rate': 9.985460848324674e-06, 'epoch': 0.37}

 12%|█▏        | 1623/13317 [16:43<1:59:31,  1.63it/s]
 12%|█▏        | 1624/13317 [16:44<1:59:44,  1.63it/s]
                                                      
{'loss': 0.2253, 'grad_norm': 0.5669989585876465, 'learning_rate': 9.985360800063671e-06, 'epoch': 0.37}

 12%|█▏        | 1624/13317 [16:44<1:59:44,  1.63it/s]
 12%|█▏        | 1625/13317 [16:45<2:00:01,  1.62it/s]
                                                      
{'loss': 0.3857, 'grad_norm': 0.5773236155509949, 'learning_rate': 9.985260409255591e-06, 'epoch': 0.37}

 12%|█▏        | 1625/13317 [16:45<2:00:01,  1.62it/s]
 12%|█▏        | 1626/13317 [16:45<1:59:47,  1.63it/s]
                                                      
{'loss': 0.3069, 'grad_norm': 0.6942152976989746, 'learning_rate': 9.985159675907338e-06, 'epoch': 0.37}

 12%|█▏        | 1626/13317 [16:45<1:59:47,  1.63it/s]
 12%|█▏        | 1627/13317 [16:46<1:59:43,  1.63it/s]
                                                      
{'loss': 0.3426, 'grad_norm': 0.7319950461387634, 'learning_rate': 9.985058600025833e-06, 'epoch': 0.37}

 12%|█▏        | 1627/13317 [16:46<1:59:43,  1.63it/s]
 12%|█▏        | 1628/13317 [16:46<1:59:38,  1.63it/s]
                                                      
{'loss': 0.2286, 'grad_norm': 0.532054603099823, 'learning_rate': 9.984957181618019e-06, 'epoch': 0.37}

 12%|█▏        | 1628/13317 [16:46<1:59:38,  1.63it/s]
 12%|█▏        | 1629/13317 [16:47<1:59:41,  1.63it/s]
                                                      
{'loss': 0.2912, 'grad_norm': 0.6126400232315063, 'learning_rate': 9.984855420690862e-06, 'epoch': 0.37}

 12%|█▏        | 1629/13317 [16:47<1:59:41,  1.63it/s]
 12%|█▏        | 1630/13317 [16:48<1:59:34,  1.63it/s]
                                                      
{'loss': 0.2429, 'grad_norm': 0.5522881746292114, 'learning_rate': 9.98475331725136e-06, 'epoch': 0.37}

 12%|█▏        | 1630/13317 [16:48<1:59:34,  1.63it/s]
 12%|█▏        | 1631/13317 [16:48<1:59:37,  1.63it/s]
                                                      
{'loss': 0.3082, 'grad_norm': 0.736790120601654, 'learning_rate': 9.984650871306526e-06, 'epoch': 0.37}

 12%|█▏        | 1631/13317 [16:48<1:59:37,  1.63it/s]
 12%|█▏        | 1632/13317 [16:49<1:59:31,  1.63it/s]
                                                      
{'loss': 0.3477, 'grad_norm': 0.7059333920478821, 'learning_rate': 9.984548082863397e-06, 'epoch': 0.37}

 12%|█▏        | 1632/13317 [16:49<1:59:31,  1.63it/s]
 12%|█▏        | 1633/13317 [16:49<1:59:25,  1.63it/s]
                                                      
{'loss': 0.377, 'grad_norm': 0.6925734877586365, 'learning_rate': 9.984444951929037e-06, 'epoch': 0.37}

 12%|█▏        | 1633/13317 [16:50<1:59:25,  1.63it/s]
 12%|█▏        | 1634/13317 [16:50<1:59:39,  1.63it/s]
                                                      
{'loss': 0.3202, 'grad_norm': 0.7043500542640686, 'learning_rate': 9.984341478510536e-06, 'epoch': 0.37}

 12%|█▏        | 1634/13317 [16:50<1:59:39,  1.63it/s]
 12%|█▏        | 1635/13317 [16:51<1:59:33,  1.63it/s]
                                                      
{'loss': 0.2571, 'grad_norm': 0.6072131991386414, 'learning_rate': 9.984237662614997e-06, 'epoch': 0.37}

 12%|█▏        | 1635/13317 [16:51<1:59:33,  1.63it/s]
 12%|█▏        | 1636/13317 [16:51<1:59:33,  1.63it/s]
                                                      
{'loss': 0.2766, 'grad_norm': 0.6065596342086792, 'learning_rate': 9.984133504249557e-06, 'epoch': 0.37}

 12%|█▏        | 1636/13317 [16:51<1:59:33,  1.63it/s]
 12%|█▏        | 1637/13317 [16:52<1:59:30,  1.63it/s]
                                                      
{'loss': 0.3229, 'grad_norm': 0.709509551525116, 'learning_rate': 9.984029003421373e-06, 'epoch': 0.37}

 12%|█▏        | 1637/13317 [16:52<1:59:30,  1.63it/s]
 12%|█▏        | 1638/13317 [16:52<1:59:30,  1.63it/s]
                                                      
{'loss': 0.2572, 'grad_norm': 0.573919415473938, 'learning_rate': 9.983924160137627e-06, 'epoch': 0.37}

 12%|█▏        | 1638/13317 [16:53<1:59:30,  1.63it/s]
 12%|█▏        | 1639/13317 [16:53<1:59:39,  1.63it/s]
                                                      
{'loss': 0.2716, 'grad_norm': 0.6051305532455444, 'learning_rate': 9.983818974405518e-06, 'epoch': 0.37}

 12%|█▏        | 1639/13317 [16:53<1:59:39,  1.63it/s]
 12%|█▏        | 1640/13317 [16:54<1:59:35,  1.63it/s]
                                                      
{'loss': 0.3182, 'grad_norm': 0.644743025302887, 'learning_rate': 9.983713446232277e-06, 'epoch': 0.37}

 12%|█▏        | 1640/13317 [16:54<1:59:35,  1.63it/s]
 12%|█▏        | 1641/13317 [16:54<1:59:30,  1.63it/s]
                                                      
{'loss': 0.2418, 'grad_norm': 0.5147328972816467, 'learning_rate': 9.983607575625152e-06, 'epoch': 0.37}

 12%|█▏        | 1641/13317 [16:54<1:59:30,  1.63it/s]
 12%|█▏        | 1642/13317 [16:55<1:59:25,  1.63it/s]
                                                      
{'loss': 0.288, 'grad_norm': 0.6153809428215027, 'learning_rate': 9.983501362591421e-06, 'epoch': 0.37}

 12%|█▏        | 1642/13317 [16:55<1:59:25,  1.63it/s]
 12%|█▏        | 1643/13317 [16:56<1:59:20,  1.63it/s]
                                                      
{'loss': 0.323, 'grad_norm': 0.64957195520401, 'learning_rate': 9.98339480713838e-06, 'epoch': 0.37}

 12%|█▏        | 1643/13317 [16:56<1:59:20,  1.63it/s]
 12%|█▏        | 1644/13317 [16:56<1:59:27,  1.63it/s]
                                                      
{'loss': 0.2571, 'grad_norm': 0.5807137489318848, 'learning_rate': 9.983287909273351e-06, 'epoch': 0.37}

 12%|█▏        | 1644/13317 [16:56<1:59:27,  1.63it/s]
 12%|█▏        | 1645/13317 [16:57<1:59:22,  1.63it/s]
                                                      
{'loss': 0.3351, 'grad_norm': 0.6333628296852112, 'learning_rate': 9.983180669003677e-06, 'epoch': 0.37}

 12%|█▏        | 1645/13317 [16:57<1:59:22,  1.63it/s]
 12%|█▏        | 1646/13317 [16:57<1:59:23,  1.63it/s]
                                                      
{'loss': 0.2865, 'grad_norm': 0.5864414572715759, 'learning_rate': 9.98307308633673e-06, 'epoch': 0.37}

 12%|█▏        | 1646/13317 [16:58<1:59:23,  1.63it/s]
 12%|█▏        | 1647/13317 [16:58<1:59:19,  1.63it/s]
                                                      
{'loss': 0.288, 'grad_norm': 0.6341243386268616, 'learning_rate': 9.982965161279901e-06, 'epoch': 0.37}

 12%|█▏        | 1647/13317 [16:58<1:59:19,  1.63it/s]
 12%|█▏        | 1648/13317 [16:59<1:59:18,  1.63it/s]
                                                      
{'loss': 0.3524, 'grad_norm': 0.7045326828956604, 'learning_rate': 9.982856893840605e-06, 'epoch': 0.37}

 12%|█▏        | 1648/13317 [16:59<1:59:18,  1.63it/s]
 12%|█▏        | 1649/13317 [16:59<1:59:55,  1.62it/s]
                                                      
{'loss': 0.303, 'grad_norm': 0.5826371908187866, 'learning_rate': 9.98274828402628e-06, 'epoch': 0.37}

 12%|█▏        | 1649/13317 [16:59<1:59:55,  1.62it/s]
 12%|█▏        | 1650/13317 [17:00<1:59:44,  1.62it/s]
                                                      
{'loss': 0.3335, 'grad_norm': 0.7415836453437805, 'learning_rate': 9.982639331844388e-06, 'epoch': 0.37}

 12%|█▏        | 1650/13317 [17:00<1:59:44,  1.62it/s]
 12%|█▏        | 1651/13317 [17:00<1:59:31,  1.63it/s]
                                                      
{'loss': 0.2914, 'grad_norm': 0.5675320625305176, 'learning_rate': 9.98253003730242e-06, 'epoch': 0.37}

 12%|█▏        | 1651/13317 [17:01<1:59:31,  1.63it/s]
 12%|█▏        | 1652/13317 [17:01<1:59:29,  1.63it/s]
                                                      
{'loss': 0.2602, 'grad_norm': 0.539586067199707, 'learning_rate': 9.98242040040788e-06, 'epoch': 0.37}

 12%|█▏        | 1652/13317 [17:01<1:59:29,  1.63it/s]
 12%|█▏        | 1653/13317 [17:02<1:59:26,  1.63it/s]
                                                      
{'loss': 0.3388, 'grad_norm': 0.600691556930542, 'learning_rate': 9.982310421168305e-06, 'epoch': 0.37}

 12%|█▏        | 1653/13317 [17:02<1:59:26,  1.63it/s]
 12%|█▏        | 1654/13317 [17:02<2:06:04,  1.54it/s]
                                                      
{'loss': 0.2974, 'grad_norm': 0.5979695916175842, 'learning_rate': 9.98220009959125e-06, 'epoch': 0.37}

 12%|█▏        | 1654/13317 [17:03<2:06:04,  1.54it/s]
 12%|█▏        | 1655/13317 [17:03<2:04:01,  1.57it/s]
                                                      
{'loss': 0.3398, 'grad_norm': 0.7592716813087463, 'learning_rate': 9.982089435684297e-06, 'epoch': 0.37}

 12%|█▏        | 1655/13317 [17:03<2:04:01,  1.57it/s]
 12%|█▏        | 1656/13317 [17:04<2:02:40,  1.58it/s]
                                                      
{'loss': 0.3509, 'grad_norm': 0.663838803768158, 'learning_rate': 9.981978429455049e-06, 'epoch': 0.37}

 12%|█▏        | 1656/13317 [17:04<2:02:40,  1.58it/s]
 12%|█▏        | 1657/13317 [17:04<2:01:35,  1.60it/s]
                                                      
{'loss': 0.2392, 'grad_norm': 0.6685572862625122, 'learning_rate': 9.98186708091113e-06, 'epoch': 0.37}

 12%|█▏        | 1657/13317 [17:04<2:01:35,  1.60it/s]
 12%|█▏        | 1658/13317 [17:05<2:00:50,  1.61it/s]
                                                      
{'loss': 0.2865, 'grad_norm': 0.6354819536209106, 'learning_rate': 9.981755390060195e-06, 'epoch': 0.37}

 12%|█▏        | 1658/13317 [17:05<2:00:50,  1.61it/s]
 12%|█▏        | 1659/13317 [17:05<2:00:35,  1.61it/s]
                                                      
{'loss': 0.2911, 'grad_norm': 0.6183892488479614, 'learning_rate': 9.981643356909918e-06, 'epoch': 0.37}

 12%|█▏        | 1659/13317 [17:06<2:00:35,  1.61it/s]
 12%|█▏        | 1660/13317 [17:06<2:00:05,  1.62it/s]
                                                      
{'loss': 0.2512, 'grad_norm': 0.6173778176307678, 'learning_rate': 9.981530981467996e-06, 'epoch': 0.37}

 12%|█▏        | 1660/13317 [17:06<2:00:05,  1.62it/s]
 12%|█▏        | 1661/13317 [17:07<1:59:48,  1.62it/s]
                                                      
{'loss': 0.3627, 'grad_norm': 0.7121611833572388, 'learning_rate': 9.981418263742148e-06, 'epoch': 0.37}

 12%|█▏        | 1661/13317 [17:07<1:59:48,  1.62it/s]
 12%|█▏        | 1662/13317 [17:07<1:59:35,  1.62it/s]
                                                      
{'loss': 0.2992, 'grad_norm': 0.5429452061653137, 'learning_rate': 9.981305203740124e-06, 'epoch': 0.37}

 12%|█▏        | 1662/13317 [17:07<1:59:35,  1.62it/s]
 12%|█▏        | 1663/13317 [17:08<1:59:24,  1.63it/s]
                                                      
{'loss': 0.2821, 'grad_norm': 0.6010006070137024, 'learning_rate': 9.981191801469687e-06, 'epoch': 0.37}

 12%|█▏        | 1663/13317 [17:08<1:59:24,  1.63it/s]
 12%|█▏        | 1664/13317 [17:09<1:59:31,  1.62it/s]
                                                      
{'loss': 0.3463, 'grad_norm': 1.0572196245193481, 'learning_rate': 9.981078056938634e-06, 'epoch': 0.37}

 12%|█▏        | 1664/13317 [17:09<1:59:31,  1.62it/s]
 13%|█▎        | 1665/13317 [17:09<1:59:25,  1.63it/s]
                                                      
{'loss': 0.3675, 'grad_norm': 0.6164083480834961, 'learning_rate': 9.980963970154775e-06, 'epoch': 0.38}

 13%|█▎        | 1665/13317 [17:09<1:59:25,  1.63it/s]
 13%|█▎        | 1666/13317 [17:10<1:59:16,  1.63it/s]
                                                      
{'loss': 0.2937, 'grad_norm': 0.6344239115715027, 'learning_rate': 9.980849541125955e-06, 'epoch': 0.38}

 13%|█▎        | 1666/13317 [17:10<1:59:16,  1.63it/s]
 13%|█▎        | 1667/13317 [17:10<1:59:19,  1.63it/s]
                                                      
{'loss': 0.251, 'grad_norm': 0.49833056330680847, 'learning_rate': 9.980734769860032e-06, 'epoch': 0.38}

 13%|█▎        | 1667/13317 [17:11<1:59:19,  1.63it/s]
 13%|█▎        | 1668/13317 [17:11<1:59:19,  1.63it/s]
                                                      
{'loss': 0.4019, 'grad_norm': 0.6822938323020935, 'learning_rate': 9.980619656364893e-06, 'epoch': 0.38}

 13%|█▎        | 1668/13317 [17:11<1:59:19,  1.63it/s]
 13%|█▎        | 1669/13317 [17:12<1:59:11,  1.63it/s]
                                                      
{'loss': 0.3002, 'grad_norm': 0.5592580437660217, 'learning_rate': 9.980504200648449e-06, 'epoch': 0.38}

 13%|█▎        | 1669/13317 [17:12<1:59:11,  1.63it/s]
 13%|█▎        | 1670/13317 [17:12<1:59:12,  1.63it/s]
                                                      
{'loss': 0.2481, 'grad_norm': 0.5319152474403381, 'learning_rate': 9.980388402718632e-06, 'epoch': 0.38}

 13%|█▎        | 1670/13317 [17:12<1:59:12,  1.63it/s]
 13%|█▎        | 1671/13317 [17:13<1:59:08,  1.63it/s]
                                                      
{'loss': 0.2601, 'grad_norm': 0.6258807182312012, 'learning_rate': 9.980272262583397e-06, 'epoch': 0.38}

 13%|█▎        | 1671/13317 [17:13<1:59:08,  1.63it/s]
 13%|█▎        | 1672/13317 [17:13<1:59:14,  1.63it/s]
                                                      
{'loss': 0.3762, 'grad_norm': 0.7656760811805725, 'learning_rate': 9.980155780250728e-06, 'epoch': 0.38}

 13%|█▎        | 1672/13317 [17:14<1:59:14,  1.63it/s]
 13%|█▎        | 1673/13317 [17:14<2:02:20,  1.59it/s]
                                                      
{'loss': 0.3057, 'grad_norm': 0.6221503615379333, 'learning_rate': 9.980038955728623e-06, 'epoch': 0.38}

 13%|█▎        | 1673/13317 [17:14<2:02:20,  1.59it/s]
 13%|█▎        | 1674/13317 [17:15<2:01:28,  1.60it/s]
                                                      
{'loss': 0.2694, 'grad_norm': 0.5879343748092651, 'learning_rate': 9.979921789025115e-06, 'epoch': 0.38}

 13%|█▎        | 1674/13317 [17:15<2:01:28,  1.60it/s]
 13%|█▎        | 1675/13317 [17:15<2:00:45,  1.61it/s]
                                                      
{'loss': 0.2787, 'grad_norm': 0.6791156530380249, 'learning_rate': 9.979804280148251e-06, 'epoch': 0.38}

 13%|█▎        | 1675/13317 [17:16<2:00:45,  1.61it/s]
 13%|█▎        | 1676/13317 [17:16<2:00:18,  1.61it/s]
                                                      
{'loss': 0.336, 'grad_norm': 0.6434783339500427, 'learning_rate': 9.979686429106105e-06, 'epoch': 0.38}

 13%|█▎        | 1676/13317 [17:16<2:00:18,  1.61it/s]
 13%|█▎        | 1677/13317 [17:17<2:00:00,  1.62it/s]
                                                      
{'loss': 0.2752, 'grad_norm': 0.6271657347679138, 'learning_rate': 9.979568235906778e-06, 'epoch': 0.38}

 13%|█▎        | 1677/13317 [17:17<2:00:00,  1.62it/s]
 13%|█▎        | 1678/13317 [17:17<1:59:39,  1.62it/s]
                                                      
{'loss': 0.3361, 'grad_norm': 0.6475008726119995, 'learning_rate': 9.979449700558386e-06, 'epoch': 0.38}

 13%|█▎        | 1678/13317 [17:17<1:59:39,  1.62it/s]
 13%|█▎        | 1679/13317 [17:18<1:59:29,  1.62it/s]
                                                      
{'loss': 0.2519, 'grad_norm': 0.5658823251724243, 'learning_rate': 9.979330823069077e-06, 'epoch': 0.38}

 13%|█▎        | 1679/13317 [17:18<1:59:29,  1.62it/s]
 13%|█▎        | 1680/13317 [17:18<1:59:22,  1.62it/s]
                                                      
{'loss': 0.2699, 'grad_norm': 0.5690644383430481, 'learning_rate': 9.979211603447022e-06, 'epoch': 0.38}

 13%|█▎        | 1680/13317 [17:19<1:59:22,  1.62it/s]
 13%|█▎        | 1681/13317 [17:19<1:59:18,  1.63it/s]
                                                      
{'loss': 0.293, 'grad_norm': 0.6581867933273315, 'learning_rate': 9.979092041700405e-06, 'epoch': 0.38}

 13%|█▎        | 1681/13317 [17:19<1:59:18,  1.63it/s]
 13%|█▎        | 1682/13317 [17:20<1:59:18,  1.63it/s]
                                                      
{'loss': 0.2928, 'grad_norm': 0.6322510242462158, 'learning_rate': 9.978972137837447e-06, 'epoch': 0.38}

 13%|█▎        | 1682/13317 [17:20<1:59:18,  1.63it/s]
 13%|█▎        | 1683/13317 [17:20<1:59:16,  1.63it/s]
                                                      
{'loss': 0.3406, 'grad_norm': 0.6607187986373901, 'learning_rate': 9.978851891866384e-06, 'epoch': 0.38}

 13%|█▎        | 1683/13317 [17:20<1:59:16,  1.63it/s]
 13%|█▎        | 1684/13317 [17:21<1:59:11,  1.63it/s]
                                                      
{'loss': 0.2472, 'grad_norm': 0.6291475296020508, 'learning_rate': 9.978731303795478e-06, 'epoch': 0.38}

 13%|█▎        | 1684/13317 [17:21<1:59:11,  1.63it/s]
 13%|█▎        | 1685/13317 [17:22<1:59:07,  1.63it/s]
                                                      
{'loss': 0.3385, 'grad_norm': 0.6611707806587219, 'learning_rate': 9.978610373633018e-06, 'epoch': 0.38}

 13%|█▎        | 1685/13317 [17:22<1:59:07,  1.63it/s]
 13%|█▎        | 1686/13317 [17:22<1:59:09,  1.63it/s]
                                                      
{'loss': 0.2697, 'grad_norm': 0.5408490300178528, 'learning_rate': 9.97848910138731e-06, 'epoch': 0.38}

 13%|█▎        | 1686/13317 [17:22<1:59:09,  1.63it/s]
 13%|█▎        | 1687/13317 [17:23<1:59:37,  1.62it/s]
                                                      
{'loss': 0.3434, 'grad_norm': 0.6419597864151001, 'learning_rate': 9.978367487066686e-06, 'epoch': 0.38}

 13%|█▎        | 1687/13317 [17:23<1:59:37,  1.62it/s]
 13%|█▎        | 1688/13317 [17:23<1:59:24,  1.62it/s]
                                                      
{'loss': 0.3486, 'grad_norm': 0.6706961989402771, 'learning_rate': 9.978245530679506e-06, 'epoch': 0.38}

 13%|█▎        | 1688/13317 [17:24<1:59:24,  1.62it/s]
 13%|█▎        | 1689/13317 [17:24<1:59:16,  1.62it/s]
                                                      
{'loss': 0.2592, 'grad_norm': 0.5426585674285889, 'learning_rate': 9.978123232234147e-06, 'epoch': 0.38}

 13%|█▎        | 1689/13317 [17:24<1:59:16,  1.62it/s]
 13%|█▎        | 1690/13317 [17:25<1:59:12,  1.63it/s]
                                                      
{'loss': 0.3372, 'grad_norm': 0.5913063883781433, 'learning_rate': 9.978000591739013e-06, 'epoch': 0.38}

 13%|█▎        | 1690/13317 [17:25<1:59:12,  1.63it/s]
 13%|█▎        | 1691/13317 [17:25<1:59:10,  1.63it/s]
                                                      
{'loss': 0.3428, 'grad_norm': 0.7101878523826599, 'learning_rate': 9.977877609202532e-06, 'epoch': 0.38}

 13%|█▎        | 1691/13317 [17:25<1:59:10,  1.63it/s]
 13%|█▎        | 1692/13317 [17:26<1:59:13,  1.63it/s]
                                                      
{'loss': 0.3179, 'grad_norm': 0.7041009068489075, 'learning_rate': 9.97775428463315e-06, 'epoch': 0.38}

 13%|█▎        | 1692/13317 [17:26<1:59:13,  1.63it/s]
 13%|█▎        | 1693/13317 [17:26<1:59:06,  1.63it/s]
                                                      
{'loss': 0.306, 'grad_norm': 0.589478611946106, 'learning_rate': 9.977630618039343e-06, 'epoch': 0.38}

 13%|█▎        | 1693/13317 [17:27<1:59:06,  1.63it/s]
 13%|█▎        | 1694/13317 [17:27<1:59:13,  1.62it/s]
                                                      
{'loss': 0.2691, 'grad_norm': 0.580897331237793, 'learning_rate': 9.97750660942961e-06, 'epoch': 0.38}

 13%|█▎        | 1694/13317 [17:27<1:59:13,  1.62it/s]
 13%|█▎        | 1695/13317 [17:28<1:59:05,  1.63it/s]
                                                      
{'loss': 0.2666, 'grad_norm': 0.5953050255775452, 'learning_rate': 9.97738225881247e-06, 'epoch': 0.38}

 13%|█▎        | 1695/13317 [17:28<1:59:05,  1.63it/s]
 13%|█▎        | 1696/13317 [17:28<1:59:00,  1.63it/s]
                                                      
{'loss': 0.2369, 'grad_norm': 0.5370635390281677, 'learning_rate': 9.977257566196467e-06, 'epoch': 0.38}

 13%|█▎        | 1696/13317 [17:28<1:59:00,  1.63it/s]
 13%|█▎        | 1697/13317 [17:29<1:58:59,  1.63it/s]
                                                      
{'loss': 0.3123, 'grad_norm': 0.6764487624168396, 'learning_rate': 9.97713253159017e-06, 'epoch': 0.38}

 13%|█▎        | 1697/13317 [17:29<1:58:59,  1.63it/s]
 13%|█▎        | 1698/13317 [17:30<1:58:59,  1.63it/s]
                                                      
{'loss': 0.3662, 'grad_norm': 0.7740838527679443, 'learning_rate': 9.97700715500217e-06, 'epoch': 0.38}

 13%|█▎        | 1698/13317 [17:30<1:58:59,  1.63it/s]
 13%|█▎        | 1699/13317 [17:30<1:59:06,  1.63it/s]
                                                      
{'loss': 0.2802, 'grad_norm': 0.5380691289901733, 'learning_rate': 9.976881436441079e-06, 'epoch': 0.38}

 13%|█▎        | 1699/13317 [17:30<1:59:06,  1.63it/s]
 13%|█▎        | 1700/13317 [17:31<1:58:58,  1.63it/s]
                                                      
{'loss': 0.2742, 'grad_norm': 0.7481679320335388, 'learning_rate': 9.976755375915539e-06, 'epoch': 0.38}

 13%|█▎        | 1700/13317 [17:31<1:58:58,  1.63it/s]
 13%|█▎        | 1701/13317 [17:31<1:58:51,  1.63it/s]
                                                      
{'loss': 0.2559, 'grad_norm': 0.5968578457832336, 'learning_rate': 9.976628973434209e-06, 'epoch': 0.38}

 13%|█▎        | 1701/13317 [17:31<1:58:51,  1.63it/s]
 13%|█▎        | 1702/13317 [17:32<1:58:57,  1.63it/s]
                                                      
{'loss': 0.3116, 'grad_norm': 0.6032739877700806, 'learning_rate': 9.976502229005776e-06, 'epoch': 0.38}

 13%|█▎        | 1702/13317 [17:32<1:58:57,  1.63it/s]
 13%|█▎        | 1703/13317 [17:33<1:58:53,  1.63it/s]
                                                      
{'loss': 0.3095, 'grad_norm': 0.6267006397247314, 'learning_rate': 9.976375142638947e-06, 'epoch': 0.38}

 13%|█▎        | 1703/13317 [17:33<1:58:53,  1.63it/s]
 13%|█▎        | 1704/13317 [17:33<1:59:02,  1.63it/s]
                                                      
{'loss': 0.2766, 'grad_norm': 0.6560315489768982, 'learning_rate': 9.976247714342456e-06, 'epoch': 0.38}

 13%|█▎        | 1704/13317 [17:33<1:59:02,  1.63it/s]
 13%|█▎        | 1705/13317 [17:34<1:58:56,  1.63it/s]
                                                      
{'loss': 0.3218, 'grad_norm': 0.66346275806427, 'learning_rate': 9.976119944125056e-06, 'epoch': 0.38}

 13%|█▎        | 1705/13317 [17:34<1:58:56,  1.63it/s]
 13%|█▎        | 1706/13317 [17:34<1:58:56,  1.63it/s]
                                                      
{'loss': 0.309, 'grad_norm': 0.7122964859008789, 'learning_rate': 9.975991831995528e-06, 'epoch': 0.38}

 13%|█▎        | 1706/13317 [17:35<1:58:56,  1.63it/s]
 13%|█▎        | 1707/13317 [17:35<1:58:51,  1.63it/s]
                                                      
{'loss': 0.3353, 'grad_norm': 0.5711252093315125, 'learning_rate': 9.975863377962675e-06, 'epoch': 0.38}

 13%|█▎        | 1707/13317 [17:35<1:58:51,  1.63it/s]
 13%|█▎        | 1708/13317 [17:36<1:58:46,  1.63it/s]
                                                      
{'loss': 0.2729, 'grad_norm': 0.5379392504692078, 'learning_rate': 9.975734582035323e-06, 'epoch': 0.38}

 13%|█▎        | 1708/13317 [17:36<1:58:46,  1.63it/s]
 13%|█▎        | 1709/13317 [17:36<1:58:56,  1.63it/s]
                                                      
{'loss': 0.3409, 'grad_norm': 0.6945775151252747, 'learning_rate': 9.975605444222321e-06, 'epoch': 0.38}

 13%|█▎        | 1709/13317 [17:36<1:58:56,  1.63it/s]
 13%|█▎        | 1710/13317 [17:37<1:58:50,  1.63it/s]
                                                      
{'loss': 0.2945, 'grad_norm': 0.5662935376167297, 'learning_rate': 9.975475964532542e-06, 'epoch': 0.39}

 13%|█▎        | 1710/13317 [17:37<1:58:50,  1.63it/s]
 13%|█▎        | 1711/13317 [17:38<1:58:44,  1.63it/s]
                                                      
{'loss': 0.2982, 'grad_norm': 0.5397766828536987, 'learning_rate': 9.975346142974884e-06, 'epoch': 0.39}

 13%|█▎        | 1711/13317 [17:38<1:58:44,  1.63it/s]
 13%|█▎        | 1712/13317 [17:38<1:58:39,  1.63it/s]
                                                      
{'loss': 0.2753, 'grad_norm': 0.5687610507011414, 'learning_rate': 9.975215979558265e-06, 'epoch': 0.39}

 13%|█▎        | 1712/13317 [17:38<1:58:39,  1.63it/s]
 13%|█▎        | 1713/13317 [17:39<1:58:39,  1.63it/s]
                                                      
{'loss': 0.2537, 'grad_norm': 0.5450707077980042, 'learning_rate': 9.975085474291631e-06, 'epoch': 0.39}

 13%|█▎        | 1713/13317 [17:39<1:58:39,  1.63it/s]
 13%|█▎        | 1714/13317 [17:39<1:58:48,  1.63it/s]
                                                      
{'loss': 0.276, 'grad_norm': 0.566221296787262, 'learning_rate': 9.974954627183948e-06, 'epoch': 0.39}

 13%|█▎        | 1714/13317 [17:39<1:58:48,  1.63it/s]
 13%|█▎        | 1715/13317 [17:40<1:58:45,  1.63it/s]
                                                      
{'loss': 0.2824, 'grad_norm': 0.6113104224205017, 'learning_rate': 9.974823438244206e-06, 'epoch': 0.39}

 13%|█▎        | 1715/13317 [17:40<1:58:45,  1.63it/s]
 13%|█▎        | 1716/13317 [17:41<1:58:48,  1.63it/s]
                                                      
{'loss': 0.2803, 'grad_norm': 0.654252827167511, 'learning_rate': 9.97469190748142e-06, 'epoch': 0.39}

 13%|█▎        | 1716/13317 [17:41<1:58:48,  1.63it/s]
 13%|█▎        | 1717/13317 [17:41<1:58:42,  1.63it/s]
                                                      
{'loss': 0.215, 'grad_norm': 0.5215024352073669, 'learning_rate': 9.974560034904625e-06, 'epoch': 0.39}

 13%|█▎        | 1717/13317 [17:41<1:58:42,  1.63it/s]
 13%|█▎        | 1718/13317 [17:42<1:58:37,  1.63it/s]
                                                      
{'loss': 0.3149, 'grad_norm': 0.6157540678977966, 'learning_rate': 9.974427820522886e-06, 'epoch': 0.39}

 13%|█▎        | 1718/13317 [17:42<1:58:37,  1.63it/s]
 13%|█▎        | 1719/13317 [17:42<1:58:48,  1.63it/s]
                                                      
{'loss': 0.3182, 'grad_norm': 0.7331597208976746, 'learning_rate': 9.974295264345285e-06, 'epoch': 0.39}

 13%|█▎        | 1719/13317 [17:43<1:58:48,  1.63it/s]
 13%|█▎        | 1720/13317 [17:43<1:58:43,  1.63it/s]
                                                      
{'loss': 0.2682, 'grad_norm': 0.6250478029251099, 'learning_rate': 9.974162366380931e-06, 'epoch': 0.39}

 13%|█▎        | 1720/13317 [17:43<1:58:43,  1.63it/s]
 13%|█▎        | 1721/13317 [17:44<1:58:44,  1.63it/s]
                                                      
{'loss': 0.3106, 'grad_norm': 0.7325892448425293, 'learning_rate': 9.974029126638955e-06, 'epoch': 0.39}

 13%|█▎        | 1721/13317 [17:44<1:58:44,  1.63it/s]
 13%|█▎        | 1722/13317 [17:44<1:58:40,  1.63it/s]
                                                      
{'loss': 0.3146, 'grad_norm': 0.616768479347229, 'learning_rate': 9.973895545128512e-06, 'epoch': 0.39}

 13%|█▎        | 1722/13317 [17:44<1:58:40,  1.63it/s]
 13%|█▎        | 1723/13317 [17:45<1:58:41,  1.63it/s]
                                                      
{'loss': 0.243, 'grad_norm': 0.5098891258239746, 'learning_rate': 9.97376162185878e-06, 'epoch': 0.39}

 13%|█▎        | 1723/13317 [17:45<1:58:41,  1.63it/s]
 13%|█▎        | 1724/13317 [17:45<1:58:53,  1.63it/s]
                                                      
{'loss': 0.3147, 'grad_norm': 0.5726875066757202, 'learning_rate': 9.973627356838963e-06, 'epoch': 0.39}

 13%|█▎        | 1724/13317 [17:46<1:58:53,  1.63it/s]
 13%|█▎        | 1725/13317 [17:46<1:58:47,  1.63it/s]
                                                      
{'loss': 0.3599, 'grad_norm': 0.7426386475563049, 'learning_rate': 9.973492750078283e-06, 'epoch': 0.39}

 13%|█▎        | 1725/13317 [17:46<1:58:47,  1.63it/s]
 13%|█▎        | 1726/13317 [17:47<1:58:52,  1.63it/s]
                                                      
{'loss': 0.3156, 'grad_norm': 0.6673934459686279, 'learning_rate': 9.973357801585993e-06, 'epoch': 0.39}

 13%|█▎        | 1726/13317 [17:47<1:58:52,  1.63it/s]
 13%|█▎        | 1727/13317 [17:47<1:58:54,  1.62it/s]
                                                      
{'loss': 0.2872, 'grad_norm': 0.5152460932731628, 'learning_rate': 9.973222511371363e-06, 'epoch': 0.39}

 13%|█▎        | 1727/13317 [17:47<1:58:54,  1.62it/s]
 13%|█▎        | 1728/13317 [17:48<1:58:43,  1.63it/s]
                                                      
{'loss': 0.3391, 'grad_norm': 0.7077203392982483, 'learning_rate': 9.973086879443688e-06, 'epoch': 0.39}

 13%|█▎        | 1728/13317 [17:48<1:58:43,  1.63it/s]
 13%|█▎        | 1729/13317 [17:49<1:58:47,  1.63it/s]
                                                      
{'loss': 0.2892, 'grad_norm': 0.6743365526199341, 'learning_rate': 9.97295090581229e-06, 'epoch': 0.39}

 13%|█▎        | 1729/13317 [17:49<1:58:47,  1.63it/s]
 13%|█▎        | 1730/13317 [17:49<1:58:39,  1.63it/s]
                                                      
{'loss': 0.2723, 'grad_norm': 0.6242380738258362, 'learning_rate': 9.972814590486509e-06, 'epoch': 0.39}

 13%|█▎        | 1730/13317 [17:49<1:58:39,  1.63it/s]
 13%|█▎        | 1731/13317 [17:50<1:59:08,  1.62it/s]
                                                      
{'loss': 0.2515, 'grad_norm': 0.5548545718193054, 'learning_rate': 9.972677933475713e-06, 'epoch': 0.39}

 13%|█▎        | 1731/13317 [17:50<1:59:08,  1.62it/s]
 13%|█▎        | 1732/13317 [17:50<1:59:00,  1.62it/s]
                                                      
{'loss': 0.2411, 'grad_norm': 0.5687015056610107, 'learning_rate': 9.972540934789292e-06, 'epoch': 0.39}

 13%|█▎        | 1732/13317 [17:51<1:59:00,  1.62it/s]
 13%|█▎        | 1733/13317 [17:51<1:58:56,  1.62it/s]
                                                      
{'loss': 0.3105, 'grad_norm': 0.7780476212501526, 'learning_rate': 9.972403594436658e-06, 'epoch': 0.39}

 13%|█▎        | 1733/13317 [17:51<1:58:56,  1.62it/s]
 13%|█▎        | 1734/13317 [17:52<1:58:59,  1.62it/s]
                                                      
{'loss': 0.2249, 'grad_norm': 0.5201689600944519, 'learning_rate': 9.972265912427249e-06, 'epoch': 0.39}

 13%|█▎        | 1734/13317 [17:52<1:58:59,  1.62it/s]
 13%|█▎        | 1735/13317 [17:52<1:58:46,  1.63it/s]
                                                      
{'loss': 0.181, 'grad_norm': 0.5291252136230469, 'learning_rate': 9.972127888770522e-06, 'epoch': 0.39}

 13%|█▎        | 1735/13317 [17:52<1:58:46,  1.63it/s]
 13%|█▎        | 1736/13317 [17:53<1:58:43,  1.63it/s]
                                                      
{'loss': 0.2844, 'grad_norm': 0.6133894324302673, 'learning_rate': 9.971989523475967e-06, 'epoch': 0.39}

 13%|█▎        | 1736/13317 [17:53<1:58:43,  1.63it/s]
 13%|█▎        | 1737/13317 [17:53<1:58:44,  1.63it/s]
                                                      
{'loss': 0.3296, 'grad_norm': 0.7589443325996399, 'learning_rate': 9.971850816553085e-06, 'epoch': 0.39}

 13%|█▎        | 1737/13317 [17:54<1:58:44,  1.63it/s]
 13%|█▎        | 1738/13317 [17:54<1:58:37,  1.63it/s]
                                                      
{'loss': 0.3664, 'grad_norm': 0.6930487751960754, 'learning_rate': 9.971711768011409e-06, 'epoch': 0.39}

 13%|█▎        | 1738/13317 [17:54<1:58:37,  1.63it/s]
 13%|█▎        | 1739/13317 [17:55<1:58:42,  1.63it/s]
                                                      
{'loss': 0.2645, 'grad_norm': 0.7059414982795715, 'learning_rate': 9.971572377860496e-06, 'epoch': 0.39}

 13%|█▎        | 1739/13317 [17:55<1:58:42,  1.63it/s]
 13%|█▎        | 1740/13317 [17:55<1:58:35,  1.63it/s]
                                                      
{'loss': 0.3017, 'grad_norm': 0.6415818929672241, 'learning_rate': 9.971432646109919e-06, 'epoch': 0.39}

 13%|█▎        | 1740/13317 [17:55<1:58:35,  1.63it/s]
 13%|█▎        | 1741/13317 [17:56<1:58:35,  1.63it/s]
                                                      
{'loss': 0.2971, 'grad_norm': 0.641218364238739, 'learning_rate': 9.97129257276928e-06, 'epoch': 0.39}

 13%|█▎        | 1741/13317 [17:56<1:58:35,  1.63it/s]
 13%|█▎        | 1742/13317 [17:57<1:58:33,  1.63it/s]
                                                      
{'loss': 0.2626, 'grad_norm': 0.5867966413497925, 'learning_rate': 9.971152157848206e-06, 'epoch': 0.39}

 13%|█▎        | 1742/13317 [17:57<1:58:33,  1.63it/s]
 13%|█▎        | 1743/13317 [17:57<1:58:30,  1.63it/s]
                                                      
{'loss': 0.3085, 'grad_norm': 0.6901381611824036, 'learning_rate': 9.971011401356342e-06, 'epoch': 0.39}

 13%|█▎        | 1743/13317 [17:57<1:58:30,  1.63it/s]
 13%|█▎        | 1744/13317 [17:58<1:58:37,  1.63it/s]
                                                      
{'loss': 0.1943, 'grad_norm': 0.5165498852729797, 'learning_rate': 9.970870303303362e-06, 'epoch': 0.39}

 13%|█▎        | 1744/13317 [17:58<1:58:37,  1.63it/s]
 13%|█▎        | 1745/13317 [17:58<1:58:40,  1.63it/s]
                                                      
{'loss': 0.2933, 'grad_norm': 0.6397223472595215, 'learning_rate': 9.970728863698958e-06, 'epoch': 0.39}

 13%|█▎        | 1745/13317 [17:59<1:58:40,  1.63it/s]
 13%|█▎        | 1746/13317 [17:59<1:58:32,  1.63it/s]
                                                      
{'loss': 0.3379, 'grad_norm': 0.7278987169265747, 'learning_rate': 9.970587082552851e-06, 'epoch': 0.39}

 13%|█▎        | 1746/13317 [17:59<1:58:32,  1.63it/s]
 13%|█▎        | 1747/13317 [18:00<1:58:26,  1.63it/s]
                                                      
{'loss': 0.2868, 'grad_norm': 0.5591513514518738, 'learning_rate': 9.970444959874784e-06, 'epoch': 0.39}

 13%|█▎        | 1747/13317 [18:00<1:58:26,  1.63it/s]
 13%|█▎        | 1748/13317 [18:00<1:58:25,  1.63it/s]
                                                      
{'loss': 0.3342, 'grad_norm': 0.6152372360229492, 'learning_rate': 9.970302495674518e-06, 'epoch': 0.39}

 13%|█▎        | 1748/13317 [18:00<1:58:25,  1.63it/s]
 13%|█▎        | 1749/13317 [18:01<1:58:30,  1.63it/s]
                                                      
{'loss': 0.2819, 'grad_norm': 0.5072128772735596, 'learning_rate': 9.970159689961845e-06, 'epoch': 0.39}

 13%|█▎        | 1749/13317 [18:01<1:58:30,  1.63it/s]
 13%|█▎        | 1750/13317 [18:01<1:58:23,  1.63it/s]
                                                      
{'loss': 0.2649, 'grad_norm': 0.6591593027114868, 'learning_rate': 9.970016542746576e-06, 'epoch': 0.39}

 13%|█▎        | 1750/13317 [18:02<1:58:23,  1.63it/s]
 13%|█▎        | 1751/13317 [18:02<1:58:21,  1.63it/s]
                                                      
{'loss': 0.2351, 'grad_norm': 0.53995680809021, 'learning_rate': 9.969873054038548e-06, 'epoch': 0.39}

 13%|█▎        | 1751/13317 [18:02<1:58:21,  1.63it/s]
 13%|█▎        | 1752/13317 [18:03<1:58:17,  1.63it/s]
                                                      
{'loss': 0.3284, 'grad_norm': 0.6365072727203369, 'learning_rate': 9.969729223847618e-06, 'epoch': 0.39}

 13%|█▎        | 1752/13317 [18:03<1:58:17,  1.63it/s]
 13%|█▎        | 1753/13317 [18:03<1:58:15,  1.63it/s]
                                                      
{'loss': 0.2558, 'grad_norm': 0.5840824842453003, 'learning_rate': 9.96958505218367e-06, 'epoch': 0.39}

 13%|█▎        | 1753/13317 [18:03<1:58:15,  1.63it/s]
 13%|█▎        | 1754/13317 [18:04<1:58:25,  1.63it/s]
                                                      
{'loss': 0.2823, 'grad_norm': 0.5428082346916199, 'learning_rate': 9.96944053905661e-06, 'epoch': 0.4}

 13%|█▎        | 1754/13317 [18:04<1:58:25,  1.63it/s]
 13%|█▎        | 1755/13317 [18:05<1:58:22,  1.63it/s]
                                                      
{'loss': 0.3255, 'grad_norm': 0.6296663880348206, 'learning_rate': 9.96929568447637e-06, 'epoch': 0.4}

 13%|█▎        | 1755/13317 [18:05<1:58:22,  1.63it/s]
 13%|█▎        | 1756/13317 [18:05<1:58:16,  1.63it/s]
                                                      
{'loss': 0.255, 'grad_norm': 0.569131076335907, 'learning_rate': 9.969150488452899e-06, 'epoch': 0.4}

 13%|█▎        | 1756/13317 [18:05<1:58:16,  1.63it/s]
 13%|█▎        | 1757/13317 [18:06<1:58:17,  1.63it/s]
                                                      
{'loss': 0.3178, 'grad_norm': 0.6757134199142456, 'learning_rate': 9.969004950996175e-06, 'epoch': 0.4}

 13%|█▎        | 1757/13317 [18:06<1:58:17,  1.63it/s]
 13%|█▎        | 1758/13317 [18:06<1:58:13,  1.63it/s]
                                                      
{'loss': 0.2956, 'grad_norm': 0.6338570713996887, 'learning_rate': 9.968859072116199e-06, 'epoch': 0.4}

 13%|█▎        | 1758/13317 [18:07<1:58:13,  1.63it/s]
 13%|█▎        | 1759/13317 [18:07<1:58:45,  1.62it/s]
                                                      
{'loss': 0.363, 'grad_norm': 0.7795665860176086, 'learning_rate': 9.968712851822993e-06, 'epoch': 0.4}

 13%|█▎        | 1759/13317 [18:07<1:58:45,  1.62it/s]
 13%|█▎        | 1760/13317 [18:08<1:58:37,  1.62it/s]
                                                      
{'loss': 0.2713, 'grad_norm': 0.5592122077941895, 'learning_rate': 9.968566290126604e-06, 'epoch': 0.4}

 13%|█▎        | 1760/13317 [18:08<1:58:37,  1.62it/s]
 13%|█▎        | 1761/13317 [18:08<1:58:26,  1.63it/s]
                                                      
{'loss': 0.2685, 'grad_norm': 0.5745306015014648, 'learning_rate': 9.968419387037104e-06, 'epoch': 0.4}

 13%|█▎        | 1761/13317 [18:08<1:58:26,  1.63it/s]
 13%|█▎        | 1762/13317 [18:09<1:58:20,  1.63it/s]
                                                      
{'loss': 0.2971, 'grad_norm': 0.5820937156677246, 'learning_rate': 9.968272142564584e-06, 'epoch': 0.4}

 13%|█▎        | 1762/13317 [18:09<1:58:20,  1.63it/s]
 13%|█▎        | 1763/13317 [18:09<1:58:20,  1.63it/s]
                                                      
{'loss': 0.2146, 'grad_norm': 0.5206307768821716, 'learning_rate': 9.968124556719165e-06, 'epoch': 0.4}

 13%|█▎        | 1763/13317 [18:10<1:58:20,  1.63it/s]
 13%|█▎        | 1764/13317 [18:10<1:58:30,  1.62it/s]
                                                      
{'loss': 0.2793, 'grad_norm': 0.5383895039558411, 'learning_rate': 9.967976629510984e-06, 'epoch': 0.4}

 13%|█▎        | 1764/13317 [18:10<1:58:30,  1.62it/s]
 13%|█▎        | 1765/13317 [18:11<1:58:23,  1.63it/s]
                                                      
{'loss': 0.2685, 'grad_norm': 0.617028534412384, 'learning_rate': 9.967828360950208e-06, 'epoch': 0.4}

 13%|█▎        | 1765/13317 [18:11<1:58:23,  1.63it/s]
 13%|█▎        | 1766/13317 [18:11<1:58:19,  1.63it/s]
                                                      
{'loss': 0.3018, 'grad_norm': 0.6076598167419434, 'learning_rate': 9.967679751047023e-06, 'epoch': 0.4}

 13%|█▎        | 1766/13317 [18:11<1:58:19,  1.63it/s]
 13%|█▎        | 1767/13317 [18:12<1:58:23,  1.63it/s]
                                                      
{'loss': 0.304, 'grad_norm': 0.6642735004425049, 'learning_rate': 9.967530799811639e-06, 'epoch': 0.4}

 13%|█▎        | 1767/13317 [18:12<1:58:23,  1.63it/s]
 13%|█▎        | 1768/13317 [18:13<1:58:18,  1.63it/s]
                                                      
{'loss': 0.2586, 'grad_norm': 0.6698780059814453, 'learning_rate': 9.967381507254295e-06, 'epoch': 0.4}

 13%|█▎        | 1768/13317 [18:13<1:58:18,  1.63it/s]
 13%|█▎        | 1769/13317 [18:13<1:58:26,  1.62it/s]
                                                      
{'loss': 0.3512, 'grad_norm': 0.6033043265342712, 'learning_rate': 9.967231873385243e-06, 'epoch': 0.4}

 13%|█▎        | 1769/13317 [18:13<1:58:26,  1.62it/s]
 13%|█▎        | 1770/13317 [18:14<1:58:19,  1.63it/s]
                                                      
{'loss': 0.2237, 'grad_norm': 0.531278669834137, 'learning_rate': 9.96708189821477e-06, 'epoch': 0.4}

 13%|█▎        | 1770/13317 [18:14<1:58:19,  1.63it/s]
 13%|█▎        | 1771/13317 [18:14<1:58:13,  1.63it/s]
                                                      
{'loss': 0.2752, 'grad_norm': 0.615602433681488, 'learning_rate': 9.966931581753176e-06, 'epoch': 0.4}

 13%|█▎        | 1771/13317 [18:15<1:58:13,  1.63it/s]
 13%|█▎        | 1772/13317 [18:15<1:58:19,  1.63it/s]
                                                      
{'loss': 0.2567, 'grad_norm': 0.5430020689964294, 'learning_rate': 9.966780924010793e-06, 'epoch': 0.4}

 13%|█▎        | 1772/13317 [18:15<1:58:19,  1.63it/s]
 13%|█▎        | 1773/13317 [18:16<1:58:10,  1.63it/s]
                                                      
{'loss': 0.2774, 'grad_norm': 0.6034227013587952, 'learning_rate': 9.966629924997971e-06, 'epoch': 0.4}

 13%|█▎        | 1773/13317 [18:16<1:58:10,  1.63it/s]
 13%|█▎        | 1774/13317 [18:16<1:58:18,  1.63it/s]
                                                      
{'loss': 0.2292, 'grad_norm': 0.8037644624710083, 'learning_rate': 9.966478584725085e-06, 'epoch': 0.4}

 13%|█▎        | 1774/13317 [18:16<1:58:18,  1.63it/s]
 13%|█▎        | 1775/13317 [18:17<1:58:10,  1.63it/s]
                                                      
{'loss': 0.409, 'grad_norm': 0.6611003875732422, 'learning_rate': 9.966326903202535e-06, 'epoch': 0.4}

 13%|█▎        | 1775/13317 [18:17<1:58:10,  1.63it/s]
 13%|█▎        | 1776/13317 [18:17<1:58:09,  1.63it/s]
                                                      
{'loss': 0.3141, 'grad_norm': 0.6145458817481995, 'learning_rate': 9.966174880440742e-06, 'epoch': 0.4}

 13%|█▎        | 1776/13317 [18:18<1:58:09,  1.63it/s]
 13%|█▎        | 1777/13317 [18:18<1:58:04,  1.63it/s]
                                                      
{'loss': 0.3223, 'grad_norm': 0.6870476603507996, 'learning_rate': 9.966022516450153e-06, 'epoch': 0.4}

 13%|█▎        | 1777/13317 [18:18<1:58:04,  1.63it/s]
 13%|█▎        | 1778/13317 [18:19<1:58:01,  1.63it/s]
                                                      
{'loss': 0.2867, 'grad_norm': 0.6189728379249573, 'learning_rate': 9.965869811241235e-06, 'epoch': 0.4}

 13%|█▎        | 1778/13317 [18:19<1:58:01,  1.63it/s]
 13%|█▎        | 1779/13317 [18:19<1:58:00,  1.63it/s]
                                                      
{'loss': 0.3567, 'grad_norm': 0.7131606936454773, 'learning_rate': 9.965716764824482e-06, 'epoch': 0.4}

 13%|█▎        | 1779/13317 [18:19<1:58:00,  1.63it/s]
 13%|█▎        | 1780/13317 [18:20<1:57:59,  1.63it/s]
                                                      
{'loss': 0.3429, 'grad_norm': 0.7275426387786865, 'learning_rate': 9.96556337721041e-06, 'epoch': 0.4}

 13%|█▎        | 1780/13317 [18:20<1:57:59,  1.63it/s]
 13%|█▎        | 1781/13317 [18:21<1:57:58,  1.63it/s]
                                                      
{'loss': 0.256, 'grad_norm': 0.6613564491271973, 'learning_rate': 9.965409648409556e-06, 'epoch': 0.4}

 13%|█▎        | 1781/13317 [18:21<1:57:58,  1.63it/s]
 13%|█▎        | 1782/13317 [18:21<1:57:57,  1.63it/s]
                                                      
{'loss': 0.3675, 'grad_norm': 0.6641497611999512, 'learning_rate': 9.965255578432487e-06, 'epoch': 0.4}

 13%|█▎        | 1782/13317 [18:21<1:57:57,  1.63it/s]
 13%|█▎        | 1783/13317 [18:22<1:57:56,  1.63it/s]
                                                      
{'loss': 0.2731, 'grad_norm': 0.6482155919075012, 'learning_rate': 9.965101167289785e-06, 'epoch': 0.4}

 13%|█▎        | 1783/13317 [18:22<1:57:56,  1.63it/s]
 13%|█▎        | 1784/13317 [18:22<1:57:56,  1.63it/s]
                                                      
{'loss': 0.3523, 'grad_norm': 0.6616619825363159, 'learning_rate': 9.964946414992062e-06, 'epoch': 0.4}

 13%|█▎        | 1784/13317 [18:23<1:57:56,  1.63it/s]
 13%|█▎        | 1785/13317 [18:23<1:57:56,  1.63it/s]
                                                      
{'loss': 0.3063, 'grad_norm': 0.6396284699440002, 'learning_rate': 9.96479132154995e-06, 'epoch': 0.4}

 13%|█▎        | 1785/13317 [18:23<1:57:56,  1.63it/s]
 13%|█▎        | 1786/13317 [18:24<1:57:54,  1.63it/s]
                                                      
{'loss': 0.2885, 'grad_norm': 0.662377119064331, 'learning_rate': 9.964635886974107e-06, 'epoch': 0.4}

 13%|█▎        | 1786/13317 [18:24<1:57:54,  1.63it/s]
 13%|█▎        | 1787/13317 [18:24<1:57:51,  1.63it/s]
                                                      
{'loss': 0.3529, 'grad_norm': 0.771767795085907, 'learning_rate': 9.964480111275212e-06, 'epoch': 0.4}

 13%|█▎        | 1787/13317 [18:24<1:57:51,  1.63it/s]
 13%|█▎        | 1788/13317 [18:25<1:57:52,  1.63it/s]
                                                      
{'loss': 0.2723, 'grad_norm': 0.6075901389122009, 'learning_rate': 9.964323994463968e-06, 'epoch': 0.4}

 13%|█▎        | 1788/13317 [18:25<1:57:52,  1.63it/s]
 13%|█▎        | 1789/13317 [18:25<1:57:51,  1.63it/s]
                                                      
{'loss': 0.2663, 'grad_norm': 0.5817909836769104, 'learning_rate': 9.964167536551104e-06, 'epoch': 0.4}

 13%|█▎        | 1789/13317 [18:26<1:57:51,  1.63it/s]
 13%|█▎        | 1790/13317 [18:26<1:57:51,  1.63it/s]
                                                      
{'loss': 0.2421, 'grad_norm': 0.5803362727165222, 'learning_rate': 9.964010737547366e-06, 'epoch': 0.4}

 13%|█▎        | 1790/13317 [18:26<1:57:51,  1.63it/s]
 13%|█▎        | 1791/13317 [18:27<1:57:51,  1.63it/s]
                                                      
{'loss': 0.3039, 'grad_norm': 0.6862210631370544, 'learning_rate': 9.963853597463533e-06, 'epoch': 0.4}

 13%|█▎        | 1791/13317 [18:27<1:57:51,  1.63it/s]
 13%|█▎        | 1792/13317 [18:27<1:57:52,  1.63it/s]
                                                      
{'loss': 0.216, 'grad_norm': 0.6673689484596252, 'learning_rate': 9.9636961163104e-06, 'epoch': 0.4}

 13%|█▎        | 1792/13317 [18:27<1:57:52,  1.63it/s]
 13%|█▎        | 1793/13317 [18:28<1:57:49,  1.63it/s]
                                                      
{'loss': 0.3414, 'grad_norm': 0.6603010296821594, 'learning_rate': 9.963538294098785e-06, 'epoch': 0.4}

 13%|█▎        | 1793/13317 [18:28<1:57:49,  1.63it/s]
 13%|█▎        | 1794/13317 [18:29<1:57:51,  1.63it/s]
                                                      
{'loss': 0.2816, 'grad_norm': 0.6482819318771362, 'learning_rate': 9.963380130839535e-06, 'epoch': 0.4}

 13%|█▎        | 1794/13317 [18:29<1:57:51,  1.63it/s]
 13%|█▎        | 1795/13317 [18:29<1:57:47,  1.63it/s]
                                                      
{'loss': 0.2679, 'grad_norm': 0.5810615420341492, 'learning_rate': 9.963221626543517e-06, 'epoch': 0.4}

 13%|█▎        | 1795/13317 [18:29<1:57:47,  1.63it/s]
 13%|█▎        | 1796/13317 [18:30<1:57:50,  1.63it/s]
                                                      
{'loss': 0.277, 'grad_norm': 0.6098554134368896, 'learning_rate': 9.963062781221624e-06, 'epoch': 0.4}

 13%|█▎        | 1796/13317 [18:30<1:57:50,  1.63it/s]
 13%|█▎        | 1797/13317 [18:30<1:57:46,  1.63it/s]
                                                      
{'loss': 0.2406, 'grad_norm': 0.5546708703041077, 'learning_rate': 9.962903594884767e-06, 'epoch': 0.4}

 13%|█▎        | 1797/13317 [18:30<1:57:46,  1.63it/s]
 14%|█▎        | 1798/13317 [18:31<1:57:47,  1.63it/s]
                                                      
{'loss': 0.231, 'grad_norm': 0.6635652780532837, 'learning_rate': 9.962744067543883e-06, 'epoch': 0.41}

 14%|█▎        | 1798/13317 [18:31<1:57:47,  1.63it/s]
 14%|█▎        | 1799/13317 [18:32<1:57:46,  1.63it/s]
                                                      
{'loss': 0.3106, 'grad_norm': 0.6594192981719971, 'learning_rate': 9.962584199209936e-06, 'epoch': 0.41}

 14%|█▎        | 1799/13317 [18:32<1:57:46,  1.63it/s]
 14%|█▎        | 1800/13317 [18:32<1:57:51,  1.63it/s]
                                                      
{'loss': 0.2685, 'grad_norm': 0.6137346029281616, 'learning_rate': 9.962423989893913e-06, 'epoch': 0.41}

 14%|█▎        | 1800/13317 [18:32<1:57:51,  1.63it/s]
 14%|█▎        | 1801/13317 [18:33<1:57:49,  1.63it/s]
                                                      
{'loss': 0.3394, 'grad_norm': 0.7322275042533875, 'learning_rate': 9.962263439606816e-06, 'epoch': 0.41}

 14%|█▎        | 1801/13317 [18:33<1:57:49,  1.63it/s]
 14%|█▎        | 1802/13317 [18:33<1:57:44,  1.63it/s]
                                                      
{'loss': 0.237, 'grad_norm': 0.5971479415893555, 'learning_rate': 9.96210254835968e-06, 'epoch': 0.41}

 14%|█▎        | 1802/13317 [18:34<1:57:44,  1.63it/s]
 14%|█▎        | 1803/13317 [18:34<1:57:42,  1.63it/s]
                                                      
{'loss': 0.3884, 'grad_norm': 0.6792637705802917, 'learning_rate': 9.96194131616356e-06, 'epoch': 0.41}

 14%|█▎        | 1803/13317 [18:34<1:57:42,  1.63it/s]
 14%|█▎        | 1804/13317 [18:35<1:57:42,  1.63it/s]
                                                      
{'loss': 0.2364, 'grad_norm': 0.5753574371337891, 'learning_rate': 9.961779743029532e-06, 'epoch': 0.41}

 14%|█▎        | 1804/13317 [18:35<1:57:42,  1.63it/s]
 14%|█▎        | 1805/13317 [18:35<1:57:43,  1.63it/s]
                                                      
{'loss': 0.2692, 'grad_norm': 0.6289042830467224, 'learning_rate': 9.9616178289687e-06, 'epoch': 0.41}

 14%|█▎        | 1805/13317 [18:35<1:57:43,  1.63it/s]
 14%|█▎        | 1806/13317 [18:36<1:57:39,  1.63it/s]
                                                      
{'loss': 0.2494, 'grad_norm': 0.6903944611549377, 'learning_rate': 9.96145557399219e-06, 'epoch': 0.41}

 14%|█▎        | 1806/13317 [18:36<1:57:39,  1.63it/s]
 14%|█▎        | 1807/13317 [18:36<1:57:37,  1.63it/s]
                                                      
{'loss': 0.211, 'grad_norm': 0.5629047751426697, 'learning_rate': 9.961292978111147e-06, 'epoch': 0.41}

 14%|█▎        | 1807/13317 [18:37<1:57:37,  1.63it/s]
 14%|█▎        | 1808/13317 [18:37<1:57:39,  1.63it/s]
                                                      
{'loss': 0.2539, 'grad_norm': 0.5599368810653687, 'learning_rate': 9.961130041336748e-06, 'epoch': 0.41}

 14%|█▎        | 1808/13317 [18:37<1:57:39,  1.63it/s]
 14%|█▎        | 1809/13317 [18:38<1:57:40,  1.63it/s]
                                                      
{'loss': 0.2567, 'grad_norm': 0.6029070615768433, 'learning_rate': 9.960966763680184e-06, 'epoch': 0.41}

 14%|█▎        | 1809/13317 [18:38<1:57:40,  1.63it/s]
 14%|█▎        | 1810/13317 [18:38<1:57:43,  1.63it/s]
                                                      
{'loss': 0.2336, 'grad_norm': 0.5909013152122498, 'learning_rate': 9.960803145152678e-06, 'epoch': 0.41}

 14%|█▎        | 1810/13317 [18:38<1:57:43,  1.63it/s]
 14%|█▎        | 1811/13317 [18:39<1:57:47,  1.63it/s]
                                                      
{'loss': 0.2338, 'grad_norm': 0.5619499683380127, 'learning_rate': 9.960639185765468e-06, 'epoch': 0.41}

 14%|█▎        | 1811/13317 [18:39<1:57:47,  1.63it/s]
 14%|█▎        | 1812/13317 [18:40<1:57:43,  1.63it/s]
                                                      
{'loss': 0.3174, 'grad_norm': 0.6438169479370117, 'learning_rate': 9.960474885529823e-06, 'epoch': 0.41}

 14%|█▎        | 1812/13317 [18:40<1:57:43,  1.63it/s]
 14%|█▎        | 1813/13317 [18:40<1:57:39,  1.63it/s]
                                                      
{'loss': 0.3026, 'grad_norm': 0.7061589360237122, 'learning_rate': 9.96031024445703e-06, 'epoch': 0.41}

 14%|█▎        | 1813/13317 [18:40<1:57:39,  1.63it/s]
 14%|█▎        | 1814/13317 [18:41<1:57:41,  1.63it/s]
                                                      
{'loss': 0.3345, 'grad_norm': 0.6672906279563904, 'learning_rate': 9.960145262558402e-06, 'epoch': 0.41}

 14%|█▎        | 1814/13317 [18:41<1:57:41,  1.63it/s]
 14%|█▎        | 1815/13317 [18:41<1:57:39,  1.63it/s]
                                                      
{'loss': 0.2656, 'grad_norm': 0.6853861212730408, 'learning_rate': 9.959979939845277e-06, 'epoch': 0.41}

 14%|█▎        | 1815/13317 [18:42<1:57:39,  1.63it/s]
 14%|█▎        | 1816/13317 [18:42<1:57:38,  1.63it/s]
                                                      
{'loss': 0.3131, 'grad_norm': 0.6883706450462341, 'learning_rate': 9.959814276329013e-06, 'epoch': 0.41}

 14%|█▎        | 1816/13317 [18:42<1:57:38,  1.63it/s]
 14%|█▎        | 1817/13317 [18:43<1:57:36,  1.63it/s]
                                                      
{'loss': 0.226, 'grad_norm': 0.6283265948295593, 'learning_rate': 9.959648272020992e-06, 'epoch': 0.41}

 14%|█▎        | 1817/13317 [18:43<1:57:36,  1.63it/s]
 14%|█▎        | 1818/13317 [18:43<1:57:33,  1.63it/s]
                                                      
{'loss': 0.2301, 'grad_norm': 0.5008438229560852, 'learning_rate': 9.959481926932621e-06, 'epoch': 0.41}

 14%|█▎        | 1818/13317 [18:43<1:57:33,  1.63it/s]
 14%|█▎        | 1819/13317 [18:44<1:57:36,  1.63it/s]
                                                      
{'loss': 0.3604, 'grad_norm': 0.6714870929718018, 'learning_rate': 9.959315241075332e-06, 'epoch': 0.41}

 14%|█▎        | 1819/13317 [18:44<1:57:36,  1.63it/s]
 14%|█▎        | 1820/13317 [18:44<1:57:35,  1.63it/s]
                                                      
{'loss': 0.2351, 'grad_norm': 0.5685430765151978, 'learning_rate': 9.959148214460574e-06, 'epoch': 0.41}

 14%|█▎        | 1820/13317 [18:45<1:57:35,  1.63it/s]
 14%|█▎        | 1821/13317 [18:45<1:57:34,  1.63it/s]
                                                      
{'loss': 0.2455, 'grad_norm': 0.5295842289924622, 'learning_rate': 9.958980847099827e-06, 'epoch': 0.41}

 14%|█▎        | 1821/13317 [18:45<1:57:34,  1.63it/s]
 14%|█▎        | 1822/13317 [18:46<1:57:31,  1.63it/s]
                                                      
{'loss': 0.3006, 'grad_norm': 0.6370148062705994, 'learning_rate': 9.958813139004586e-06, 'epoch': 0.41}

 14%|█▎        | 1822/13317 [18:46<1:57:31,  1.63it/s]
 14%|█▎        | 1823/13317 [18:46<1:57:29,  1.63it/s]
                                                      
{'loss': 0.2168, 'grad_norm': 0.5461023449897766, 'learning_rate': 9.95864509018638e-06, 'epoch': 0.41}

 14%|█▎        | 1823/13317 [18:46<1:57:29,  1.63it/s]
 14%|█▎        | 1824/13317 [18:47<1:57:28,  1.63it/s]
                                                      
{'loss': 0.2161, 'grad_norm': 0.5541934370994568, 'learning_rate': 9.958476700656752e-06, 'epoch': 0.41}

 14%|█▎        | 1824/13317 [18:47<1:57:28,  1.63it/s]
 14%|█▎        | 1825/13317 [18:48<1:57:29,  1.63it/s]
                                                      
{'loss': 0.1957, 'grad_norm': 0.5747409462928772, 'learning_rate': 9.958307970427276e-06, 'epoch': 0.41}

 14%|█▎        | 1825/13317 [18:48<1:57:29,  1.63it/s]
 14%|█▎        | 1826/13317 [18:48<1:57:52,  1.62it/s]
                                                      
{'loss': 0.2964, 'grad_norm': 0.5930110216140747, 'learning_rate': 9.958138899509541e-06, 'epoch': 0.41}

 14%|█▎        | 1826/13317 [18:48<1:57:52,  1.62it/s]
 14%|█▎        | 1827/13317 [18:49<1:57:46,  1.63it/s]
                                                      
{'loss': 0.3863, 'grad_norm': 0.6421357989311218, 'learning_rate': 9.957969487915166e-06, 'epoch': 0.41}

 14%|█▎        | 1827/13317 [18:49<1:57:46,  1.63it/s]
 14%|█▎        | 1828/13317 [18:49<1:57:44,  1.63it/s]
                                                      
{'loss': 0.331, 'grad_norm': 0.6508557796478271, 'learning_rate': 9.957799735655793e-06, 'epoch': 0.41}

 14%|█▎        | 1828/13317 [18:50<1:57:44,  1.63it/s]
 14%|█▎        | 1829/13317 [18:50<1:57:45,  1.63it/s]
                                                      
{'loss': 0.3154, 'grad_norm': 0.6724609136581421, 'learning_rate': 9.957629642743082e-06, 'epoch': 0.41}

 14%|█▎        | 1829/13317 [18:50<1:57:45,  1.63it/s]
 14%|█▎        | 1830/13317 [18:51<1:57:35,  1.63it/s]
                                                      
{'loss': 0.2639, 'grad_norm': 0.6171687841415405, 'learning_rate': 9.957459209188722e-06, 'epoch': 0.41}

 14%|█▎        | 1830/13317 [18:51<1:57:35,  1.63it/s]
 14%|█▎        | 1831/13317 [18:51<1:57:29,  1.63it/s]
                                                      
{'loss': 0.3193, 'grad_norm': 0.716738760471344, 'learning_rate': 9.957288435004427e-06, 'epoch': 0.41}

 14%|█▎        | 1831/13317 [18:51<1:57:29,  1.63it/s]
 14%|█▍        | 1832/13317 [18:52<1:57:39,  1.63it/s]
                                                      
{'loss': 0.2955, 'grad_norm': 0.6207910180091858, 'learning_rate': 9.957117320201925e-06, 'epoch': 0.41}

 14%|█▍        | 1832/13317 [18:52<1:57:39,  1.63it/s]
 14%|█▍        | 1833/13317 [18:52<1:57:35,  1.63it/s]
                                                      
{'loss': 0.246, 'grad_norm': 0.6008695363998413, 'learning_rate': 9.956945864792978e-06, 'epoch': 0.41}

 14%|█▍        | 1833/13317 [18:53<1:57:35,  1.63it/s]
 14%|█▍        | 1834/13317 [18:53<1:58:07,  1.62it/s]
                                                      
{'loss': 0.2918, 'grad_norm': 0.5982924699783325, 'learning_rate': 9.956774068789363e-06, 'epoch': 0.41}

 14%|█▍        | 1834/13317 [18:53<1:58:07,  1.62it/s]
 14%|█▍        | 1835/13317 [18:54<1:57:59,  1.62it/s]
                                                      
{'loss': 0.2823, 'grad_norm': 0.5943489074707031, 'learning_rate': 9.956601932202888e-06, 'epoch': 0.41}

 14%|█▍        | 1835/13317 [18:54<1:57:59,  1.62it/s]
 14%|█▍        | 1836/13317 [18:54<1:58:19,  1.62it/s]
                                                      
{'loss': 0.2586, 'grad_norm': 0.5580497980117798, 'learning_rate': 9.956429455045379e-06, 'epoch': 0.41}

 14%|█▍        | 1836/13317 [18:54<1:58:19,  1.62it/s]
 14%|█▍        | 1837/13317 [18:55<1:58:11,  1.62it/s]
                                                      
{'loss': 0.2264, 'grad_norm': 0.5297930240631104, 'learning_rate': 9.956256637328687e-06, 'epoch': 0.41}

 14%|█▍        | 1837/13317 [18:55<1:58:11,  1.62it/s]
 14%|█▍        | 1838/13317 [18:56<1:57:57,  1.62it/s]
                                                      
{'loss': 0.2228, 'grad_norm': 0.5260700583457947, 'learning_rate': 9.956083479064686e-06, 'epoch': 0.41}

 14%|█▍        | 1838/13317 [18:56<1:57:57,  1.62it/s]
 14%|█▍        | 1839/13317 [18:56<1:58:25,  1.62it/s]
                                                      
{'loss': 0.3621, 'grad_norm': 0.6894851326942444, 'learning_rate': 9.955909980265274e-06, 'epoch': 0.41}

 14%|█▍        | 1839/13317 [18:56<1:58:25,  1.62it/s]
 14%|█▍        | 1840/13317 [18:57<1:58:09,  1.62it/s]
                                                      
{'loss': 0.2759, 'grad_norm': 0.555934488773346, 'learning_rate': 9.955736140942371e-06, 'epoch': 0.41}

 14%|█▍        | 1840/13317 [18:57<1:58:09,  1.62it/s]
 14%|█▍        | 1841/13317 [18:57<1:57:59,  1.62it/s]
                                                      
{'loss': 0.268, 'grad_norm': 0.5544992089271545, 'learning_rate': 9.955561961107924e-06, 'epoch': 0.41}

 14%|█▍        | 1841/13317 [18:58<1:57:59,  1.62it/s]
 14%|█▍        | 1842/13317 [18:58<1:57:55,  1.62it/s]
                                                      
{'loss': 0.2664, 'grad_norm': 0.5935534238815308, 'learning_rate': 9.955387440773902e-06, 'epoch': 0.41}

 14%|█▍        | 1842/13317 [18:58<1:57:55,  1.62it/s]
 14%|█▍        | 1843/13317 [18:59<1:57:44,  1.62it/s]
                                                      
{'loss': 0.2896, 'grad_norm': 0.5572241544723511, 'learning_rate': 9.955212579952292e-06, 'epoch': 0.42}

 14%|█▍        | 1843/13317 [18:59<1:57:44,  1.62it/s]
 14%|█▍        | 1844/13317 [18:59<1:57:56,  1.62it/s]
                                                      
{'loss': 0.2234, 'grad_norm': 0.5543793439865112, 'learning_rate': 9.955037378655112e-06, 'epoch': 0.42}

 14%|█▍        | 1844/13317 [18:59<1:57:56,  1.62it/s]
 14%|█▍        | 1845/13317 [19:00<1:57:47,  1.62it/s]
                                                      
{'loss': 0.2733, 'grad_norm': 0.6046830415725708, 'learning_rate': 9.954861836894398e-06, 'epoch': 0.42}

 14%|█▍        | 1845/13317 [19:00<1:57:47,  1.62it/s]
 14%|█▍        | 1846/13317 [19:00<1:57:43,  1.62it/s]
                                                      
{'loss': 0.7201, 'grad_norm': 0.7080265879631042, 'learning_rate': 9.954685954682216e-06, 'epoch': 0.42}

 14%|█▍        | 1846/13317 [19:01<1:57:43,  1.62it/s]
 14%|█▍        | 1847/13317 [19:01<1:57:45,  1.62it/s]
                                                      
{'loss': 0.3141, 'grad_norm': 0.603529155254364, 'learning_rate': 9.954509732030644e-06, 'epoch': 0.42}

 14%|█▍        | 1847/13317 [19:01<1:57:45,  1.62it/s]
 14%|█▍        | 1848/13317 [19:02<1:57:35,  1.63it/s]
                                                      
{'loss': 0.2363, 'grad_norm': 0.5648319721221924, 'learning_rate': 9.954333168951796e-06, 'epoch': 0.42}

 14%|█▍        | 1848/13317 [19:02<1:57:35,  1.63it/s]
 14%|█▍        | 1849/13317 [19:02<1:57:34,  1.63it/s]
                                                      
{'loss': 0.3106, 'grad_norm': 0.7018384337425232, 'learning_rate': 9.954156265457801e-06, 'epoch': 0.42}

 14%|█▍        | 1849/13317 [19:02<1:57:34,  1.63it/s]
 14%|█▍        | 1850/13317 [19:03<1:57:25,  1.63it/s]
                                                      
{'loss': 0.2913, 'grad_norm': 0.6035786271095276, 'learning_rate': 9.953979021560818e-06, 'epoch': 0.42}

 14%|█▍        | 1850/13317 [19:03<1:57:25,  1.63it/s]
 14%|█▍        | 1851/13317 [19:04<1:58:03,  1.62it/s]
                                                      
{'loss': 0.2859, 'grad_norm': 0.7239078283309937, 'learning_rate': 9.953801437273021e-06, 'epoch': 0.42}

 14%|█▍        | 1851/13317 [19:04<1:58:03,  1.62it/s]
 14%|█▍        | 1852/13317 [19:04<1:57:50,  1.62it/s]
                                                      
{'loss': 0.2754, 'grad_norm': 0.5523841977119446, 'learning_rate': 9.953623512606612e-06, 'epoch': 0.42}

 14%|█▍        | 1852/13317 [19:04<1:57:50,  1.62it/s]
 14%|█▍        | 1853/13317 [19:05<1:57:42,  1.62it/s]
                                                      
{'loss': 0.3619, 'grad_norm': 0.6143205165863037, 'learning_rate': 9.953445247573819e-06, 'epoch': 0.42}

 14%|█▍        | 1853/13317 [19:05<1:57:42,  1.62it/s]
 14%|█▍        | 1854/13317 [19:05<1:57:43,  1.62it/s]
                                                      
{'loss': 0.2664, 'grad_norm': 0.5838739275932312, 'learning_rate': 9.95326664218689e-06, 'epoch': 0.42}

 14%|█▍        | 1854/13317 [19:06<1:57:43,  1.62it/s]
 14%|█▍        | 1855/13317 [19:06<1:57:32,  1.63it/s]
                                                      
{'loss': 0.2461, 'grad_norm': 0.5812797546386719, 'learning_rate': 9.953087696458096e-06, 'epoch': 0.42}

 14%|█▍        | 1855/13317 [19:06<1:57:32,  1.63it/s]
 14%|█▍        | 1856/13317 [19:07<1:57:31,  1.63it/s]
                                                      
{'loss': 0.2648, 'grad_norm': 0.5798954963684082, 'learning_rate': 9.952908410399734e-06, 'epoch': 0.42}

 14%|█▍        | 1856/13317 [19:07<1:57:31,  1.63it/s]
 14%|█▍        | 1857/13317 [19:07<1:57:26,  1.63it/s]
                                                      
{'loss': 0.2531, 'grad_norm': 0.5247767567634583, 'learning_rate': 9.952728784024122e-06, 'epoch': 0.42}

 14%|█▍        | 1857/13317 [19:07<1:57:26,  1.63it/s]
 14%|█▍        | 1858/13317 [19:08<1:57:16,  1.63it/s]
                                                      
{'loss': 0.2954, 'grad_norm': 0.5952466130256653, 'learning_rate': 9.952548817343602e-06, 'epoch': 0.42}

 14%|█▍        | 1858/13317 [19:08<1:57:16,  1.63it/s]
 14%|█▍        | 1859/13317 [19:08<1:57:52,  1.62it/s]
                                                      
{'loss': 0.2332, 'grad_norm': 0.5360962748527527, 'learning_rate': 9.952368510370538e-06, 'epoch': 0.42}

 14%|█▍        | 1859/13317 [19:09<1:57:52,  1.62it/s]
 14%|█▍        | 1860/13317 [19:09<1:57:39,  1.62it/s]
                                                      
{'loss': 0.2059, 'grad_norm': 0.5312371850013733, 'learning_rate': 9.952187863117322e-06, 'epoch': 0.42}

 14%|█▍        | 1860/13317 [19:09<1:57:39,  1.62it/s]
 14%|█▍        | 1861/13317 [19:10<1:57:37,  1.62it/s]
                                                      
{'loss': 0.3228, 'grad_norm': 0.8153566122055054, 'learning_rate': 9.952006875596367e-06, 'epoch': 0.42}

 14%|█▍        | 1861/13317 [19:10<1:57:37,  1.62it/s]
 14%|█▍        | 1862/13317 [19:10<1:57:42,  1.62it/s]
                                                      
{'loss': 0.2072, 'grad_norm': 0.5757837891578674, 'learning_rate': 9.951825547820104e-06, 'epoch': 0.42}

 14%|█▍        | 1862/13317 [19:10<1:57:42,  1.62it/s]
 14%|█▍        | 1863/13317 [19:11<1:57:29,  1.62it/s]
                                                      
{'loss': 0.2616, 'grad_norm': 0.6155881881713867, 'learning_rate': 9.951643879800997e-06, 'epoch': 0.42}

 14%|█▍        | 1863/13317 [19:11<1:57:29,  1.62it/s]
 14%|█▍        | 1864/13317 [19:12<1:57:21,  1.63it/s]
                                                      
{'loss': 0.2644, 'grad_norm': 0.5571503043174744, 'learning_rate': 9.951461871551527e-06, 'epoch': 0.42}

 14%|█▍        | 1864/13317 [19:12<1:57:21,  1.63it/s]
 14%|█▍        | 1865/13317 [19:12<1:57:16,  1.63it/s]
                                                      
{'loss': 0.2623, 'grad_norm': 0.6214870810508728, 'learning_rate': 9.951279523084199e-06, 'epoch': 0.42}

 14%|█▍        | 1865/13317 [19:12<1:57:16,  1.63it/s]
 14%|█▍        | 1866/13317 [19:13<1:57:14,  1.63it/s]
                                                      
{'loss': 0.3376, 'grad_norm': 0.6624147295951843, 'learning_rate': 9.951096834411542e-06, 'epoch': 0.42}

 14%|█▍        | 1866/13317 [19:13<1:57:14,  1.63it/s]
 14%|█▍        | 1867/13317 [19:13<1:57:13,  1.63it/s]
                                                      
{'loss': 0.2822, 'grad_norm': 0.6577680110931396, 'learning_rate': 9.950913805546109e-06, 'epoch': 0.42}

 14%|█▍        | 1867/13317 [19:14<1:57:13,  1.63it/s]
 14%|█▍        | 1868/13317 [19:14<1:57:07,  1.63it/s]
                                                      
{'loss': 0.2165, 'grad_norm': 0.5633938908576965, 'learning_rate': 9.950730436500477e-06, 'epoch': 0.42}

 14%|█▍        | 1868/13317 [19:14<1:57:07,  1.63it/s]
 14%|█▍        | 1869/13317 [19:15<1:57:02,  1.63it/s]
                                                      
{'loss': 0.2958, 'grad_norm': 0.6383453011512756, 'learning_rate': 9.950546727287247e-06, 'epoch': 0.42}

 14%|█▍        | 1869/13317 [19:15<1:57:02,  1.63it/s]
 14%|█▍        | 1870/13317 [19:15<1:57:01,  1.63it/s]
                                                      
{'loss': 0.3497, 'grad_norm': 0.5879560112953186, 'learning_rate': 9.950362677919036e-06, 'epoch': 0.42}

 14%|█▍        | 1870/13317 [19:15<1:57:01,  1.63it/s]
 14%|█▍        | 1871/13317 [19:16<1:57:11,  1.63it/s]
                                                      
{'loss': 0.2197, 'grad_norm': 0.6106364130973816, 'learning_rate': 9.950178288408496e-06, 'epoch': 0.42}

 14%|█▍        | 1871/13317 [19:16<1:57:11,  1.63it/s]
 14%|█▍        | 1872/13317 [19:16<1:57:10,  1.63it/s]
                                                      
{'loss': 0.2994, 'grad_norm': 0.8491281270980835, 'learning_rate': 9.949993558768295e-06, 'epoch': 0.42}

 14%|█▍        | 1872/13317 [19:17<1:57:10,  1.63it/s]
 14%|█▍        | 1873/13317 [19:17<1:57:07,  1.63it/s]
                                                      
{'loss': 0.2356, 'grad_norm': 0.576330840587616, 'learning_rate': 9.949808489011124e-06, 'epoch': 0.42}

 14%|█▍        | 1873/13317 [19:17<1:57:07,  1.63it/s]
 14%|█▍        | 1874/13317 [19:18<1:57:03,  1.63it/s]
                                                      
{'loss': 0.3439, 'grad_norm': 0.7051506638526917, 'learning_rate': 9.9496230791497e-06, 'epoch': 0.42}

 14%|█▍        | 1874/13317 [19:18<1:57:03,  1.63it/s]
 14%|█▍        | 1875/13317 [19:18<1:57:00,  1.63it/s]
                                                      
{'loss': 0.3007, 'grad_norm': 0.6525668501853943, 'learning_rate': 9.949437329196763e-06, 'epoch': 0.42}

 14%|█▍        | 1875/13317 [19:18<1:57:00,  1.63it/s]
 14%|█▍        | 1876/13317 [19:19<1:57:08,  1.63it/s]
                                                      
{'loss': 0.3419, 'grad_norm': 0.6805566549301147, 'learning_rate': 9.949251239165076e-06, 'epoch': 0.42}

 14%|█▍        | 1876/13317 [19:19<1:57:08,  1.63it/s]
 14%|█▍        | 1877/13317 [19:20<1:57:03,  1.63it/s]
                                                      
{'loss': 0.2343, 'grad_norm': 0.5262712836265564, 'learning_rate': 9.949064809067425e-06, 'epoch': 0.42}

 14%|█▍        | 1877/13317 [19:20<1:57:03,  1.63it/s]
 14%|█▍        | 1878/13317 [19:20<1:57:05,  1.63it/s]
                                                      
{'loss': 0.2941, 'grad_norm': 0.6509297490119934, 'learning_rate': 9.948878038916623e-06, 'epoch': 0.42}

 14%|█▍        | 1878/13317 [19:20<1:57:05,  1.63it/s]
 14%|█▍        | 1879/13317 [19:21<1:56:59,  1.63it/s]
                                                      
{'loss': 0.4335, 'grad_norm': 0.7457350492477417, 'learning_rate': 9.948690928725497e-06, 'epoch': 0.42}

 14%|█▍        | 1879/13317 [19:21<1:56:59,  1.63it/s]
 14%|█▍        | 1880/13317 [19:21<1:56:57,  1.63it/s]
                                                      
{'loss': 0.2553, 'grad_norm': 0.6094668507575989, 'learning_rate': 9.948503478506908e-06, 'epoch': 0.42}

 14%|█▍        | 1880/13317 [19:22<1:56:57,  1.63it/s]
 14%|█▍        | 1881/13317 [19:22<1:56:55,  1.63it/s]
                                                      
{'loss': 0.3216, 'grad_norm': 0.6785441040992737, 'learning_rate': 9.948315688273733e-06, 'epoch': 0.42}

 14%|█▍        | 1881/13317 [19:22<1:56:55,  1.63it/s]
 14%|█▍        | 1882/13317 [19:23<1:56:55,  1.63it/s]
                                                      
{'loss': 0.2434, 'grad_norm': 0.5729135274887085, 'learning_rate': 9.948127558038878e-06, 'epoch': 0.42}

 14%|█▍        | 1882/13317 [19:23<1:56:55,  1.63it/s]
 14%|█▍        | 1883/13317 [19:23<1:56:56,  1.63it/s]
                                                      
{'loss': 0.2148, 'grad_norm': 0.461558997631073, 'learning_rate': 9.947939087815268e-06, 'epoch': 0.42}

 14%|█▍        | 1883/13317 [19:23<1:56:56,  1.63it/s]
 14%|█▍        | 1884/13317 [19:24<1:56:54,  1.63it/s]
                                                      
{'loss': 0.2868, 'grad_norm': 0.5643543004989624, 'learning_rate': 9.947750277615853e-06, 'epoch': 0.42}

 14%|█▍        | 1884/13317 [19:24<1:56:54,  1.63it/s]
 14%|█▍        | 1885/13317 [19:24<1:56:58,  1.63it/s]
                                                      
{'loss': 0.2086, 'grad_norm': 0.49859923124313354, 'learning_rate': 9.947561127453606e-06, 'epoch': 0.42}

 14%|█▍        | 1885/13317 [19:25<1:56:58,  1.63it/s]
 14%|█▍        | 1886/13317 [19:25<1:56:54,  1.63it/s]
                                                      
{'loss': 0.3051, 'grad_norm': 0.6435400247573853, 'learning_rate': 9.947371637341523e-06, 'epoch': 0.42}

 14%|█▍        | 1886/13317 [19:25<1:56:54,  1.63it/s]
 14%|█▍        | 1887/13317 [19:26<1:56:51,  1.63it/s]
                                                      
{'loss': 0.2962, 'grad_norm': 0.5313014388084412, 'learning_rate': 9.947181807292626e-06, 'epoch': 0.43}

 14%|█▍        | 1887/13317 [19:26<1:56:51,  1.63it/s]
 14%|█▍        | 1888/13317 [19:26<1:56:50,  1.63it/s]
                                                      
{'loss': 0.2789, 'grad_norm': 0.5513039231300354, 'learning_rate': 9.946991637319958e-06, 'epoch': 0.43}

 14%|█▍        | 1888/13317 [19:26<1:56:50,  1.63it/s]
 14%|█▍        | 1889/13317 [19:27<1:56:49,  1.63it/s]
                                                      
{'loss': 0.298, 'grad_norm': 0.5980415940284729, 'learning_rate': 9.946801127436583e-06, 'epoch': 0.43}

 14%|█▍        | 1889/13317 [19:27<1:56:49,  1.63it/s]
 14%|█▍        | 1890/13317 [19:28<1:56:48,  1.63it/s]
                                                      
{'loss': 0.2621, 'grad_norm': 0.5491133332252502, 'learning_rate': 9.946610277655596e-06, 'epoch': 0.43}

 14%|█▍        | 1890/13317 [19:28<1:56:48,  1.63it/s]
 14%|█▍        | 1891/13317 [19:28<1:56:48,  1.63it/s]
                                                      
{'loss': 0.2778, 'grad_norm': 0.5640999674797058, 'learning_rate': 9.946419087990105e-06, 'epoch': 0.43}

 14%|█▍        | 1891/13317 [19:28<1:56:48,  1.63it/s]
 14%|█▍        | 1892/13317 [19:29<1:56:48,  1.63it/s]
                                                      
{'loss': 0.2274, 'grad_norm': 0.5903883576393127, 'learning_rate': 9.946227558453249e-06, 'epoch': 0.43}

 14%|█▍        | 1892/13317 [19:29<1:56:48,  1.63it/s]
 14%|█▍        | 1893/13317 [19:29<1:56:49,  1.63it/s]
                                                      
{'loss': 0.2961, 'grad_norm': 0.6198226809501648, 'learning_rate': 9.946035689058189e-06, 'epoch': 0.43}

 14%|█▍        | 1893/13317 [19:29<1:56:49,  1.63it/s]
 14%|█▍        | 1894/13317 [19:30<1:56:53,  1.63it/s]
                                                      
{'loss': 0.2095, 'grad_norm': 0.45424899458885193, 'learning_rate': 9.945843479818106e-06, 'epoch': 0.43}

 14%|█▍        | 1894/13317 [19:30<1:56:53,  1.63it/s]
 14%|█▍        | 1895/13317 [19:31<1:56:57,  1.63it/s]
                                                      
{'loss': 0.4663, 'grad_norm': 0.7974310517311096, 'learning_rate': 9.945650930746211e-06, 'epoch': 0.43}

 14%|█▍        | 1895/13317 [19:31<1:56:57,  1.63it/s]
 14%|█▍        | 1896/13317 [19:31<1:57:00,  1.63it/s]
                                                      
{'loss': 0.4083, 'grad_norm': 0.6792088150978088, 'learning_rate': 9.945458041855732e-06, 'epoch': 0.43}

 14%|█▍        | 1896/13317 [19:31<1:57:00,  1.63it/s]
 14%|█▍        | 1897/13317 [19:32<1:56:57,  1.63it/s]
                                                      
{'loss': 0.2151, 'grad_norm': 0.5081570148468018, 'learning_rate': 9.94526481315992e-06, 'epoch': 0.43}

 14%|█▍        | 1897/13317 [19:32<1:56:57,  1.63it/s]
 14%|█▍        | 1898/13317 [19:32<1:56:51,  1.63it/s]
                                                      
{'loss': 0.2628, 'grad_norm': 0.5649418234825134, 'learning_rate': 9.945071244672055e-06, 'epoch': 0.43}

 14%|█▍        | 1898/13317 [19:33<1:56:51,  1.63it/s]
 14%|█▍        | 1899/13317 [19:33<1:56:52,  1.63it/s]
                                                      
{'loss': 0.2467, 'grad_norm': 0.6111924648284912, 'learning_rate': 9.944877336405436e-06, 'epoch': 0.43}

 14%|█▍        | 1899/13317 [19:33<1:56:52,  1.63it/s]
 14%|█▍        | 1900/13317 [19:34<1:56:52,  1.63it/s]
                                                      
{'loss': 0.3425, 'grad_norm': 0.7352547645568848, 'learning_rate': 9.944683088373385e-06, 'epoch': 0.43}

 14%|█▍        | 1900/13317 [19:34<1:56:52,  1.63it/s]
 14%|█▍        | 1901/13317 [19:34<1:56:50,  1.63it/s]
                                                      
{'loss': 0.2651, 'grad_norm': 0.5441684126853943, 'learning_rate': 9.944488500589253e-06, 'epoch': 0.43}

 14%|█▍        | 1901/13317 [19:34<1:56:50,  1.63it/s]
 14%|█▍        | 1902/13317 [19:35<1:56:57,  1.63it/s]
                                                      
{'loss': 0.3831, 'grad_norm': 0.7112727165222168, 'learning_rate': 9.944293573066408e-06, 'epoch': 0.43}

 14%|█▍        | 1902/13317 [19:35<1:56:57,  1.63it/s]
 14%|█▍        | 1903/13317 [19:35<1:56:55,  1.63it/s]
                                                      
{'loss': 0.2758, 'grad_norm': 0.6316356062889099, 'learning_rate': 9.944098305818243e-06, 'epoch': 0.43}

 14%|█▍        | 1903/13317 [19:36<1:56:55,  1.63it/s]
 14%|█▍        | 1904/13317 [19:36<1:56:49,  1.63it/s]
                                                      
{'loss': 0.3254, 'grad_norm': 0.7580543160438538, 'learning_rate': 9.943902698858177e-06, 'epoch': 0.43}

 14%|█▍        | 1904/13317 [19:36<1:56:49,  1.63it/s]
 14%|█▍        | 1905/13317 [19:37<1:56:45,  1.63it/s]
                                                      
{'loss': 0.2229, 'grad_norm': 0.5722782015800476, 'learning_rate': 9.943706752199647e-06, 'epoch': 0.43}

 14%|█▍        | 1905/13317 [19:37<1:56:45,  1.63it/s]
 14%|█▍        | 1906/13317 [19:37<1:56:43,  1.63it/s]
                                                      
{'loss': 0.241, 'grad_norm': 0.6591605544090271, 'learning_rate': 9.943510465856118e-06, 'epoch': 0.43}

 14%|█▍        | 1906/13317 [19:37<1:56:43,  1.63it/s]
 14%|█▍        | 1907/13317 [19:38<1:56:47,  1.63it/s]
                                                      
{'loss': 0.3392, 'grad_norm': 0.7256247997283936, 'learning_rate': 9.943313839841078e-06, 'epoch': 0.43}

 14%|█▍        | 1907/13317 [19:38<1:56:47,  1.63it/s]
 14%|█▍        | 1908/13317 [19:39<1:56:47,  1.63it/s]
                                                      
{'loss': 0.2844, 'grad_norm': 0.6416723132133484, 'learning_rate': 9.943116874168036e-06, 'epoch': 0.43}

 14%|█▍        | 1908/13317 [19:39<1:56:47,  1.63it/s]
 14%|█▍        | 1909/13317 [19:39<1:56:42,  1.63it/s]
                                                      
{'loss': 0.1936, 'grad_norm': 0.5214923024177551, 'learning_rate': 9.942919568850527e-06, 'epoch': 0.43}

 14%|█▍        | 1909/13317 [19:39<1:56:42,  1.63it/s]
 14%|█▍        | 1910/13317 [19:40<1:56:44,  1.63it/s]
                                                      
{'loss': 0.2853, 'grad_norm': 0.6103485822677612, 'learning_rate': 9.942721923902105e-06, 'epoch': 0.43}

 14%|█▍        | 1910/13317 [19:40<1:56:44,  1.63it/s]
 14%|█▍        | 1911/13317 [19:40<1:56:40,  1.63it/s]
                                                      
{'loss': 0.254, 'grad_norm': 0.5224648118019104, 'learning_rate': 9.942523939336355e-06, 'epoch': 0.43}

 14%|█▍        | 1911/13317 [19:41<1:56:40,  1.63it/s]
 14%|█▍        | 1912/13317 [19:41<1:56:51,  1.63it/s]
                                                      
{'loss': 0.2687, 'grad_norm': 0.6135900616645813, 'learning_rate': 9.942325615166876e-06, 'epoch': 0.43}

 14%|█▍        | 1912/13317 [19:41<1:56:51,  1.63it/s]
 14%|█▍        | 1913/13317 [19:42<1:56:47,  1.63it/s]
                                                      
{'loss': 0.3042, 'grad_norm': 0.6109628081321716, 'learning_rate': 9.9421269514073e-06, 'epoch': 0.43}

 14%|█▍        | 1913/13317 [19:42<1:56:47,  1.63it/s]
 14%|█▍        | 1914/13317 [19:42<1:56:44,  1.63it/s]
                                                      
{'loss': 0.2622, 'grad_norm': 0.5767208933830261, 'learning_rate': 9.941927948071273e-06, 'epoch': 0.43}

 14%|█▍        | 1914/13317 [19:42<1:56:44,  1.63it/s]
 14%|█▍        | 1915/13317 [19:43<1:56:40,  1.63it/s]
                                                      
{'loss': 0.3115, 'grad_norm': 0.6362483501434326, 'learning_rate': 9.941728605172469e-06, 'epoch': 0.43}

 14%|█▍        | 1915/13317 [19:43<1:56:40,  1.63it/s]
 14%|█▍        | 1916/13317 [19:43<1:56:36,  1.63it/s]
                                                      
{'loss': 0.3221, 'grad_norm': 0.7455350160598755, 'learning_rate': 9.941528922724585e-06, 'epoch': 0.43}

 14%|█▍        | 1916/13317 [19:44<1:56:36,  1.63it/s]
 14%|█▍        | 1917/13317 [19:44<1:56:32,  1.63it/s]
                                                      
{'loss': 0.3254, 'grad_norm': 0.703851044178009, 'learning_rate': 9.941328900741345e-06, 'epoch': 0.43}

 14%|█▍        | 1917/13317 [19:44<1:56:32,  1.63it/s]
 14%|█▍        | 1918/13317 [19:45<1:56:34,  1.63it/s]
                                                      
{'loss': 0.2653, 'grad_norm': 0.5348911881446838, 'learning_rate': 9.941128539236489e-06, 'epoch': 0.43}

 14%|█▍        | 1918/13317 [19:45<1:56:34,  1.63it/s]
 14%|█▍        | 1919/13317 [19:45<1:56:33,  1.63it/s]
                                                      
{'loss': 0.2834, 'grad_norm': 0.6112419962882996, 'learning_rate': 9.940927838223783e-06, 'epoch': 0.43}

 14%|█▍        | 1919/13317 [19:45<1:56:33,  1.63it/s]
 14%|█▍        | 1920/13317 [19:46<1:56:33,  1.63it/s]
                                                      
{'loss': 0.2266, 'grad_norm': 0.5933027863502502, 'learning_rate': 9.940726797717021e-06, 'epoch': 0.43}

 14%|█▍        | 1920/13317 [19:46<1:56:33,  1.63it/s]
 14%|█▍        | 1921/13317 [19:47<1:56:31,  1.63it/s]
                                                      
{'loss': 0.2553, 'grad_norm': 0.654266357421875, 'learning_rate': 9.940525417730016e-06, 'epoch': 0.43}

 14%|█▍        | 1921/13317 [19:47<1:56:31,  1.63it/s]
 14%|█▍        | 1922/13317 [19:47<1:56:39,  1.63it/s]
                                                      
{'loss': 0.2941, 'grad_norm': 0.6782610416412354, 'learning_rate': 9.940323698276601e-06, 'epoch': 0.43}

 14%|█▍        | 1922/13317 [19:47<1:56:39,  1.63it/s]
 14%|█▍        | 1923/13317 [19:48<1:56:33,  1.63it/s]
                                                      
{'loss': 0.248, 'grad_norm': 0.6376633048057556, 'learning_rate': 9.94012163937064e-06, 'epoch': 0.43}

 14%|█▍        | 1923/13317 [19:48<1:56:33,  1.63it/s]
 14%|█▍        | 1924/13317 [19:48<1:56:31,  1.63it/s]
                                                      
{'loss': 0.3098, 'grad_norm': 0.7693906426429749, 'learning_rate': 9.939919241026017e-06, 'epoch': 0.43}

 14%|█▍        | 1924/13317 [19:49<1:56:31,  1.63it/s]
 14%|█▍        | 1925/13317 [19:49<1:56:36,  1.63it/s]
                                                      
{'loss': 0.291, 'grad_norm': 0.5399231910705566, 'learning_rate': 9.939716503256636e-06, 'epoch': 0.43}

 14%|█▍        | 1925/13317 [19:49<1:56:36,  1.63it/s]
 14%|█▍        | 1926/13317 [19:50<1:56:36,  1.63it/s]
                                                      
{'loss': 0.2699, 'grad_norm': 0.6243401765823364, 'learning_rate': 9.939513426076428e-06, 'epoch': 0.43}

 14%|█▍        | 1926/13317 [19:50<1:56:36,  1.63it/s]
 14%|█▍        | 1927/13317 [19:50<1:56:31,  1.63it/s]
                                                      
{'loss': 0.2511, 'grad_norm': 0.475876122713089, 'learning_rate': 9.939310009499348e-06, 'epoch': 0.43}

 14%|█▍        | 1927/13317 [19:50<1:56:31,  1.63it/s]
 14%|█▍        | 1928/13317 [19:51<1:56:30,  1.63it/s]
                                                      
{'loss': 0.2871, 'grad_norm': 0.5676286816596985, 'learning_rate': 9.939106253539373e-06, 'epoch': 0.43}

 14%|█▍        | 1928/13317 [19:51<1:56:30,  1.63it/s]
 14%|█▍        | 1929/13317 [19:51<1:56:28,  1.63it/s]
                                                      
{'loss': 0.4238, 'grad_norm': 0.6261544227600098, 'learning_rate': 9.9389021582105e-06, 'epoch': 0.43}

 14%|█▍        | 1929/13317 [19:52<1:56:28,  1.63it/s]
 14%|█▍        | 1930/13317 [19:52<1:56:28,  1.63it/s]
                                                      
{'loss': 0.2257, 'grad_norm': 0.6121737360954285, 'learning_rate': 9.938697723526757e-06, 'epoch': 0.43}

 14%|█▍        | 1930/13317 [19:52<1:56:28,  1.63it/s]
 15%|█▍        | 1931/13317 [19:53<1:56:26,  1.63it/s]
                                                      
{'loss': 0.3176, 'grad_norm': 0.6844605803489685, 'learning_rate': 9.938492949502187e-06, 'epoch': 0.44}

 15%|█▍        | 1931/13317 [19:53<1:56:26,  1.63it/s]
 15%|█▍        | 1932/13317 [19:53<1:56:33,  1.63it/s]
                                                      
{'loss': 0.2568, 'grad_norm': 0.6231881976127625, 'learning_rate': 9.938287836150863e-06, 'epoch': 0.44}

 15%|█▍        | 1932/13317 [19:53<1:56:33,  1.63it/s]
 15%|█▍        | 1933/13317 [19:54<1:56:25,  1.63it/s]
                                                      
{'loss': 0.243, 'grad_norm': 0.5433745980262756, 'learning_rate': 9.938082383486877e-06, 'epoch': 0.44}

 15%|█▍        | 1933/13317 [19:54<1:56:25,  1.63it/s]
 15%|█▍        | 1934/13317 [19:55<1:56:23,  1.63it/s]
                                                      
{'loss': 0.2281, 'grad_norm': 0.5823758244514465, 'learning_rate': 9.937876591524345e-06, 'epoch': 0.44}

 15%|█▍        | 1934/13317 [19:55<1:56:23,  1.63it/s]
 15%|█▍        | 1935/13317 [19:55<1:56:21,  1.63it/s]
                                                      
{'loss': 0.3815, 'grad_norm': 0.6324708461761475, 'learning_rate': 9.937670460277408e-06, 'epoch': 0.44}

 15%|█▍        | 1935/13317 [19:55<1:56:21,  1.63it/s]
 15%|█▍        | 1936/13317 [19:56<1:56:18,  1.63it/s]
                                                      
{'loss': 0.2608, 'grad_norm': 0.539962112903595, 'learning_rate': 9.937463989760231e-06, 'epoch': 0.44}

 15%|█▍        | 1936/13317 [19:56<1:56:18,  1.63it/s]
 15%|█▍        | 1937/13317 [19:56<1:56:21,  1.63it/s]
                                                      
{'loss': 0.2496, 'grad_norm': 0.5822400450706482, 'learning_rate': 9.937257179987e-06, 'epoch': 0.44}

 15%|█▍        | 1937/13317 [19:56<1:56:21,  1.63it/s]
 15%|█▍        | 1938/13317 [19:57<1:56:24,  1.63it/s]
                                                      
{'loss': 0.2926, 'grad_norm': 0.6294212341308594, 'learning_rate': 9.937050030971922e-06, 'epoch': 0.44}

 15%|█▍        | 1938/13317 [19:57<1:56:24,  1.63it/s]
 15%|█▍        | 1939/13317 [19:58<1:56:26,  1.63it/s]
                                                      
{'loss': 0.2455, 'grad_norm': 0.5583479404449463, 'learning_rate': 9.936842542729233e-06, 'epoch': 0.44}

 15%|█▍        | 1939/13317 [19:58<1:56:26,  1.63it/s]
 15%|█▍        | 1940/13317 [19:58<1:56:22,  1.63it/s]
                                                      
{'loss': 0.3701, 'grad_norm': 0.65872722864151, 'learning_rate': 9.93663471527319e-06, 'epoch': 0.44}

 15%|█▍        | 1940/13317 [19:58<1:56:22,  1.63it/s]
 15%|█▍        | 1941/13317 [19:59<1:56:21,  1.63it/s]
                                                      
{'loss': 0.2728, 'grad_norm': 0.6092521548271179, 'learning_rate': 9.936426548618071e-06, 'epoch': 0.44}

 15%|█▍        | 1941/13317 [19:59<1:56:21,  1.63it/s]
 15%|█▍        | 1942/13317 [19:59<1:56:18,  1.63it/s]
                                                      
{'loss': 0.2498, 'grad_norm': 0.5526002049446106, 'learning_rate': 9.93621804277818e-06, 'epoch': 0.44}

 15%|█▍        | 1942/13317 [20:00<1:56:18,  1.63it/s]
 15%|█▍        | 1943/13317 [20:00<1:56:18,  1.63it/s]
                                                      
{'loss': 0.2229, 'grad_norm': 0.6137045621871948, 'learning_rate': 9.936009197767847e-06, 'epoch': 0.44}

 15%|█▍        | 1943/13317 [20:00<1:56:18,  1.63it/s]
 15%|█▍        | 1944/13317 [20:01<1:56:18,  1.63it/s]
                                                      
{'loss': 0.2644, 'grad_norm': 0.5807635188102722, 'learning_rate': 9.935800013601415e-06, 'epoch': 0.44}

 15%|█▍        | 1944/13317 [20:01<1:56:18,  1.63it/s]
 15%|█▍        | 1945/13317 [20:01<1:56:14,  1.63it/s]
                                                      
{'loss': 0.3819, 'grad_norm': 0.6720389723777771, 'learning_rate': 9.935590490293263e-06, 'epoch': 0.44}

 15%|█▍        | 1945/13317 [20:01<1:56:14,  1.63it/s]
 15%|█▍        | 1946/13317 [20:02<1:56:10,  1.63it/s]
                                                      
{'loss': 0.2684, 'grad_norm': 0.5838550329208374, 'learning_rate': 9.935380627857785e-06, 'epoch': 0.44}

 15%|█▍        | 1946/13317 [20:02<1:56:10,  1.63it/s]
 15%|█▍        | 1947/13317 [20:02<1:56:11,  1.63it/s]
                                                      
{'loss': 0.3046, 'grad_norm': 0.6823102235794067, 'learning_rate': 9.935170426309401e-06, 'epoch': 0.44}

 15%|█▍        | 1947/13317 [20:03<1:56:11,  1.63it/s]
 15%|█▍        | 1948/13317 [20:03<1:56:09,  1.63it/s]
                                                      
{'loss': 0.3266, 'grad_norm': 0.6552139520645142, 'learning_rate': 9.934959885662556e-06, 'epoch': 0.44}

 15%|█▍        | 1948/13317 [20:03<1:56:09,  1.63it/s]
 15%|█▍        | 1949/13317 [20:04<1:56:09,  1.63it/s]
                                                      
{'loss': 0.3083, 'grad_norm': 0.6557247042655945, 'learning_rate': 9.934749005931711e-06, 'epoch': 0.44}

 15%|█▍        | 1949/13317 [20:04<1:56:09,  1.63it/s]
 15%|█▍        | 1950/13317 [20:04<1:56:12,  1.63it/s]
                                                      
{'loss': 0.3088, 'grad_norm': 0.6213516592979431, 'learning_rate': 9.934537787131364e-06, 'epoch': 0.44}

 15%|█▍        | 1950/13317 [20:04<1:56:12,  1.63it/s]
 15%|█▍        | 1951/13317 [20:05<1:56:11,  1.63it/s]
                                                      
{'loss': 0.3276, 'grad_norm': 0.5924139022827148, 'learning_rate': 9.934326229276022e-06, 'epoch': 0.44}

 15%|█▍        | 1951/13317 [20:05<1:56:11,  1.63it/s]
 15%|█▍        | 1952/13317 [20:06<1:56:11,  1.63it/s]
                                                      
{'loss': 0.3474, 'grad_norm': 0.6113287806510925, 'learning_rate': 9.93411433238022e-06, 'epoch': 0.44}

 15%|█▍        | 1952/13317 [20:06<1:56:11,  1.63it/s]
 15%|█▍        | 1953/13317 [20:06<1:56:17,  1.63it/s]
                                                      
{'loss': 0.2795, 'grad_norm': 0.5744065642356873, 'learning_rate': 9.933902096458522e-06, 'epoch': 0.44}

 15%|█▍        | 1953/13317 [20:06<1:56:17,  1.63it/s]
 15%|█▍        | 1954/13317 [20:07<1:56:16,  1.63it/s]
                                                      
{'loss': 0.3466, 'grad_norm': 0.654914379119873, 'learning_rate': 9.933689521525508e-06, 'epoch': 0.44}

 15%|█▍        | 1954/13317 [20:07<1:56:16,  1.63it/s]
 15%|█▍        | 1955/13317 [20:07<1:56:09,  1.63it/s]
                                                      
{'loss': 0.2773, 'grad_norm': 0.5840253233909607, 'learning_rate': 9.933476607595786e-06, 'epoch': 0.44}

 15%|█▍        | 1955/13317 [20:08<1:56:09,  1.63it/s]
 15%|█▍        | 1956/13317 [20:08<1:56:11,  1.63it/s]
                                                      
{'loss': 0.2619, 'grad_norm': 0.615969181060791, 'learning_rate': 9.933263354683983e-06, 'epoch': 0.44}

 15%|█▍        | 1956/13317 [20:08<1:56:11,  1.63it/s]
 15%|█▍        | 1957/13317 [20:09<1:56:11,  1.63it/s]
                                                      
{'loss': 0.2795, 'grad_norm': 0.6504784822463989, 'learning_rate': 9.933049762804757e-06, 'epoch': 0.44}

 15%|█▍        | 1957/13317 [20:09<1:56:11,  1.63it/s]
 15%|█▍        | 1958/13317 [20:09<1:56:09,  1.63it/s]
                                                      
{'loss': 0.2744, 'grad_norm': 0.5965577363967896, 'learning_rate': 9.932835831972778e-06, 'epoch': 0.44}

 15%|█▍        | 1958/13317 [20:09<1:56:09,  1.63it/s]
 15%|█▍        | 1959/13317 [20:10<1:56:06,  1.63it/s]
                                                      
{'loss': 0.2005, 'grad_norm': 0.4859795868396759, 'learning_rate': 9.932621562202746e-06, 'epoch': 0.44}

 15%|█▍        | 1959/13317 [20:10<1:56:06,  1.63it/s]
 15%|█▍        | 1960/13317 [20:10<1:56:06,  1.63it/s]
                                                      
{'loss': 0.507, 'grad_norm': 0.7525908946990967, 'learning_rate': 9.932406953509388e-06, 'epoch': 0.44}

 15%|█▍        | 1960/13317 [20:11<1:56:06,  1.63it/s]
 15%|█▍        | 1961/13317 [20:11<1:56:03,  1.63it/s]
                                                      
{'loss': 0.2958, 'grad_norm': 0.5851914882659912, 'learning_rate': 9.932192005907446e-06, 'epoch': 0.44}

 15%|█▍        | 1961/13317 [20:11<1:56:03,  1.63it/s]
 15%|█▍        | 1962/13317 [20:12<1:56:02,  1.63it/s]
                                                      
{'loss': 0.2414, 'grad_norm': 0.6183757781982422, 'learning_rate': 9.93197671941169e-06, 'epoch': 0.44}

 15%|█▍        | 1962/13317 [20:12<1:56:02,  1.63it/s]
 15%|█▍        | 1963/13317 [20:12<1:56:02,  1.63it/s]
                                                      
{'loss': 0.2557, 'grad_norm': 0.5506025552749634, 'learning_rate': 9.931761094036912e-06, 'epoch': 0.44}

 15%|█▍        | 1963/13317 [20:12<1:56:02,  1.63it/s]
 15%|█▍        | 1964/13317 [20:13<1:55:58,  1.63it/s]
                                                      
{'loss': 0.2784, 'grad_norm': 0.5543888807296753, 'learning_rate': 9.93154512979793e-06, 'epoch': 0.44}

 15%|█▍        | 1964/13317 [20:13<1:55:58,  1.63it/s]
 15%|█▍        | 1965/13317 [20:14<1:56:02,  1.63it/s]
                                                      
{'loss': 0.3093, 'grad_norm': 0.6204724311828613, 'learning_rate': 9.93132882670958e-06, 'epoch': 0.44}

 15%|█▍        | 1965/13317 [20:14<1:56:02,  1.63it/s]
 15%|█▍        | 1966/13317 [20:14<1:56:01,  1.63it/s]
                                                      
{'loss': 0.3118, 'grad_norm': 0.6136650443077087, 'learning_rate': 9.931112184786729e-06, 'epoch': 0.44}

 15%|█▍        | 1966/13317 [20:14<1:56:01,  1.63it/s]
 15%|█▍        | 1967/13317 [20:15<1:55:58,  1.63it/s]
                                                      
{'loss': 0.2424, 'grad_norm': 0.5888246893882751, 'learning_rate': 9.930895204044257e-06, 'epoch': 0.44}

 15%|█▍        | 1967/13317 [20:15<1:55:58,  1.63it/s]
 15%|█▍        | 1968/13317 [20:15<1:56:03,  1.63it/s]
                                                      
{'loss': 0.2242, 'grad_norm': 0.5722543597221375, 'learning_rate': 9.930677884497076e-06, 'epoch': 0.44}

 15%|█▍        | 1968/13317 [20:16<1:56:03,  1.63it/s]
 15%|█▍        | 1969/13317 [20:16<1:56:03,  1.63it/s]
                                                      
{'loss': 0.2889, 'grad_norm': 0.5902222394943237, 'learning_rate': 9.930460226160118e-06, 'epoch': 0.44}

 15%|█▍        | 1969/13317 [20:16<1:56:03,  1.63it/s]
 15%|█▍        | 1970/13317 [20:17<1:56:03,  1.63it/s]
                                                      
{'loss': 0.3902, 'grad_norm': 0.667751133441925, 'learning_rate': 9.930242229048336e-06, 'epoch': 0.44}

 15%|█▍        | 1970/13317 [20:17<1:56:03,  1.63it/s]
 15%|█▍        | 1971/13317 [20:17<1:55:58,  1.63it/s]
                                                      
{'loss': 0.2802, 'grad_norm': 0.5843353271484375, 'learning_rate': 9.930023893176713e-06, 'epoch': 0.44}

 15%|█▍        | 1971/13317 [20:17<1:55:58,  1.63it/s]
 15%|█▍        | 1972/13317 [20:18<1:55:57,  1.63it/s]
                                                      
{'loss': 0.2933, 'grad_norm': 0.5891402959823608, 'learning_rate': 9.929805218560247e-06, 'epoch': 0.44}

 15%|█▍        | 1972/13317 [20:18<1:55:57,  1.63it/s]
 15%|█▍        | 1973/13317 [20:18<1:55:55,  1.63it/s]
                                                      
{'loss': 0.2347, 'grad_norm': 0.49416235089302063, 'learning_rate': 9.929586205213964e-06, 'epoch': 0.44}

 15%|█▍        | 1973/13317 [20:19<1:55:55,  1.63it/s]
 15%|█▍        | 1974/13317 [20:19<1:55:59,  1.63it/s]
                                                      
{'loss': 0.2659, 'grad_norm': 0.5889056921005249, 'learning_rate': 9.929366853152915e-06, 'epoch': 0.44}

 15%|█▍        | 1974/13317 [20:19<1:55:59,  1.63it/s]
 15%|█▍        | 1975/13317 [20:20<1:55:59,  1.63it/s]
                                                      
{'loss': 0.2023, 'grad_norm': 0.5689650774002075, 'learning_rate': 9.92914716239217e-06, 'epoch': 0.44}

 15%|█▍        | 1975/13317 [20:20<1:55:59,  1.63it/s]
 15%|█▍        | 1976/13317 [20:20<1:55:59,  1.63it/s]
                                                      
{'loss': 0.2029, 'grad_norm': 0.5325412750244141, 'learning_rate': 9.928927132946823e-06, 'epoch': 0.45}

 15%|█▍        | 1976/13317 [20:20<1:55:59,  1.63it/s]
 15%|█▍        | 1977/13317 [20:21<1:55:58,  1.63it/s]
                                                      
{'loss': 0.3422, 'grad_norm': 0.747582733631134, 'learning_rate': 9.928706764831996e-06, 'epoch': 0.45}

 15%|█▍        | 1977/13317 [20:21<1:55:58,  1.63it/s]
 15%|█▍        | 1978/13317 [20:22<1:55:54,  1.63it/s]
                                                      
{'loss': 0.3096, 'grad_norm': 0.6857290863990784, 'learning_rate': 9.928486058062828e-06, 'epoch': 0.45}

 15%|█▍        | 1978/13317 [20:22<1:55:54,  1.63it/s]
 15%|█▍        | 1979/13317 [20:22<1:55:56,  1.63it/s]
                                                      
{'loss': 0.2587, 'grad_norm': 0.5386132597923279, 'learning_rate': 9.928265012654483e-06, 'epoch': 0.45}

 15%|█▍        | 1979/13317 [20:22<1:55:56,  1.63it/s]
 15%|█▍        | 1980/13317 [20:23<1:55:56,  1.63it/s]
                                                      
{'loss': 0.2384, 'grad_norm': 0.5883188843727112, 'learning_rate': 9.92804362862215e-06, 'epoch': 0.45}

 15%|█▍        | 1980/13317 [20:23<1:55:56,  1.63it/s]
 15%|█▍        | 1981/13317 [20:23<1:55:57,  1.63it/s]
                                                      
{'loss': 0.2806, 'grad_norm': 0.58869469165802, 'learning_rate': 9.92782190598104e-06, 'epoch': 0.45}

 15%|█▍        | 1981/13317 [20:23<1:55:57,  1.63it/s]
 15%|█▍        | 1982/13317 [20:24<1:55:58,  1.63it/s]
                                                      
{'loss': 0.369, 'grad_norm': 0.7091907262802124, 'learning_rate': 9.927599844746392e-06, 'epoch': 0.45}

 15%|█▍        | 1982/13317 [20:24<1:55:58,  1.63it/s]
 15%|█▍        | 1983/13317 [20:25<1:56:01,  1.63it/s]
                                                      
{'loss': 0.2338, 'grad_norm': 0.4985731840133667, 'learning_rate': 9.927377444933457e-06, 'epoch': 0.45}

 15%|█▍        | 1983/13317 [20:25<1:56:01,  1.63it/s]
 15%|█▍        | 1984/13317 [20:25<1:55:58,  1.63it/s]
                                                      
{'loss': 0.2816, 'grad_norm': 0.5639927983283997, 'learning_rate': 9.92715470655752e-06, 'epoch': 0.45}

 15%|█▍        | 1984/13317 [20:25<1:55:58,  1.63it/s]
 15%|█▍        | 1985/13317 [20:26<1:55:51,  1.63it/s]
                                                      
{'loss': 0.289, 'grad_norm': 0.633503794670105, 'learning_rate': 9.926931629633887e-06, 'epoch': 0.45}

 15%|█▍        | 1985/13317 [20:26<1:55:51,  1.63it/s]
 15%|█▍        | 1986/13317 [20:26<1:55:52,  1.63it/s]
                                                      
{'loss': 0.3044, 'grad_norm': 0.624982476234436, 'learning_rate': 9.926708214177883e-06, 'epoch': 0.45}

 15%|█▍        | 1986/13317 [20:27<1:55:52,  1.63it/s]
 15%|█▍        | 1987/13317 [20:27<1:55:49,  1.63it/s]
                                                      
{'loss': 0.3857, 'grad_norm': 0.7496258020401001, 'learning_rate': 9.926484460204859e-06, 'epoch': 0.45}

 15%|█▍        | 1987/13317 [20:27<1:55:49,  1.63it/s]
 15%|█▍        | 1988/13317 [20:28<1:55:49,  1.63it/s]
                                                      
{'loss': 0.234, 'grad_norm': 0.5448757410049438, 'learning_rate': 9.926260367730192e-06, 'epoch': 0.45}

 15%|█▍        | 1988/13317 [20:28<1:55:49,  1.63it/s]
 15%|█▍        | 1989/13317 [20:28<1:55:47,  1.63it/s]
                                                      
{'loss': 0.2139, 'grad_norm': 0.5103499889373779, 'learning_rate': 9.926035936769276e-06, 'epoch': 0.45}

 15%|█▍        | 1989/13317 [20:28<1:55:47,  1.63it/s]
 15%|█▍        | 1990/13317 [20:29<1:55:46,  1.63it/s]
                                                      
{'loss': 0.302, 'grad_norm': 0.629708468914032, 'learning_rate': 9.925811167337533e-06, 'epoch': 0.45}

 15%|█▍        | 1990/13317 [20:29<1:55:46,  1.63it/s]
 15%|█▍        | 1991/13317 [20:29<1:55:47,  1.63it/s]
                                                      
{'loss': 0.2855, 'grad_norm': 0.5942504405975342, 'learning_rate': 9.925586059450409e-06, 'epoch': 0.45}

 15%|█▍        | 1991/13317 [20:30<1:55:47,  1.63it/s]
 15%|█▍        | 1992/13317 [20:30<1:55:43,  1.63it/s]
                                                      
{'loss': 0.3515, 'grad_norm': 0.6885195374488831, 'learning_rate': 9.92536061312337e-06, 'epoch': 0.45}

 15%|█▍        | 1992/13317 [20:30<1:55:43,  1.63it/s]
 15%|█▍        | 1993/13317 [20:31<1:55:42,  1.63it/s]
                                                      
{'loss': 0.2531, 'grad_norm': 0.6333690285682678, 'learning_rate': 9.925134828371904e-06, 'epoch': 0.45}

 15%|█▍        | 1993/13317 [20:31<1:55:42,  1.63it/s]
 15%|█▍        | 1994/13317 [20:31<1:55:42,  1.63it/s]
                                                      
{'loss': 0.2697, 'grad_norm': 0.5366863012313843, 'learning_rate': 9.92490870521153e-06, 'epoch': 0.45}

 15%|█▍        | 1994/13317 [20:31<1:55:42,  1.63it/s]
 15%|█▍        | 1995/13317 [20:32<1:55:46,  1.63it/s]
                                                      
{'loss': 0.3174, 'grad_norm': 0.673125147819519, 'learning_rate': 9.92468224365778e-06, 'epoch': 0.45}

 15%|█▍        | 1995/13317 [20:32<1:55:46,  1.63it/s]
 15%|█▍        | 1996/13317 [20:33<1:55:44,  1.63it/s]
                                                      
{'loss': 0.2157, 'grad_norm': 0.5154394507408142, 'learning_rate': 9.924455443726217e-06, 'epoch': 0.45}

 15%|█▍        | 1996/13317 [20:33<1:55:44,  1.63it/s]
 15%|█▍        | 1997/13317 [20:33<1:55:46,  1.63it/s]
                                                      
{'loss': 0.2909, 'grad_norm': 0.6692171692848206, 'learning_rate': 9.924228305432424e-06, 'epoch': 0.45}

 15%|█▍        | 1997/13317 [20:33<1:55:46,  1.63it/s]
 15%|█▌        | 1998/13317 [20:34<1:55:40,  1.63it/s]
                                                      
{'loss': 0.2615, 'grad_norm': 0.5652177333831787, 'learning_rate': 9.924000828792008e-06, 'epoch': 0.45}

 15%|█▌        | 1998/13317 [20:34<1:55:40,  1.63it/s]
 15%|█▌        | 1999/13317 [20:34<1:55:39,  1.63it/s]
                                                      
{'loss': 0.3099, 'grad_norm': 0.7051103115081787, 'learning_rate': 9.923773013820599e-06, 'epoch': 0.45}

 15%|█▌        | 1999/13317 [20:35<1:55:39,  1.63it/s]
 15%|█▌        | 2000/13317 [20:35<1:55:43,  1.63it/s]
                                                      
{'loss': 0.3056, 'grad_norm': 0.7217023372650146, 'learning_rate': 9.92354486053385e-06, 'epoch': 0.45}

 15%|█▌        | 2000/13317 [20:35<1:55:43,  1.63it/s]

  0%|          | 0/13 [00:00<?, ?it/s]

 15%|█▌        | 2/13 [00:00<00:00, 13.37it/s]

 31%|███       | 4/13 [00:00<00:01,  8.41it/s]

 38%|███▊      | 5/13 [00:00<00:01,  7.81it/s]

 46%|████▌     | 6/13 [00:00<00:00,  7.43it/s]

 54%|█████▍    | 7/13 [00:00<00:00,  7.19it/s]

 62%|██████▏   | 8/13 [00:01<00:00,  7.02it/s]

 69%|██████▉   | 9/13 [00:01<00:00,  6.91it/s]

 77%|███████▋  | 10/13 [00:01<00:00,  6.83it/s]

 85%|████████▍ | 11/13 [00:01<00:00,  6.78it/s]

 92%|█████████▏| 12/13 [00:01<00:00,  6.75it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]
                                                      

                                               
{'eval_loss': 0.7398889064788818, 'eval_runtime': 1.952, 'eval_samples_per_second': 102.458, 'eval_steps_per_second': 6.66, 'epoch': 0.45}

 15%|█▌        | 2000/13317 [20:37<1:55:43,  1.63it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]

                                               
 15%|█▌        | 2001/13317 [20:38<3:46:25,  1.20s/it]
                                                      
{'loss': 0.2574, 'grad_norm': 0.5632222890853882, 'learning_rate': 9.923316368947439e-06, 'epoch': 0.45}

 15%|█▌        | 2001/13317 [20:38<3:46:25,  1.20s/it]
 15%|█▌        | 2002/13317 [20:38<3:13:11,  1.02s/it]
                                                      
{'loss': 0.2591, 'grad_norm': 0.5718333721160889, 'learning_rate': 9.923087539077063e-06, 'epoch': 0.45}

 15%|█▌        | 2002/13317 [20:38<3:13:11,  1.02s/it]
 15%|█▌        | 2003/13317 [20:39<2:49:57,  1.11it/s]
                                                      
{'loss': 0.2969, 'grad_norm': 0.5756542682647705, 'learning_rate': 9.922858370938448e-06, 'epoch': 0.45}

 15%|█▌        | 2003/13317 [20:39<2:49:57,  1.11it/s]
 15%|█▌        | 2004/13317 [20:39<2:33:38,  1.23it/s]
                                                      
{'loss': 0.2725, 'grad_norm': 0.6623870730400085, 'learning_rate': 9.922628864547336e-06, 'epoch': 0.45}

 15%|█▌        | 2004/13317 [20:40<2:33:38,  1.23it/s]
 15%|█▌        | 2005/13317 [20:40<2:22:16,  1.33it/s]
                                                      
{'loss': 0.3843, 'grad_norm': 0.6988383531570435, 'learning_rate': 9.922399019919502e-06, 'epoch': 0.45}

 15%|█▌        | 2005/13317 [20:40<2:22:16,  1.33it/s]
 15%|█▌        | 2006/13317 [20:41<2:14:15,  1.40it/s]
                                                      
{'loss': 0.252, 'grad_norm': 0.6031769514083862, 'learning_rate': 9.922168837070736e-06, 'epoch': 0.45}

 15%|█▌        | 2006/13317 [20:41<2:14:15,  1.40it/s]
 15%|█▌        | 2007/13317 [20:41<2:08:39,  1.47it/s]
                                                      
{'loss': 0.2687, 'grad_norm': 0.5224857330322266, 'learning_rate': 9.921938316016854e-06, 'epoch': 0.45}

 15%|█▌        | 2007/13317 [20:41<2:08:39,  1.47it/s]
 15%|█▌        | 2008/13317 [20:42<2:04:45,  1.51it/s]
                                                      
{'loss': 0.2749, 'grad_norm': 0.5937696099281311, 'learning_rate': 9.921707456773695e-06, 'epoch': 0.45}

 15%|█▌        | 2008/13317 [20:42<2:04:45,  1.51it/s]
 15%|█▌        | 2009/13317 [20:42<2:02:02,  1.54it/s]
                                                      
{'loss': 0.2508, 'grad_norm': 0.5421342849731445, 'learning_rate': 9.921476259357122e-06, 'epoch': 0.45}

 15%|█▌        | 2009/13317 [20:43<2:02:02,  1.54it/s]
 15%|█▌        | 2010/13317 [20:43<2:00:05,  1.57it/s]
                                                      
{'loss': 0.2446, 'grad_norm': 0.5200783014297485, 'learning_rate': 9.92124472378302e-06, 'epoch': 0.45}

 15%|█▌        | 2010/13317 [20:43<2:00:05,  1.57it/s]
 15%|█▌        | 2011/13317 [20:44<1:58:43,  1.59it/s]
                                                      
{'loss': 0.2345, 'grad_norm': 0.5128250122070312, 'learning_rate': 9.921012850067299e-06, 'epoch': 0.45}

 15%|█▌        | 2011/13317 [20:44<1:58:43,  1.59it/s]
 15%|█▌        | 2012/13317 [20:44<1:57:45,  1.60it/s]
                                                      
{'loss': 0.3734, 'grad_norm': 0.6046855449676514, 'learning_rate': 9.92078063822589e-06, 'epoch': 0.45}

 15%|█▌        | 2012/13317 [20:44<1:57:45,  1.60it/s]
 15%|█▌        | 2013/13317 [20:45<1:57:06,  1.61it/s]
                                                      
{'loss': 0.2567, 'grad_norm': 0.5709905624389648, 'learning_rate': 9.92054808827475e-06, 'epoch': 0.45}

 15%|█▌        | 2013/13317 [20:45<1:57:06,  1.61it/s]
 15%|█▌        | 2014/13317 [20:46<1:56:38,  1.62it/s]
                                                      
{'loss': 0.2928, 'grad_norm': 0.7237128019332886, 'learning_rate': 9.920315200229856e-06, 'epoch': 0.45}

 15%|█▌        | 2014/13317 [20:46<1:56:38,  1.62it/s]
 15%|█▌        | 2015/13317 [20:46<1:56:18,  1.62it/s]
                                                      
{'loss': 0.2908, 'grad_norm': 0.6870774626731873, 'learning_rate': 9.920081974107212e-06, 'epoch': 0.45}

 15%|█▌        | 2015/13317 [20:46<1:56:18,  1.62it/s]
 15%|█▌        | 2016/13317 [20:47<1:55:59,  1.62it/s]
                                                      
{'loss': 0.2564, 'grad_norm': 0.5286972522735596, 'learning_rate': 9.91984840992284e-06, 'epoch': 0.45}

 15%|█▌        | 2016/13317 [20:47<1:55:59,  1.62it/s]
 15%|█▌        | 2017/13317 [20:47<1:55:51,  1.63it/s]
                                                      
{'loss': 0.2577, 'grad_norm': 0.5703082084655762, 'learning_rate': 9.919614507692793e-06, 'epoch': 0.45}

 15%|█▌        | 2017/13317 [20:48<1:55:51,  1.63it/s]
 15%|█▌        | 2018/13317 [20:48<1:55:44,  1.63it/s]
                                                      
{'loss': 0.2348, 'grad_norm': 0.5582262873649597, 'learning_rate': 9.919380267433135e-06, 'epoch': 0.45}

 15%|█▌        | 2018/13317 [20:48<1:55:44,  1.63it/s]
 15%|█▌        | 2019/13317 [20:49<1:55:41,  1.63it/s]
                                                      
{'loss': 0.262, 'grad_norm': 0.5566505193710327, 'learning_rate': 9.919145689159968e-06, 'epoch': 0.45}

 15%|█▌        | 2019/13317 [20:49<1:55:41,  1.63it/s]
 15%|█▌        | 2020/13317 [20:49<1:55:40,  1.63it/s]
                                                      
{'loss': 0.302, 'grad_norm': 0.7310200333595276, 'learning_rate': 9.918910772889407e-06, 'epoch': 0.46}

 15%|█▌        | 2020/13317 [20:49<1:55:40,  1.63it/s]
 15%|█▌        | 2021/13317 [20:50<1:55:38,  1.63it/s]
                                                      
{'loss': 0.2828, 'grad_norm': 0.6160504221916199, 'learning_rate': 9.918675518637595e-06, 'epoch': 0.46}

 15%|█▌        | 2021/13317 [20:50<1:55:38,  1.63it/s]
 15%|█▌        | 2022/13317 [20:50<1:55:35,  1.63it/s]
                                                      
{'loss': 0.2335, 'grad_norm': 0.5732446312904358, 'learning_rate': 9.918439926420693e-06, 'epoch': 0.46}

 15%|█▌        | 2022/13317 [20:51<1:55:35,  1.63it/s]
 15%|█▌        | 2023/13317 [20:51<1:55:34,  1.63it/s]
                                                      
{'loss': 0.2835, 'grad_norm': 0.5235902070999146, 'learning_rate': 9.918203996254892e-06, 'epoch': 0.46}

 15%|█▌        | 2023/13317 [20:51<1:55:34,  1.63it/s]
 15%|█▌        | 2024/13317 [20:52<1:55:33,  1.63it/s]
                                                      
{'loss': 0.3118, 'grad_norm': 0.5858473181724548, 'learning_rate': 9.917967728156402e-06, 'epoch': 0.46}

 15%|█▌        | 2024/13317 [20:52<1:55:33,  1.63it/s]
 15%|█▌        | 2025/13317 [20:52<1:55:30,  1.63it/s]
                                                      
{'loss': 0.3028, 'grad_norm': 0.6644707918167114, 'learning_rate': 9.917731122141456e-06, 'epoch': 0.46}

 15%|█▌        | 2025/13317 [20:52<1:55:30,  1.63it/s]
 15%|█▌        | 2026/13317 [20:53<1:55:26,  1.63it/s]
                                                      
{'loss': 0.3323, 'grad_norm': 0.6342570781707764, 'learning_rate': 9.917494178226313e-06, 'epoch': 0.46}

 15%|█▌        | 2026/13317 [20:53<1:55:26,  1.63it/s]
 15%|█▌        | 2027/13317 [20:54<1:55:25,  1.63it/s]
                                                      
{'loss': 0.2215, 'grad_norm': 0.5575411915779114, 'learning_rate': 9.91725689642725e-06, 'epoch': 0.46}

 15%|█▌        | 2027/13317 [20:54<1:55:25,  1.63it/s]
 15%|█▌        | 2028/13317 [20:54<1:55:21,  1.63it/s]
                                                      
{'loss': 0.2848, 'grad_norm': 0.5910139679908752, 'learning_rate': 9.917019276760577e-06, 'epoch': 0.46}

 15%|█▌        | 2028/13317 [20:54<1:55:21,  1.63it/s]
 15%|█▌        | 2029/13317 [20:55<1:55:23,  1.63it/s]
                                                      
{'loss': 0.2684, 'grad_norm': 0.5820047855377197, 'learning_rate': 9.916781319242616e-06, 'epoch': 0.46}

 15%|█▌        | 2029/13317 [20:55<1:55:23,  1.63it/s]
 15%|█▌        | 2030/13317 [20:55<1:55:26,  1.63it/s]
                                                      
{'loss': 0.3128, 'grad_norm': 0.6368503570556641, 'learning_rate': 9.916543023889717e-06, 'epoch': 0.46}

 15%|█▌        | 2030/13317 [20:56<1:55:26,  1.63it/s]
 15%|█▌        | 2031/13317 [20:56<1:55:23,  1.63it/s]
                                                      
{'loss': 0.3508, 'grad_norm': 0.6723087430000305, 'learning_rate': 9.916304390718258e-06, 'epoch': 0.46}

 15%|█▌        | 2031/13317 [20:56<1:55:23,  1.63it/s]
 15%|█▌        | 2032/13317 [20:57<1:55:21,  1.63it/s]
                                                      
{'loss': 0.3161, 'grad_norm': 0.6536437273025513, 'learning_rate': 9.916065419744632e-06, 'epoch': 0.46}

 15%|█▌        | 2032/13317 [20:57<1:55:21,  1.63it/s]
 15%|█▌        | 2033/13317 [20:57<1:55:21,  1.63it/s]
                                                      
{'loss': 0.2676, 'grad_norm': 0.6112756133079529, 'learning_rate': 9.915826110985259e-06, 'epoch': 0.46}

 15%|█▌        | 2033/13317 [20:57<1:55:21,  1.63it/s]
 15%|█▌        | 2034/13317 [20:58<1:55:22,  1.63it/s]
                                                      
{'loss': 0.2913, 'grad_norm': 0.5926722884178162, 'learning_rate': 9.915586464456582e-06, 'epoch': 0.46}

 15%|█▌        | 2034/13317 [20:58<1:55:22,  1.63it/s]
 15%|█▌        | 2035/13317 [20:58<1:55:18,  1.63it/s]
                                                      
{'loss': 0.281, 'grad_norm': 0.5960965752601624, 'learning_rate': 9.915346480175067e-06, 'epoch': 0.46}

 15%|█▌        | 2035/13317 [20:59<1:55:18,  1.63it/s]
 15%|█▌        | 2036/13317 [20:59<1:55:16,  1.63it/s]
                                                      
{'loss': 0.2899, 'grad_norm': 0.689972460269928, 'learning_rate': 9.915106158157207e-06, 'epoch': 0.46}

 15%|█▌        | 2036/13317 [20:59<1:55:16,  1.63it/s]
 15%|█▌        | 2037/13317 [21:00<1:55:17,  1.63it/s]
                                                      
{'loss': 0.2732, 'grad_norm': 0.604254424571991, 'learning_rate': 9.91486549841951e-06, 'epoch': 0.46}

 15%|█▌        | 2037/13317 [21:00<1:55:17,  1.63it/s]
 15%|█▌        | 2038/13317 [21:00<1:55:19,  1.63it/s]
                                                      
{'loss': 0.3763, 'grad_norm': 0.6495276093482971, 'learning_rate': 9.914624500978513e-06, 'epoch': 0.46}

 15%|█▌        | 2038/13317 [21:00<1:55:19,  1.63it/s]
 15%|█▌        | 2039/13317 [21:01<1:55:17,  1.63it/s]
                                                      
{'loss': 0.2541, 'grad_norm': 0.5726681351661682, 'learning_rate': 9.914383165850779e-06, 'epoch': 0.46}

 15%|█▌        | 2039/13317 [21:01<1:55:17,  1.63it/s]
 15%|█▌        | 2040/13317 [21:02<1:55:17,  1.63it/s]
                                                      
{'loss': 0.2108, 'grad_norm': 0.47319495677948, 'learning_rate': 9.914141493052884e-06, 'epoch': 0.46}

 15%|█▌        | 2040/13317 [21:02<1:55:17,  1.63it/s]
 15%|█▌        | 2041/13317 [21:02<1:55:18,  1.63it/s]
                                                      
{'loss': 0.1902, 'grad_norm': 0.5366383194923401, 'learning_rate': 9.913899482601439e-06, 'epoch': 0.46}

 15%|█▌        | 2041/13317 [21:02<1:55:18,  1.63it/s]
 15%|█▌        | 2042/13317 [21:03<1:55:16,  1.63it/s]
                                                      
{'loss': 0.3035, 'grad_norm': 0.6589699387550354, 'learning_rate': 9.91365713451307e-06, 'epoch': 0.46}

 15%|█▌        | 2042/13317 [21:03<1:55:16,  1.63it/s]
 15%|█▌        | 2043/13317 [21:03<1:55:12,  1.63it/s]
                                                      
{'loss': 0.3697, 'grad_norm': 0.5730133056640625, 'learning_rate': 9.913414448804428e-06, 'epoch': 0.46}

 15%|█▌        | 2043/13317 [21:03<1:55:12,  1.63it/s]
 15%|█▌        | 2044/13317 [21:04<1:55:16,  1.63it/s]
                                                      
{'loss': 0.2994, 'grad_norm': 0.5455162525177002, 'learning_rate': 9.913171425492189e-06, 'epoch': 0.46}

 15%|█▌        | 2044/13317 [21:04<1:55:16,  1.63it/s]
 15%|█▌        | 2045/13317 [21:05<1:55:15,  1.63it/s]
                                                      
{'loss': 0.3865, 'grad_norm': 0.6457644104957581, 'learning_rate': 9.912928064593053e-06, 'epoch': 0.46}

 15%|█▌        | 2045/13317 [21:05<1:55:15,  1.63it/s]
 15%|█▌        | 2046/13317 [21:05<1:55:13,  1.63it/s]
                                                      
{'loss': 0.2646, 'grad_norm': 0.5573134422302246, 'learning_rate': 9.91268436612374e-06, 'epoch': 0.46}

 15%|█▌        | 2046/13317 [21:05<1:55:13,  1.63it/s]
 15%|█▌        | 2047/13317 [21:06<1:55:07,  1.63it/s]
                                                      
{'loss': 0.2298, 'grad_norm': 0.5228017568588257, 'learning_rate': 9.912440330100994e-06, 'epoch': 0.46}

 15%|█▌        | 2047/13317 [21:06<1:55:07,  1.63it/s]
 15%|█▌        | 2048/13317 [21:06<1:55:07,  1.63it/s]
                                                      
{'loss': 0.256, 'grad_norm': 0.5548675060272217, 'learning_rate': 9.912195956541584e-06, 'epoch': 0.46}

 15%|█▌        | 2048/13317 [21:07<1:55:07,  1.63it/s]
 15%|█▌        | 2049/13317 [21:07<1:55:09,  1.63it/s]
                                                      
{'loss': 0.2264, 'grad_norm': 0.4940539002418518, 'learning_rate': 9.911951245462301e-06, 'epoch': 0.46}

 15%|█▌        | 2049/13317 [21:07<1:55:09,  1.63it/s]
 15%|█▌        | 2050/13317 [21:08<1:55:10,  1.63it/s]
                                                      
{'loss': 0.2248, 'grad_norm': 0.5970724821090698, 'learning_rate': 9.911706196879957e-06, 'epoch': 0.46}

 15%|█▌        | 2050/13317 [21:08<1:55:10,  1.63it/s]
 15%|█▌        | 2051/13317 [21:08<1:55:07,  1.63it/s]
                                                      
{'loss': 0.3543, 'grad_norm': 0.6852599382400513, 'learning_rate': 9.911460810811395e-06, 'epoch': 0.46}

 15%|█▌        | 2051/13317 [21:08<1:55:07,  1.63it/s]
 15%|█▌        | 2052/13317 [21:09<1:55:04,  1.63it/s]
                                                      
{'loss': 0.2864, 'grad_norm': 0.678033173084259, 'learning_rate': 9.911215087273469e-06, 'epoch': 0.46}

 15%|█▌        | 2052/13317 [21:09<1:55:04,  1.63it/s]
 15%|█▌        | 2053/13317 [21:09<1:55:08,  1.63it/s]
                                                      
{'loss': 0.2922, 'grad_norm': 0.6447487473487854, 'learning_rate': 9.910969026283065e-06, 'epoch': 0.46}

 15%|█▌        | 2053/13317 [21:10<1:55:08,  1.63it/s]
 15%|█▌        | 2054/13317 [21:10<1:55:15,  1.63it/s]
                                                      
{'loss': 0.1878, 'grad_norm': 0.45753124356269836, 'learning_rate': 9.910722627857092e-06, 'epoch': 0.46}

 15%|█▌        | 2054/13317 [21:10<1:55:15,  1.63it/s]
 15%|█▌        | 2055/13317 [21:11<1:55:13,  1.63it/s]
                                                      
{'loss': 0.3451, 'grad_norm': 0.6390776634216309, 'learning_rate': 9.91047589201248e-06, 'epoch': 0.46}

 15%|█▌        | 2055/13317 [21:11<1:55:13,  1.63it/s]
 15%|█▌        | 2056/13317 [21:11<1:55:11,  1.63it/s]
                                                      
{'loss': 0.2472, 'grad_norm': 0.5569659471511841, 'learning_rate': 9.910228818766179e-06, 'epoch': 0.46}

 15%|█▌        | 2056/13317 [21:11<1:55:11,  1.63it/s]
 15%|█▌        | 2057/13317 [21:12<1:55:08,  1.63it/s]
                                                      
{'loss': 0.3186, 'grad_norm': 0.5460555553436279, 'learning_rate': 9.90998140813517e-06, 'epoch': 0.46}

 15%|█▌        | 2057/13317 [21:12<1:55:08,  1.63it/s]
 15%|█▌        | 2058/13317 [21:13<1:55:09,  1.63it/s]
                                                      
{'loss': 0.2186, 'grad_norm': 0.4692433178424835, 'learning_rate': 9.909733660136448e-06, 'epoch': 0.46}

 15%|█▌        | 2058/13317 [21:13<1:55:09,  1.63it/s]
 15%|█▌        | 2059/13317 [21:13<1:55:40,  1.62it/s]
                                                      
{'loss': 0.2351, 'grad_norm': 0.5578592419624329, 'learning_rate': 9.90948557478704e-06, 'epoch': 0.46}

 15%|█▌        | 2059/13317 [21:13<1:55:40,  1.62it/s]
 15%|█▌        | 2060/13317 [21:14<1:55:28,  1.62it/s]
                                                      
{'loss': 0.2804, 'grad_norm': 0.6927225589752197, 'learning_rate': 9.909237152103993e-06, 'epoch': 0.46}

 15%|█▌        | 2060/13317 [21:14<1:55:28,  1.62it/s]
 15%|█▌        | 2061/13317 [21:14<1:55:24,  1.63it/s]
                                                      
{'loss': 0.2213, 'grad_norm': 0.4937034249305725, 'learning_rate': 9.90898839210437e-06, 'epoch': 0.46}

 15%|█▌        | 2061/13317 [21:15<1:55:24,  1.63it/s]
 15%|█▌        | 2062/13317 [21:15<1:55:19,  1.63it/s]
                                                      
{'loss': 0.2704, 'grad_norm': 0.578766405582428, 'learning_rate': 9.90873929480527e-06, 'epoch': 0.46}

 15%|█▌        | 2062/13317 [21:15<1:55:19,  1.63it/s]
 15%|█▌        | 2063/13317 [21:16<1:55:14,  1.63it/s]
                                                      
{'loss': 0.2587, 'grad_norm': 0.6005515456199646, 'learning_rate': 9.908489860223805e-06, 'epoch': 0.46}

 15%|█▌        | 2063/13317 [21:16<1:55:14,  1.63it/s]
 15%|█▌        | 2064/13317 [21:16<1:55:22,  1.63it/s]
                                                      
{'loss': 0.2506, 'grad_norm': 0.5742387771606445, 'learning_rate': 9.908240088377113e-06, 'epoch': 0.46}

 15%|█▌        | 2064/13317 [21:16<1:55:22,  1.63it/s]
 16%|█▌        | 2065/13317 [21:17<1:55:20,  1.63it/s]
                                                      
{'loss': 0.2918, 'grad_norm': 0.5823510885238647, 'learning_rate': 9.907989979282358e-06, 'epoch': 0.47}

 16%|█▌        | 2065/13317 [21:17<1:55:20,  1.63it/s]
 16%|█▌        | 2066/13317 [21:17<1:55:27,  1.62it/s]
                                                      
{'loss': 0.3147, 'grad_norm': 0.6589888334274292, 'learning_rate': 9.907739532956725e-06, 'epoch': 0.47}

 16%|█▌        | 2066/13317 [21:18<1:55:27,  1.62it/s]
 16%|█▌        | 2067/13317 [21:18<1:55:23,  1.62it/s]
                                                      
{'loss': 0.3654, 'grad_norm': 0.6351408958435059, 'learning_rate': 9.907488749417425e-06, 'epoch': 0.47}

 16%|█▌        | 2067/13317 [21:18<1:55:23,  1.62it/s]
 16%|█▌        | 2068/13317 [21:19<1:55:16,  1.63it/s]
                                                      
{'loss': 0.3525, 'grad_norm': 0.7341656684875488, 'learning_rate': 9.907237628681682e-06, 'epoch': 0.47}

 16%|█▌        | 2068/13317 [21:19<1:55:16,  1.63it/s]
 16%|█▌        | 2069/13317 [21:19<1:55:18,  1.63it/s]
                                                      
{'loss': 0.2406, 'grad_norm': 0.5567678809165955, 'learning_rate': 9.906986170766759e-06, 'epoch': 0.47}

 16%|█▌        | 2069/13317 [21:19<1:55:18,  1.63it/s]
 16%|█▌        | 2070/13317 [21:20<1:55:28,  1.62it/s]
                                                      
{'loss': 0.2299, 'grad_norm': 0.569022536277771, 'learning_rate': 9.906734375689928e-06, 'epoch': 0.47}

 16%|█▌        | 2070/13317 [21:20<1:55:28,  1.62it/s]
 16%|█▌        | 2071/13317 [21:21<1:55:26,  1.62it/s]
                                                      
{'loss': 0.3569, 'grad_norm': 0.8311614990234375, 'learning_rate': 9.906482243468494e-06, 'epoch': 0.47}

 16%|█▌        | 2071/13317 [21:21<1:55:26,  1.62it/s]
 16%|█▌        | 2072/13317 [21:21<1:55:18,  1.63it/s]
                                                      
{'loss': 0.4607, 'grad_norm': 0.7328338027000427, 'learning_rate': 9.906229774119778e-06, 'epoch': 0.47}

 16%|█▌        | 2072/13317 [21:21<1:55:18,  1.63it/s]
 16%|█▌        | 2073/13317 [21:22<1:55:13,  1.63it/s]
                                                      
{'loss': 0.211, 'grad_norm': 0.49636217951774597, 'learning_rate': 9.90597696766113e-06, 'epoch': 0.47}

 16%|█▌        | 2073/13317 [21:22<1:55:13,  1.63it/s]
 16%|█▌        | 2074/13317 [21:22<1:55:19,  1.62it/s]
                                                      
{'loss': 0.2609, 'grad_norm': 0.6146278381347656, 'learning_rate': 9.905723824109915e-06, 'epoch': 0.47}

 16%|█▌        | 2074/13317 [21:23<1:55:19,  1.62it/s]
 16%|█▌        | 2075/13317 [21:23<1:55:10,  1.63it/s]
                                                      
{'loss': 0.2961, 'grad_norm': 0.6149910092353821, 'learning_rate': 9.905470343483534e-06, 'epoch': 0.47}

 16%|█▌        | 2075/13317 [21:23<1:55:10,  1.63it/s]
 16%|█▌        | 2076/13317 [21:24<1:55:10,  1.63it/s]
                                                      
{'loss': 0.2151, 'grad_norm': 0.503352165222168, 'learning_rate': 9.905216525799399e-06, 'epoch': 0.47}

 16%|█▌        | 2076/13317 [21:24<1:55:10,  1.63it/s]
 16%|█▌        | 2077/13317 [21:24<1:55:16,  1.63it/s]
                                                      
{'loss': 0.2154, 'grad_norm': 0.46159958839416504, 'learning_rate': 9.904962371074951e-06, 'epoch': 0.47}

 16%|█▌        | 2077/13317 [21:24<1:55:16,  1.63it/s]
 16%|█▌        | 2078/13317 [21:25<1:55:08,  1.63it/s]
                                                      
{'loss': 0.2639, 'grad_norm': 0.4679913818836212, 'learning_rate': 9.904707879327655e-06, 'epoch': 0.47}

 16%|█▌        | 2078/13317 [21:25<1:55:08,  1.63it/s]
 16%|█▌        | 2079/13317 [21:25<1:55:13,  1.63it/s]
                                                      
{'loss': 0.344, 'grad_norm': 0.6530001759529114, 'learning_rate': 9.904453050574995e-06, 'epoch': 0.47}

 16%|█▌        | 2079/13317 [21:26<1:55:13,  1.63it/s]
 16%|█▌        | 2080/13317 [21:26<1:55:08,  1.63it/s]
                                                      
{'loss': 0.2637, 'grad_norm': 0.5499759316444397, 'learning_rate': 9.904197884834482e-06, 'epoch': 0.47}

 16%|█▌        | 2080/13317 [21:26<1:55:08,  1.63it/s]
 16%|█▌        | 2081/13317 [21:27<1:55:02,  1.63it/s]
                                                      
{'loss': 0.2554, 'grad_norm': 0.5822736620903015, 'learning_rate': 9.903942382123647e-06, 'epoch': 0.47}

 16%|█▌        | 2081/13317 [21:27<1:55:02,  1.63it/s]
 16%|█▌        | 2082/13317 [21:27<1:54:57,  1.63it/s]
                                                      
{'loss': 0.294, 'grad_norm': 0.7404043078422546, 'learning_rate': 9.903686542460047e-06, 'epoch': 0.47}

 16%|█▌        | 2082/13317 [21:27<1:54:57,  1.63it/s]
 16%|█▌        | 2083/13317 [21:28<1:54:58,  1.63it/s]
                                                      
{'loss': 0.3479, 'grad_norm': 0.6074334979057312, 'learning_rate': 9.903430365861261e-06, 'epoch': 0.47}

 16%|█▌        | 2083/13317 [21:28<1:54:58,  1.63it/s]
 16%|█▌        | 2084/13317 [21:29<1:54:53,  1.63it/s]
                                                      
{'loss': 0.3335, 'grad_norm': 0.6216788291931152, 'learning_rate': 9.903173852344889e-06, 'epoch': 0.47}

 16%|█▌        | 2084/13317 [21:29<1:54:53,  1.63it/s]
 16%|█▌        | 2085/13317 [21:29<1:54:49,  1.63it/s]
                                                      
{'loss': 0.329, 'grad_norm': 0.5555008053779602, 'learning_rate': 9.90291700192856e-06, 'epoch': 0.47}

 16%|█▌        | 2085/13317 [21:29<1:54:49,  1.63it/s]
 16%|█▌        | 2086/13317 [21:30<1:54:49,  1.63it/s]
                                                      
{'loss': 0.2847, 'grad_norm': 0.5732804536819458, 'learning_rate': 9.902659814629917e-06, 'epoch': 0.47}

 16%|█▌        | 2086/13317 [21:30<1:54:49,  1.63it/s]
 16%|█▌        | 2087/13317 [21:30<1:54:56,  1.63it/s]
                                                      
{'loss': 0.2978, 'grad_norm': 0.6836506724357605, 'learning_rate': 9.902402290466638e-06, 'epoch': 0.47}

 16%|█▌        | 2087/13317 [21:31<1:54:56,  1.63it/s]
 16%|█▌        | 2088/13317 [21:31<1:54:55,  1.63it/s]
                                                      
{'loss': 0.2157, 'grad_norm': 0.5057666301727295, 'learning_rate': 9.902144429456412e-06, 'epoch': 0.47}

 16%|█▌        | 2088/13317 [21:31<1:54:55,  1.63it/s]
 16%|█▌        | 2089/13317 [21:32<1:54:53,  1.63it/s]
                                                      
{'loss': 0.2305, 'grad_norm': 0.5164790153503418, 'learning_rate': 9.901886231616959e-06, 'epoch': 0.47}

 16%|█▌        | 2089/13317 [21:32<1:54:53,  1.63it/s]
 16%|█▌        | 2090/13317 [21:32<1:54:56,  1.63it/s]
                                                      
{'loss': 0.3421, 'grad_norm': 0.5836398601531982, 'learning_rate': 9.901627696966021e-06, 'epoch': 0.47}

 16%|█▌        | 2090/13317 [21:32<1:54:56,  1.63it/s]
 16%|█▌        | 2091/13317 [21:33<1:54:49,  1.63it/s]
                                                      
{'loss': 0.2525, 'grad_norm': 0.5903003811836243, 'learning_rate': 9.901368825521359e-06, 'epoch': 0.47}

 16%|█▌        | 2091/13317 [21:33<1:54:49,  1.63it/s]
 16%|█▌        | 2092/13317 [21:33<1:54:58,  1.63it/s]
                                                      
{'loss': 0.2503, 'grad_norm': 0.5681645274162292, 'learning_rate': 9.901109617300763e-06, 'epoch': 0.47}

 16%|█▌        | 2092/13317 [21:34<1:54:58,  1.63it/s]
 16%|█▌        | 2093/13317 [21:34<1:54:57,  1.63it/s]
                                                      
{'loss': 0.3003, 'grad_norm': 0.595439076423645, 'learning_rate': 9.900850072322043e-06, 'epoch': 0.47}

 16%|█▌        | 2093/13317 [21:34<1:54:57,  1.63it/s]
 16%|█▌        | 2094/13317 [21:35<1:54:49,  1.63it/s]
                                                      
{'loss': 0.3673, 'grad_norm': 0.7174631953239441, 'learning_rate': 9.900590190603032e-06, 'epoch': 0.47}

 16%|█▌        | 2094/13317 [21:35<1:54:49,  1.63it/s]
 16%|█▌        | 2095/13317 [21:35<1:54:48,  1.63it/s]
                                                      
{'loss': 0.2925, 'grad_norm': 0.7798371315002441, 'learning_rate': 9.900329972161586e-06, 'epoch': 0.47}

 16%|█▌        | 2095/13317 [21:35<1:54:48,  1.63it/s]
 16%|█▌        | 2096/13317 [21:36<1:54:43,  1.63it/s]
                                                      
{'loss': 0.2829, 'grad_norm': 0.5792388916015625, 'learning_rate': 9.900069417015588e-06, 'epoch': 0.47}

 16%|█▌        | 2096/13317 [21:36<1:54:43,  1.63it/s]
 16%|█▌        | 2097/13317 [21:37<1:54:51,  1.63it/s]
                                                      
{'loss': 0.2524, 'grad_norm': 0.5284010171890259, 'learning_rate': 9.899808525182935e-06, 'epoch': 0.47}

 16%|█▌        | 2097/13317 [21:37<1:54:51,  1.63it/s]
 16%|█▌        | 2098/13317 [21:37<1:54:48,  1.63it/s]
                                                      
{'loss': 0.2821, 'grad_norm': 0.568049967288971, 'learning_rate': 9.899547296681558e-06, 'epoch': 0.47}

 16%|█▌        | 2098/13317 [21:37<1:54:48,  1.63it/s]
 16%|█▌        | 2099/13317 [21:38<1:54:45,  1.63it/s]
                                                      
{'loss': 0.2131, 'grad_norm': 0.5909082293510437, 'learning_rate': 9.899285731529403e-06, 'epoch': 0.47}

 16%|█▌        | 2099/13317 [21:38<1:54:45,  1.63it/s]
 16%|█▌        | 2100/13317 [21:38<1:54:43,  1.63it/s]
                                                      
{'loss': 0.3419, 'grad_norm': 0.6747695803642273, 'learning_rate': 9.899023829744445e-06, 'epoch': 0.47}

 16%|█▌        | 2100/13317 [21:38<1:54:43,  1.63it/s]
 16%|█▌        | 2101/13317 [21:39<1:54:42,  1.63it/s]
                                                      
{'loss': 0.2702, 'grad_norm': 0.5476119518280029, 'learning_rate': 9.898761591344679e-06, 'epoch': 0.47}

 16%|█▌        | 2101/13317 [21:39<1:54:42,  1.63it/s]
 16%|█▌        | 2102/13317 [21:40<1:54:48,  1.63it/s]
                                                      
{'loss': 0.3108, 'grad_norm': 0.6813405156135559, 'learning_rate': 9.898499016348121e-06, 'epoch': 0.47}

 16%|█▌        | 2102/13317 [21:40<1:54:48,  1.63it/s]
 16%|█▌        | 2103/13317 [21:40<1:54:45,  1.63it/s]
                                                      
{'loss': 0.3002, 'grad_norm': 0.7767387628555298, 'learning_rate': 9.898236104772815e-06, 'epoch': 0.47}

 16%|█▌        | 2103/13317 [21:40<1:54:45,  1.63it/s]
 16%|█▌        | 2104/13317 [21:41<1:54:44,  1.63it/s]
                                                      
{'loss': 0.322, 'grad_norm': 0.7078081965446472, 'learning_rate': 9.897972856636825e-06, 'epoch': 0.47}

 16%|█▌        | 2104/13317 [21:41<1:54:44,  1.63it/s]
 16%|█▌        | 2105/13317 [21:41<1:54:44,  1.63it/s]
                                                      
{'loss': 0.282, 'grad_norm': 0.5290752053260803, 'learning_rate': 9.89770927195824e-06, 'epoch': 0.47}

 16%|█▌        | 2105/13317 [21:42<1:54:44,  1.63it/s]
 16%|█▌        | 2106/13317 [21:42<1:54:42,  1.63it/s]
                                                      
{'loss': 0.2614, 'grad_norm': 0.6035284399986267, 'learning_rate': 9.897445350755169e-06, 'epoch': 0.47}

 16%|█▌        | 2106/13317 [21:42<1:54:42,  1.63it/s]
 16%|█▌        | 2107/13317 [21:43<1:54:47,  1.63it/s]
                                                      
{'loss': 0.29, 'grad_norm': 0.5801594853401184, 'learning_rate': 9.89718109304575e-06, 'epoch': 0.47}

 16%|█▌        | 2107/13317 [21:43<1:54:47,  1.63it/s]
 16%|█▌        | 2108/13317 [21:43<1:54:46,  1.63it/s]
                                                      
{'loss': 0.3339, 'grad_norm': 0.6343690156936646, 'learning_rate': 9.896916498848135e-06, 'epoch': 0.47}

 16%|█▌        | 2108/13317 [21:43<1:54:46,  1.63it/s]
 16%|█▌        | 2109/13317 [21:44<1:54:41,  1.63it/s]
                                                      
{'loss': 0.2666, 'grad_norm': 0.5744673013687134, 'learning_rate': 9.896651568180508e-06, 'epoch': 0.48}

 16%|█▌        | 2109/13317 [21:44<1:54:41,  1.63it/s]
 16%|█▌        | 2110/13317 [21:45<1:54:40,  1.63it/s]
                                                      
{'loss': 0.3138, 'grad_norm': 0.62062668800354, 'learning_rate': 9.896386301061072e-06, 'epoch': 0.48}

 16%|█▌        | 2110/13317 [21:45<1:54:40,  1.63it/s]
 16%|█▌        | 2111/13317 [21:45<1:54:39,  1.63it/s]
                                                      
{'loss': 0.2623, 'grad_norm': 0.5773950815200806, 'learning_rate': 9.896120697508052e-06, 'epoch': 0.48}

 16%|█▌        | 2111/13317 [21:45<1:54:39,  1.63it/s]
 16%|█▌        | 2112/13317 [21:46<1:54:36,  1.63it/s]
                                                      
{'loss': 0.2456, 'grad_norm': 0.5267162322998047, 'learning_rate': 9.8958547575397e-06, 'epoch': 0.48}

 16%|█▌        | 2112/13317 [21:46<1:54:36,  1.63it/s]
 16%|█▌        | 2113/13317 [21:46<1:54:37,  1.63it/s]
                                                      
{'loss': 0.2633, 'grad_norm': 0.5084396600723267, 'learning_rate': 9.89558848117429e-06, 'epoch': 0.48}

 16%|█▌        | 2113/13317 [21:46<1:54:37,  1.63it/s]
 16%|█▌        | 2114/13317 [21:47<1:54:35,  1.63it/s]
                                                      
{'loss': 0.3087, 'grad_norm': 0.6199885010719299, 'learning_rate': 9.895321868430113e-06, 'epoch': 0.48}

 16%|█▌        | 2114/13317 [21:47<1:54:35,  1.63it/s]
 16%|█▌        | 2115/13317 [21:48<1:54:35,  1.63it/s]
                                                      
{'loss': 0.2598, 'grad_norm': 0.5928699374198914, 'learning_rate': 9.895054919325494e-06, 'epoch': 0.48}

 16%|█▌        | 2115/13317 [21:48<1:54:35,  1.63it/s]
 16%|█▌        | 2116/13317 [21:48<1:54:34,  1.63it/s]
                                                      
{'loss': 0.2354, 'grad_norm': 0.5028877854347229, 'learning_rate': 9.89478763387877e-06, 'epoch': 0.48}

 16%|█▌        | 2116/13317 [21:48<1:54:34,  1.63it/s]
 16%|█▌        | 2117/13317 [21:49<1:54:36,  1.63it/s]
                                                      
{'loss': 0.2895, 'grad_norm': 0.5750977993011475, 'learning_rate': 9.894520012108308e-06, 'epoch': 0.48}

 16%|█▌        | 2117/13317 [21:49<1:54:36,  1.63it/s]
 16%|█▌        | 2118/13317 [21:49<1:54:30,  1.63it/s]
                                                      
{'loss': 0.2314, 'grad_norm': 0.5803858637809753, 'learning_rate': 9.894252054032501e-06, 'epoch': 0.48}

 16%|█▌        | 2118/13317 [21:50<1:54:30,  1.63it/s]
 16%|█▌        | 2119/13317 [21:50<1:54:30,  1.63it/s]
                                                      
{'loss': 0.2587, 'grad_norm': 0.5581138134002686, 'learning_rate': 9.893983759669753e-06, 'epoch': 0.48}

 16%|█▌        | 2119/13317 [21:50<1:54:30,  1.63it/s]
 16%|█▌        | 2120/13317 [21:51<1:54:29,  1.63it/s]
                                                      
{'loss': 0.2546, 'grad_norm': 0.5595778822898865, 'learning_rate': 9.893715129038504e-06, 'epoch': 0.48}

 16%|█▌        | 2120/13317 [21:51<1:54:29,  1.63it/s]
 16%|█▌        | 2121/13317 [21:51<1:54:22,  1.63it/s]
                                                      
{'loss': 0.3571, 'grad_norm': 0.5978675484657288, 'learning_rate': 9.89344616215721e-06, 'epoch': 0.48}

 16%|█▌        | 2121/13317 [21:51<1:54:22,  1.63it/s]
 16%|█▌        | 2122/13317 [21:52<1:54:26,  1.63it/s]
                                                      
{'loss': 0.3443, 'grad_norm': 0.6195209622383118, 'learning_rate': 9.893176859044352e-06, 'epoch': 0.48}

 16%|█▌        | 2122/13317 [21:52<1:54:26,  1.63it/s]
 16%|█▌        | 2123/13317 [21:52<1:54:28,  1.63it/s]
                                                      
{'loss': 0.3091, 'grad_norm': 0.706692099571228, 'learning_rate': 9.892907219718435e-06, 'epoch': 0.48}

 16%|█▌        | 2123/13317 [21:53<1:54:28,  1.63it/s]
 16%|█▌        | 2124/13317 [21:53<1:54:26,  1.63it/s]
                                                      
{'loss': 0.2931, 'grad_norm': 0.6190691590309143, 'learning_rate': 9.892637244197983e-06, 'epoch': 0.48}

 16%|█▌        | 2124/13317 [21:53<1:54:26,  1.63it/s]
 16%|█▌        | 2125/13317 [21:54<1:54:24,  1.63it/s]
                                                      
{'loss': 0.2197, 'grad_norm': 0.6330646872520447, 'learning_rate': 9.892366932501549e-06, 'epoch': 0.48}

 16%|█▌        | 2125/13317 [21:54<1:54:24,  1.63it/s]
 16%|█▌        | 2126/13317 [21:54<1:54:24,  1.63it/s]
                                                      
{'loss': 0.3486, 'grad_norm': 0.6886559128761292, 'learning_rate': 9.892096284647706e-06, 'epoch': 0.48}

 16%|█▌        | 2126/13317 [21:54<1:54:24,  1.63it/s]
 16%|█▌        | 2127/13317 [21:55<1:54:23,  1.63it/s]
                                                      
{'loss': 0.3682, 'grad_norm': 0.6401602029800415, 'learning_rate': 9.891825300655048e-06, 'epoch': 0.48}

 16%|█▌        | 2127/13317 [21:55<1:54:23,  1.63it/s]
 16%|█▌        | 2128/13317 [21:56<1:54:21,  1.63it/s]
                                                      
{'loss': 0.3189, 'grad_norm': 0.6773753762245178, 'learning_rate': 9.891553980542199e-06, 'epoch': 0.48}

 16%|█▌        | 2128/13317 [21:56<1:54:21,  1.63it/s]
 16%|█▌        | 2129/13317 [21:56<1:54:21,  1.63it/s]
                                                      
{'loss': 0.3488, 'grad_norm': 0.5955212712287903, 'learning_rate': 9.891282324327798e-06, 'epoch': 0.48}

 16%|█▌        | 2129/13317 [21:56<1:54:21,  1.63it/s]
 16%|█▌        | 2130/13317 [21:57<1:54:18,  1.63it/s]
                                                      
{'loss': 0.3199, 'grad_norm': 0.6876688599586487, 'learning_rate': 9.891010332030513e-06, 'epoch': 0.48}

 16%|█▌        | 2130/13317 [21:57<1:54:18,  1.63it/s]
 16%|█▌        | 2131/13317 [21:57<1:54:14,  1.63it/s]
                                                      
{'loss': 0.2947, 'grad_norm': 0.6291860342025757, 'learning_rate': 9.890738003669029e-06, 'epoch': 0.48}

 16%|█▌        | 2131/13317 [21:58<1:54:14,  1.63it/s]
 16%|█▌        | 2132/13317 [21:58<1:54:21,  1.63it/s]
                                                      
{'loss': 0.2543, 'grad_norm': 0.6033007502555847, 'learning_rate': 9.890465339262061e-06, 'epoch': 0.48}

 16%|█▌        | 2132/13317 [21:58<1:54:21,  1.63it/s]
 16%|█▌        | 2133/13317 [21:59<1:54:23,  1.63it/s]
                                                      
{'loss': 0.3298, 'grad_norm': 0.6235203146934509, 'learning_rate': 9.890192338828345e-06, 'epoch': 0.48}

 16%|█▌        | 2133/13317 [21:59<1:54:23,  1.63it/s]
 16%|█▌        | 2134/13317 [21:59<1:54:22,  1.63it/s]
                                                      
{'loss': 0.2582, 'grad_norm': 0.5867703557014465, 'learning_rate': 9.889919002386634e-06, 'epoch': 0.48}

 16%|█▌        | 2134/13317 [21:59<1:54:22,  1.63it/s]
 16%|█▌        | 2135/13317 [22:00<1:54:23,  1.63it/s]
                                                      
{'loss': 0.2177, 'grad_norm': 0.46200641989707947, 'learning_rate': 9.889645329955712e-06, 'epoch': 0.48}

 16%|█▌        | 2135/13317 [22:00<1:54:23,  1.63it/s]
 16%|█▌        | 2136/13317 [22:00<1:54:21,  1.63it/s]
                                                      
{'loss': 0.2364, 'grad_norm': 0.5480154156684875, 'learning_rate': 9.889371321554386e-06, 'epoch': 0.48}

 16%|█▌        | 2136/13317 [22:01<1:54:21,  1.63it/s]
 16%|█▌        | 2137/13317 [22:01<1:54:23,  1.63it/s]
                                                      
{'loss': 0.3187, 'grad_norm': 0.5947777628898621, 'learning_rate': 9.88909697720148e-06, 'epoch': 0.48}

 16%|█▌        | 2137/13317 [22:01<1:54:23,  1.63it/s]
 16%|█▌        | 2138/13317 [22:02<1:54:25,  1.63it/s]
                                                      
{'loss': 0.2972, 'grad_norm': 0.5735918879508972, 'learning_rate': 9.888822296915846e-06, 'epoch': 0.48}

 16%|█▌        | 2138/13317 [22:02<1:54:25,  1.63it/s]
 16%|█▌        | 2139/13317 [22:02<1:54:23,  1.63it/s]
                                                      
{'loss': 0.2664, 'grad_norm': 0.5686464309692383, 'learning_rate': 9.888547280716353e-06, 'epoch': 0.48}

 16%|█▌        | 2139/13317 [22:02<1:54:23,  1.63it/s]
 16%|█▌        | 2140/13317 [22:03<1:54:19,  1.63it/s]
                                                      
{'loss': 0.3241, 'grad_norm': 0.5631359815597534, 'learning_rate': 9.888271928621904e-06, 'epoch': 0.48}

 16%|█▌        | 2140/13317 [22:03<1:54:19,  1.63it/s]
 16%|█▌        | 2141/13317 [22:04<1:54:19,  1.63it/s]
                                                      
{'loss': 0.3082, 'grad_norm': 0.6029781103134155, 'learning_rate': 9.887996240651417e-06, 'epoch': 0.48}

 16%|█▌        | 2141/13317 [22:04<1:54:19,  1.63it/s]
 16%|█▌        | 2142/13317 [22:04<1:54:22,  1.63it/s]
                                                      
{'loss': 0.3162, 'grad_norm': 0.6014645099639893, 'learning_rate': 9.887720216823831e-06, 'epoch': 0.48}

 16%|█▌        | 2142/13317 [22:04<1:54:22,  1.63it/s]
 16%|█▌        | 2143/13317 [22:05<1:54:22,  1.63it/s]
                                                      
{'loss': 0.4096, 'grad_norm': 0.7307090163230896, 'learning_rate': 9.887443857158114e-06, 'epoch': 0.48}

 16%|█▌        | 2143/13317 [22:05<1:54:22,  1.63it/s]
 16%|█▌        | 2144/13317 [22:05<1:54:19,  1.63it/s]
                                                      
{'loss': 0.2098, 'grad_norm': 0.4607015550136566, 'learning_rate': 9.887167161673254e-06, 'epoch': 0.48}

 16%|█▌        | 2144/13317 [22:05<1:54:19,  1.63it/s]
 16%|█▌        | 2145/13317 [22:06<1:54:17,  1.63it/s]
                                                      
{'loss': 0.257, 'grad_norm': 0.5642063021659851, 'learning_rate': 9.886890130388267e-06, 'epoch': 0.48}

 16%|█▌        | 2145/13317 [22:06<1:54:17,  1.63it/s]
 16%|█▌        | 2146/13317 [22:07<1:54:15,  1.63it/s]
                                                      
{'loss': 0.3397, 'grad_norm': 0.624472439289093, 'learning_rate': 9.886612763322183e-06, 'epoch': 0.48}

 16%|█▌        | 2146/13317 [22:07<1:54:15,  1.63it/s]
 16%|█▌        | 2147/13317 [22:07<1:54:15,  1.63it/s]
                                                      
{'loss': 0.2284, 'grad_norm': 0.5457583665847778, 'learning_rate': 9.886335060494061e-06, 'epoch': 0.48}

 16%|█▌        | 2147/13317 [22:07<1:54:15,  1.63it/s]
 16%|█▌        | 2148/13317 [22:08<1:54:13,  1.63it/s]
                                                      
{'loss': 0.2669, 'grad_norm': 0.6772779822349548, 'learning_rate': 9.886057021922984e-06, 'epoch': 0.48}

 16%|█▌        | 2148/13317 [22:08<1:54:13,  1.63it/s]
 16%|█▌        | 2149/13317 [22:08<1:54:13,  1.63it/s]
                                                      
{'loss': 0.2363, 'grad_norm': 0.5538378357887268, 'learning_rate': 9.885778647628055e-06, 'epoch': 0.48}

 16%|█▌        | 2149/13317 [22:09<1:54:13,  1.63it/s]
 16%|█▌        | 2150/13317 [22:09<1:54:48,  1.62it/s]
                                                      
{'loss': 0.2804, 'grad_norm': 0.5655017495155334, 'learning_rate': 9.885499937628402e-06, 'epoch': 0.48}

 16%|█▌        | 2150/13317 [22:09<1:54:48,  1.62it/s]
 16%|█▌        | 2151/13317 [22:10<1:54:38,  1.62it/s]
                                                      
{'loss': 0.3095, 'grad_norm': 0.5605271458625793, 'learning_rate': 9.885220891943174e-06, 'epoch': 0.48}

 16%|█▌        | 2151/13317 [22:10<1:54:38,  1.62it/s]
 16%|█▌        | 2152/13317 [22:10<1:54:28,  1.63it/s]
                                                      
{'loss': 0.2768, 'grad_norm': 0.6042520999908447, 'learning_rate': 9.884941510591547e-06, 'epoch': 0.48}

 16%|█▌        | 2152/13317 [22:10<1:54:28,  1.63it/s]
 16%|█▌        | 2153/13317 [22:11<1:54:27,  1.63it/s]
                                                      
{'loss': 0.2232, 'grad_norm': 0.5467307567596436, 'learning_rate': 9.884661793592714e-06, 'epoch': 0.49}

 16%|█▌        | 2153/13317 [22:11<1:54:27,  1.63it/s]
 16%|█▌        | 2154/13317 [22:12<1:54:23,  1.63it/s]
                                                      
{'loss': 0.4354, 'grad_norm': 0.7750148177146912, 'learning_rate': 9.884381740965895e-06, 'epoch': 0.49}

 16%|█▌        | 2154/13317 [22:12<1:54:23,  1.63it/s]
 16%|█▌        | 2155/13317 [22:12<1:54:16,  1.63it/s]
                                                      
{'loss': 0.3002, 'grad_norm': 0.6211086511611938, 'learning_rate': 9.884101352730335e-06, 'epoch': 0.49}

 16%|█▌        | 2155/13317 [22:12<1:54:16,  1.63it/s]
 16%|█▌        | 2156/13317 [22:13<1:54:20,  1.63it/s]
                                                      
{'loss': 0.2714, 'grad_norm': 0.6379814743995667, 'learning_rate': 9.883820628905299e-06, 'epoch': 0.49}

 16%|█▌        | 2156/13317 [22:13<1:54:20,  1.63it/s]
 16%|█▌        | 2157/13317 [22:13<1:54:25,  1.63it/s]
                                                      
{'loss': 0.2255, 'grad_norm': 0.5785325169563293, 'learning_rate': 9.883539569510074e-06, 'epoch': 0.49}

 16%|█▌        | 2157/13317 [22:13<1:54:25,  1.63it/s]
 16%|█▌        | 2158/13317 [22:14<1:54:17,  1.63it/s]
                                                      
{'loss': 0.314, 'grad_norm': 0.6287438273429871, 'learning_rate': 9.883258174563974e-06, 'epoch': 0.49}

 16%|█▌        | 2158/13317 [22:14<1:54:17,  1.63it/s]
 16%|█▌        | 2159/13317 [22:15<1:54:22,  1.63it/s]
                                                      
{'loss': 0.3255, 'grad_norm': 0.6405089497566223, 'learning_rate': 9.882976444086331e-06, 'epoch': 0.49}

 16%|█▌        | 2159/13317 [22:15<1:54:22,  1.63it/s]
 16%|█▌        | 2160/13317 [22:15<1:54:16,  1.63it/s]
                                                      
{'loss': 0.2201, 'grad_norm': 0.8842615485191345, 'learning_rate': 9.882694378096505e-06, 'epoch': 0.49}

 16%|█▌        | 2160/13317 [22:15<1:54:16,  1.63it/s]
 16%|█▌        | 2161/13317 [22:16<1:54:19,  1.63it/s]
                                                      
{'loss': 0.2367, 'grad_norm': 0.5100123882293701, 'learning_rate': 9.882411976613878e-06, 'epoch': 0.49}

 16%|█▌        | 2161/13317 [22:16<1:54:19,  1.63it/s]
 16%|█▌        | 2162/13317 [22:16<1:54:25,  1.62it/s]
                                                      
{'loss': 0.3634, 'grad_norm': 0.6049882769584656, 'learning_rate': 9.882129239657851e-06, 'epoch': 0.49}

 16%|█▌        | 2162/13317 [22:17<1:54:25,  1.62it/s]
 16%|█▌        | 2163/13317 [22:17<1:54:16,  1.63it/s]
                                                      
{'loss': 0.2909, 'grad_norm': 0.5776451826095581, 'learning_rate': 9.881846167247853e-06, 'epoch': 0.49}

 16%|█▌        | 2163/13317 [22:17<1:54:16,  1.63it/s]
 16%|█▌        | 2164/13317 [22:18<1:54:22,  1.63it/s]
                                                      
{'loss': 0.2911, 'grad_norm': 0.6416159272193909, 'learning_rate': 9.881562759403333e-06, 'epoch': 0.49}

 16%|█▌        | 2164/13317 [22:18<1:54:22,  1.63it/s]
 16%|█▋        | 2165/13317 [22:18<1:54:17,  1.63it/s]
                                                      
{'loss': 0.2618, 'grad_norm': 0.5905357003211975, 'learning_rate': 9.881279016143766e-06, 'epoch': 0.49}

 16%|█▋        | 2165/13317 [22:18<1:54:17,  1.63it/s]
 16%|█▋        | 2166/13317 [22:19<1:54:19,  1.63it/s]
                                                      
{'loss': 0.2299, 'grad_norm': 0.5213228464126587, 'learning_rate': 9.880994937488644e-06, 'epoch': 0.49}

 16%|█▋        | 2166/13317 [22:19<1:54:19,  1.63it/s]
 16%|█▋        | 2167/13317 [22:20<1:54:25,  1.62it/s]
                                                      
{'loss': 0.3092, 'grad_norm': 0.6188309192657471, 'learning_rate': 9.88071052345749e-06, 'epoch': 0.49}

 16%|█▋        | 2167/13317 [22:20<1:54:25,  1.62it/s]
 16%|█▋        | 2168/13317 [22:20<1:54:17,  1.63it/s]
                                                      
{'loss': 0.3206, 'grad_norm': 0.7189080715179443, 'learning_rate': 9.880425774069845e-06, 'epoch': 0.49}

 16%|█▋        | 2168/13317 [22:20<1:54:17,  1.63it/s]
 16%|█▋        | 2169/13317 [22:21<1:54:18,  1.63it/s]
                                                      
{'loss': 0.286, 'grad_norm': 0.5846171975135803, 'learning_rate': 9.880140689345277e-06, 'epoch': 0.49}

 16%|█▋        | 2169/13317 [22:21<1:54:18,  1.63it/s]
 16%|█▋        | 2170/13317 [22:21<1:54:09,  1.63it/s]
                                                      
{'loss': 0.2444, 'grad_norm': 0.55199134349823, 'learning_rate': 9.879855269303372e-06, 'epoch': 0.49}

 16%|█▋        | 2170/13317 [22:21<1:54:09,  1.63it/s]
 16%|█▋        | 2171/13317 [22:22<1:54:15,  1.63it/s]
                                                      
{'loss': 0.2509, 'grad_norm': 0.520616352558136, 'learning_rate': 9.879569513963739e-06, 'epoch': 0.49}

 16%|█▋        | 2171/13317 [22:22<1:54:15,  1.63it/s]
 16%|█▋        | 2172/13317 [22:23<1:54:19,  1.62it/s]
                                                      
{'loss': 0.3035, 'grad_norm': 0.5865454077720642, 'learning_rate': 9.879283423346014e-06, 'epoch': 0.49}

 16%|█▋        | 2172/13317 [22:23<1:54:19,  1.62it/s]
 16%|█▋        | 2173/13317 [22:23<1:54:12,  1.63it/s]
                                                      
{'loss': 0.2801, 'grad_norm': 0.5743669271469116, 'learning_rate': 9.878996997469857e-06, 'epoch': 0.49}

 16%|█▋        | 2173/13317 [22:23<1:54:12,  1.63it/s]
 16%|█▋        | 2174/13317 [22:24<1:54:05,  1.63it/s]
                                                      
{'loss': 0.2656, 'grad_norm': 0.5758652687072754, 'learning_rate': 9.878710236354945e-06, 'epoch': 0.49}

 16%|█▋        | 2174/13317 [22:24<1:54:05,  1.63it/s]
 16%|█▋        | 2175/13317 [22:24<1:54:00,  1.63it/s]
                                                      
{'loss': 0.3179, 'grad_norm': 0.8084157109260559, 'learning_rate': 9.878423140020985e-06, 'epoch': 0.49}

 16%|█▋        | 2175/13317 [22:25<1:54:00,  1.63it/s]
 16%|█▋        | 2176/13317 [22:25<1:54:08,  1.63it/s]
                                                      
{'loss': 0.357, 'grad_norm': 0.6072846055030823, 'learning_rate': 9.8781357084877e-06, 'epoch': 0.49}

 16%|█▋        | 2176/13317 [22:25<1:54:08,  1.63it/s]
 16%|█▋        | 2177/13317 [22:26<1:54:09,  1.63it/s]
                                                      
{'loss': 0.2851, 'grad_norm': 0.5596015453338623, 'learning_rate': 9.877847941774844e-06, 'epoch': 0.49}

 16%|█▋        | 2177/13317 [22:26<1:54:09,  1.63it/s]
 16%|█▋        | 2178/13317 [22:26<1:54:05,  1.63it/s]
                                                      
{'loss': 0.3373, 'grad_norm': 0.6691579222679138, 'learning_rate': 9.877559839902185e-06, 'epoch': 0.49}

 16%|█▋        | 2178/13317 [22:26<1:54:05,  1.63it/s]
 16%|█▋        | 2179/13317 [22:27<1:54:06,  1.63it/s]
                                                      
{'loss': 0.2695, 'grad_norm': 0.6092930436134338, 'learning_rate': 9.877271402889519e-06, 'epoch': 0.49}

 16%|█▋        | 2179/13317 [22:27<1:54:06,  1.63it/s]
 16%|█▋        | 2180/13317 [22:27<1:54:01,  1.63it/s]
                                                      
{'loss': 0.2864, 'grad_norm': 0.577244758605957, 'learning_rate': 9.876982630756668e-06, 'epoch': 0.49}

 16%|█▋        | 2180/13317 [22:28<1:54:01,  1.63it/s]
 16%|█▋        | 2181/13317 [22:28<1:54:00,  1.63it/s]
                                                      
{'loss': 0.244, 'grad_norm': 0.5042018294334412, 'learning_rate': 9.876693523523472e-06, 'epoch': 0.49}

 16%|█▋        | 2181/13317 [22:28<1:54:00,  1.63it/s]
 16%|█▋        | 2182/13317 [22:29<1:54:04,  1.63it/s]
                                                      
{'loss': 0.3114, 'grad_norm': 0.5875144600868225, 'learning_rate': 9.876404081209796e-06, 'epoch': 0.49}

 16%|█▋        | 2182/13317 [22:29<1:54:04,  1.63it/s]
 16%|█▋        | 2183/13317 [22:29<1:53:59,  1.63it/s]
                                                      
{'loss': 0.2294, 'grad_norm': 0.5020793676376343, 'learning_rate': 9.876114303835526e-06, 'epoch': 0.49}

 16%|█▋        | 2183/13317 [22:29<1:53:59,  1.63it/s]
 16%|█▋        | 2184/13317 [22:30<1:53:57,  1.63it/s]
                                                      
{'loss': 0.2292, 'grad_norm': 0.5704085826873779, 'learning_rate': 9.875824191420576e-06, 'epoch': 0.49}

 16%|█▋        | 2184/13317 [22:30<1:53:57,  1.63it/s]
 16%|█▋        | 2185/13317 [22:31<1:53:50,  1.63it/s]
                                                      
{'loss': 0.3688, 'grad_norm': 0.7162659168243408, 'learning_rate': 9.875533743984878e-06, 'epoch': 0.49}

 16%|█▋        | 2185/13317 [22:31<1:53:50,  1.63it/s]
 16%|█▋        | 2186/13317 [22:31<1:53:51,  1.63it/s]
                                                      
{'loss': 0.2879, 'grad_norm': 0.6059628129005432, 'learning_rate': 9.875242961548386e-06, 'epoch': 0.49}

 16%|█▋        | 2186/13317 [22:31<1:53:51,  1.63it/s]
 16%|█▋        | 2187/13317 [22:32<1:53:54,  1.63it/s]
                                                      
{'loss': 0.2922, 'grad_norm': 0.5637176036834717, 'learning_rate': 9.874951844131086e-06, 'epoch': 0.49}

 16%|█▋        | 2187/13317 [22:32<1:53:54,  1.63it/s]
 16%|█▋        | 2188/13317 [22:32<1:53:55,  1.63it/s]
                                                      
{'loss': 0.2265, 'grad_norm': 0.503640353679657, 'learning_rate': 9.874660391752975e-06, 'epoch': 0.49}

 16%|█▋        | 2188/13317 [22:33<1:53:55,  1.63it/s]
 16%|█▋        | 2189/13317 [22:33<1:53:55,  1.63it/s]
                                                      
{'loss': 0.2207, 'grad_norm': 0.5097084641456604, 'learning_rate': 9.874368604434084e-06, 'epoch': 0.49}

 16%|█▋        | 2189/13317 [22:33<1:53:55,  1.63it/s]
 16%|█▋        | 2190/13317 [22:34<1:53:53,  1.63it/s]
                                                      
{'loss': 0.2506, 'grad_norm': 0.5715931057929993, 'learning_rate': 9.874076482194456e-06, 'epoch': 0.49}

 16%|█▋        | 2190/13317 [22:34<1:53:53,  1.63it/s]
 16%|█▋        | 2191/13317 [22:34<1:53:53,  1.63it/s]
                                                      
{'loss': 0.3087, 'grad_norm': 0.7025182843208313, 'learning_rate': 9.873784025054169e-06, 'epoch': 0.49}

 16%|█▋        | 2191/13317 [22:34<1:53:53,  1.63it/s]
 16%|█▋        | 2192/13317 [22:35<1:53:53,  1.63it/s]
                                                      
{'loss': 0.3, 'grad_norm': 0.6897812485694885, 'learning_rate': 9.873491233033314e-06, 'epoch': 0.49}

 16%|█▋        | 2192/13317 [22:35<1:53:53,  1.63it/s]
 16%|█▋        | 2193/13317 [22:35<1:54:06,  1.62it/s]
                                                      
{'loss': 0.2565, 'grad_norm': 0.5496415495872498, 'learning_rate': 9.873198106152009e-06, 'epoch': 0.49}

 16%|█▋        | 2193/13317 [22:36<1:54:06,  1.62it/s]
 16%|█▋        | 2194/13317 [22:36<1:54:12,  1.62it/s]
                                                      
{'loss': 0.2145, 'grad_norm': 0.45043566823005676, 'learning_rate': 9.872904644430397e-06, 'epoch': 0.49}

 16%|█▋        | 2194/13317 [22:36<1:54:12,  1.62it/s]
 16%|█▋        | 2195/13317 [22:37<1:54:05,  1.62it/s]
                                                      
{'loss': 0.2402, 'grad_norm': 0.5546428561210632, 'learning_rate': 9.87261084788864e-06, 'epoch': 0.49}

 16%|█▋        | 2195/13317 [22:37<1:54:05,  1.62it/s]
 16%|█▋        | 2196/13317 [22:37<1:53:54,  1.63it/s]
                                                      
{'loss': 0.2644, 'grad_norm': 0.6016594767570496, 'learning_rate': 9.872316716546926e-06, 'epoch': 0.49}

 16%|█▋        | 2196/13317 [22:37<1:53:54,  1.63it/s]
 16%|█▋        | 2197/13317 [22:38<1:53:54,  1.63it/s]
                                                      
{'loss': 0.2091, 'grad_norm': 0.5002607107162476, 'learning_rate': 9.872022250425465e-06, 'epoch': 0.49}

 16%|█▋        | 2197/13317 [22:38<1:53:54,  1.63it/s]
 17%|█▋        | 2198/13317 [22:39<1:53:49,  1.63it/s]
                                                      
{'loss': 0.3337, 'grad_norm': 0.7418053150177002, 'learning_rate': 9.87172744954449e-06, 'epoch': 0.5}

 17%|█▋        | 2198/13317 [22:39<1:53:49,  1.63it/s]
 17%|█▋        | 2199/13317 [22:39<1:53:55,  1.63it/s]
                                                      
{'loss': 0.2332, 'grad_norm': 0.5373680591583252, 'learning_rate': 9.871432313924255e-06, 'epoch': 0.5}

 17%|█▋        | 2199/13317 [22:39<1:53:55,  1.63it/s]
 17%|█▋        | 2200/13317 [22:40<1:53:55,  1.63it/s]
                                                      
{'loss': 0.2944, 'grad_norm': 0.6239293813705444, 'learning_rate': 9.871136843585042e-06, 'epoch': 0.5}

 17%|█▋        | 2200/13317 [22:40<1:53:55,  1.63it/s]
 17%|█▋        | 2201/13317 [22:40<1:53:53,  1.63it/s]
                                                      
{'loss': 0.2186, 'grad_norm': 0.4780564606189728, 'learning_rate': 9.870841038547149e-06, 'epoch': 0.5}

 17%|█▋        | 2201/13317 [22:41<1:53:53,  1.63it/s]
 17%|█▋        | 2202/13317 [22:41<1:53:59,  1.63it/s]
                                                      
{'loss': 0.249, 'grad_norm': 0.506135880947113, 'learning_rate': 9.870544898830908e-06, 'epoch': 0.5}

 17%|█▋        | 2202/13317 [22:41<1:53:59,  1.63it/s]
 17%|█▋        | 2203/13317 [22:42<1:53:51,  1.63it/s]
                                                      
{'loss': 0.3029, 'grad_norm': 0.6319773197174072, 'learning_rate': 9.870248424456659e-06, 'epoch': 0.5}

 17%|█▋        | 2203/13317 [22:42<1:53:51,  1.63it/s]
 17%|█▋        | 2204/13317 [22:42<1:53:54,  1.63it/s]
                                                      
{'loss': 0.2662, 'grad_norm': 0.5233315825462341, 'learning_rate': 9.869951615444774e-06, 'epoch': 0.5}

 17%|█▋        | 2204/13317 [22:42<1:53:54,  1.63it/s]
 17%|█▋        | 2205/13317 [22:43<1:53:45,  1.63it/s]
                                                      
{'loss': 0.3263, 'grad_norm': 0.5745076537132263, 'learning_rate': 9.869654471815653e-06, 'epoch': 0.5}

 17%|█▋        | 2205/13317 [22:43<1:53:45,  1.63it/s]
 17%|█▋        | 2206/13317 [22:43<1:53:44,  1.63it/s]
                                                      
{'loss': 0.5071, 'grad_norm': 0.7381264567375183, 'learning_rate': 9.869356993589709e-06, 'epoch': 0.5}

 17%|█▋        | 2206/13317 [22:44<1:53:44,  1.63it/s]
 17%|█▋        | 2207/13317 [22:44<1:53:49,  1.63it/s]
                                                      
{'loss': 0.267, 'grad_norm': 0.6287693977355957, 'learning_rate': 9.86905918078738e-06, 'epoch': 0.5}

 17%|█▋        | 2207/13317 [22:44<1:53:49,  1.63it/s]
 17%|█▋        | 2208/13317 [22:45<1:53:47,  1.63it/s]
                                                      
{'loss': 0.251, 'grad_norm': 0.5520833730697632, 'learning_rate': 9.868761033429132e-06, 'epoch': 0.5}

 17%|█▋        | 2208/13317 [22:45<1:53:47,  1.63it/s]
 17%|█▋        | 2209/13317 [22:45<1:53:42,  1.63it/s]
                                                      
{'loss': 0.3645, 'grad_norm': 0.688110888004303, 'learning_rate': 9.86846255153545e-06, 'epoch': 0.5}

 17%|█▋        | 2209/13317 [22:45<1:53:42,  1.63it/s]
 17%|█▋        | 2210/13317 [22:46<1:53:39,  1.63it/s]
                                                      
{'loss': 0.3338, 'grad_norm': 0.6496300101280212, 'learning_rate': 9.86816373512684e-06, 'epoch': 0.5}

 17%|█▋        | 2210/13317 [22:46<1:53:39,  1.63it/s]
 17%|█▋        | 2211/13317 [22:47<1:53:32,  1.63it/s]
                                                      
{'loss': 0.3143, 'grad_norm': 0.6685534715652466, 'learning_rate': 9.86786458422384e-06, 'epoch': 0.5}

 17%|█▋        | 2211/13317 [22:47<1:53:32,  1.63it/s]
 17%|█▋        | 2212/13317 [22:47<1:53:35,  1.63it/s]
                                                      
{'loss': 0.2661, 'grad_norm': 0.5884599089622498, 'learning_rate': 9.867565098847e-06, 'epoch': 0.5}

 17%|█▋        | 2212/13317 [22:47<1:53:35,  1.63it/s]
 17%|█▋        | 2213/13317 [22:48<1:53:33,  1.63it/s]
                                                      
{'loss': 0.3089, 'grad_norm': 0.5485038161277771, 'learning_rate': 9.867265279016899e-06, 'epoch': 0.5}

 17%|█▋        | 2213/13317 [22:48<1:53:33,  1.63it/s]
 17%|█▋        | 2214/13317 [22:48<1:53:32,  1.63it/s]
                                                      
{'loss': 0.2734, 'grad_norm': 0.5997791886329651, 'learning_rate': 9.866965124754138e-06, 'epoch': 0.5}

 17%|█▋        | 2214/13317 [22:49<1:53:32,  1.63it/s]
 17%|█▋        | 2215/13317 [22:49<1:53:33,  1.63it/s]
                                                      
{'loss': 0.4163, 'grad_norm': 0.7114030122756958, 'learning_rate': 9.866664636079339e-06, 'epoch': 0.5}

 17%|█▋        | 2215/13317 [22:49<1:53:33,  1.63it/s]
 17%|█▋        | 2216/13317 [22:50<1:53:33,  1.63it/s]
                                                      
{'loss': 0.2608, 'grad_norm': 0.5368509292602539, 'learning_rate': 9.866363813013153e-06, 'epoch': 0.5}

 17%|█▋        | 2216/13317 [22:50<1:53:33,  1.63it/s]
 17%|█▋        | 2217/13317 [22:50<1:53:38,  1.63it/s]
                                                      
{'loss': 0.3128, 'grad_norm': 0.6802924275398254, 'learning_rate': 9.866062655576245e-06, 'epoch': 0.5}

 17%|█▋        | 2217/13317 [22:50<1:53:38,  1.63it/s]
 17%|█▋        | 2218/13317 [22:51<1:53:34,  1.63it/s]
                                                      
{'loss': 0.319, 'grad_norm': 0.6169158816337585, 'learning_rate': 9.865761163789312e-06, 'epoch': 0.5}

 17%|█▋        | 2218/13317 [22:51<1:53:34,  1.63it/s]
 17%|█▋        | 2219/13317 [22:51<1:53:32,  1.63it/s]
                                                      
{'loss': 0.255, 'grad_norm': 0.5240515470504761, 'learning_rate': 9.865459337673066e-06, 'epoch': 0.5}

 17%|█▋        | 2219/13317 [22:52<1:53:32,  1.63it/s]
 17%|█▋        | 2220/13317 [22:52<1:53:28,  1.63it/s]
                                                      
{'loss': 0.2736, 'grad_norm': 0.5335531830787659, 'learning_rate': 9.865157177248248e-06, 'epoch': 0.5}

 17%|█▋        | 2220/13317 [22:52<1:53:28,  1.63it/s]
 17%|█▋        | 2221/13317 [22:53<1:53:32,  1.63it/s]
                                                      
{'loss': 0.2538, 'grad_norm': 0.5413637757301331, 'learning_rate': 9.864854682535619e-06, 'epoch': 0.5}

 17%|█▋        | 2221/13317 [22:53<1:53:32,  1.63it/s]
 17%|█▋        | 2222/13317 [22:53<1:53:43,  1.63it/s]
                                                      
{'loss': 0.3208, 'grad_norm': 0.6550132632255554, 'learning_rate': 9.864551853555961e-06, 'epoch': 0.5}

 17%|█▋        | 2222/13317 [22:53<1:53:43,  1.63it/s]
 17%|█▋        | 2223/13317 [22:54<1:53:38,  1.63it/s]
                                                      
{'loss': 0.2405, 'grad_norm': 0.5300983786582947, 'learning_rate': 9.864248690330088e-06, 'epoch': 0.5}

 17%|█▋        | 2223/13317 [22:54<1:53:38,  1.63it/s]
 17%|█▋        | 2224/13317 [22:55<1:53:33,  1.63it/s]
                                                      
{'loss': 0.2435, 'grad_norm': 0.5990498661994934, 'learning_rate': 9.863945192878823e-06, 'epoch': 0.5}

 17%|█▋        | 2224/13317 [22:55<1:53:33,  1.63it/s]
 17%|█▋        | 2225/13317 [22:55<1:53:31,  1.63it/s]
                                                      
{'loss': 0.3012, 'grad_norm': 0.5846682786941528, 'learning_rate': 9.863641361223025e-06, 'epoch': 0.5}

 17%|█▋        | 2225/13317 [22:55<1:53:31,  1.63it/s]
 17%|█▋        | 2226/13317 [22:56<1:53:26,  1.63it/s]
                                                      
{'loss': 0.2477, 'grad_norm': 0.5477436184883118, 'learning_rate': 9.863337195383567e-06, 'epoch': 0.5}

 17%|█▋        | 2226/13317 [22:56<1:53:26,  1.63it/s]
 17%|█▋        | 2227/13317 [22:56<1:53:33,  1.63it/s]
                                                      
{'loss': 0.2678, 'grad_norm': 0.5316845178604126, 'learning_rate': 9.863032695381352e-06, 'epoch': 0.5}

 17%|█▋        | 2227/13317 [22:57<1:53:33,  1.63it/s]
 17%|█▋        | 2228/13317 [22:57<1:53:29,  1.63it/s]
                                                      
{'loss': 0.2275, 'grad_norm': 0.4822985529899597, 'learning_rate': 9.862727861237299e-06, 'epoch': 0.5}

 17%|█▋        | 2228/13317 [22:57<1:53:29,  1.63it/s]
 17%|█▋        | 2229/13317 [22:58<1:53:26,  1.63it/s]
                                                      
{'loss': 0.2056, 'grad_norm': 0.47822853922843933, 'learning_rate': 9.862422692972353e-06, 'epoch': 0.5}

 17%|█▋        | 2229/13317 [22:58<1:53:26,  1.63it/s]
 17%|█▋        | 2230/13317 [22:58<1:53:25,  1.63it/s]
                                                      
{'loss': 0.2945, 'grad_norm': 0.5783000588417053, 'learning_rate': 9.862117190607485e-06, 'epoch': 0.5}

 17%|█▋        | 2230/13317 [22:58<1:53:25,  1.63it/s]
 17%|█▋        | 2231/13317 [22:59<1:53:22,  1.63it/s]
                                                      
{'loss': 0.3578, 'grad_norm': 0.7058320641517639, 'learning_rate': 9.861811354163685e-06, 'epoch': 0.5}

 17%|█▋        | 2231/13317 [22:59<1:53:22,  1.63it/s]
 17%|█▋        | 2232/13317 [22:59<1:53:25,  1.63it/s]
                                                      
{'loss': 0.2912, 'grad_norm': 0.6229597330093384, 'learning_rate': 9.861505183661968e-06, 'epoch': 0.5}

 17%|█▋        | 2232/13317 [23:00<1:53:25,  1.63it/s]
 17%|█▋        | 2233/13317 [23:00<1:53:23,  1.63it/s]
                                                      
{'loss': 0.3098, 'grad_norm': 0.608830988407135, 'learning_rate': 9.86119867912337e-06, 'epoch': 0.5}

 17%|█▋        | 2233/13317 [23:00<1:53:23,  1.63it/s]
 17%|█▋        | 2234/13317 [23:01<1:53:29,  1.63it/s]
                                                      
{'loss': 0.2442, 'grad_norm': 0.5142486095428467, 'learning_rate': 9.860891840568952e-06, 'epoch': 0.5}

 17%|█▋        | 2234/13317 [23:01<1:53:29,  1.63it/s]
 17%|█▋        | 2235/13317 [23:01<1:53:29,  1.63it/s]
                                                      
{'loss': 0.2974, 'grad_norm': 0.6216627359390259, 'learning_rate': 9.860584668019794e-06, 'epoch': 0.5}

 17%|█▋        | 2235/13317 [23:01<1:53:29,  1.63it/s]
 17%|█▋        | 2236/13317 [23:02<1:53:26,  1.63it/s]
                                                      
{'loss': 0.3234, 'grad_norm': 0.5873440504074097, 'learning_rate': 9.860277161497007e-06, 'epoch': 0.5}

 17%|█▋        | 2236/13317 [23:02<1:53:26,  1.63it/s]
 17%|█▋        | 2237/13317 [23:03<1:53:27,  1.63it/s]
                                                      
{'loss': 0.2256, 'grad_norm': 0.5553790330886841, 'learning_rate': 9.859969321021718e-06, 'epoch': 0.5}

 17%|█▋        | 2237/13317 [23:03<1:53:27,  1.63it/s]
 17%|█▋        | 2238/13317 [23:03<1:53:24,  1.63it/s]
                                                      
{'loss': 0.2545, 'grad_norm': 0.5313760638237, 'learning_rate': 9.859661146615079e-06, 'epoch': 0.5}

 17%|█▋        | 2238/13317 [23:03<1:53:24,  1.63it/s]
 17%|█▋        | 2239/13317 [23:04<1:53:21,  1.63it/s]
                                                      
{'loss': 0.2255, 'grad_norm': 0.5682799220085144, 'learning_rate': 9.859352638298264e-06, 'epoch': 0.5}

 17%|█▋        | 2239/13317 [23:04<1:53:21,  1.63it/s]
 17%|█▋        | 2240/13317 [23:04<1:53:23,  1.63it/s]
                                                      
{'loss': 0.2629, 'grad_norm': 0.5060707330703735, 'learning_rate': 9.85904379609247e-06, 'epoch': 0.5}

 17%|█▋        | 2240/13317 [23:04<1:53:23,  1.63it/s]
 17%|█▋        | 2241/13317 [23:05<1:53:19,  1.63it/s]
                                                      
{'loss': 0.2443, 'grad_norm': 0.4563635289669037, 'learning_rate': 9.858734620018918e-06, 'epoch': 0.5}

 17%|█▋        | 2241/13317 [23:05<1:53:19,  1.63it/s]
 17%|█▋        | 2242/13317 [23:06<1:53:28,  1.63it/s]
                                                      
{'loss': 0.2895, 'grad_norm': 0.5686892867088318, 'learning_rate': 9.858425110098852e-06, 'epoch': 0.51}

 17%|█▋        | 2242/13317 [23:06<1:53:28,  1.63it/s]
 17%|█▋        | 2243/13317 [23:06<1:53:22,  1.63it/s]
                                                      
{'loss': 0.2889, 'grad_norm': 0.6875563859939575, 'learning_rate': 9.858115266353541e-06, 'epoch': 0.51}

 17%|█▋        | 2243/13317 [23:06<1:53:22,  1.63it/s]
 17%|█▋        | 2244/13317 [23:07<1:53:16,  1.63it/s]
                                                      
{'loss': 0.3115, 'grad_norm': 0.6379311084747314, 'learning_rate': 9.857805088804274e-06, 'epoch': 0.51}

 17%|█▋        | 2244/13317 [23:07<1:53:16,  1.63it/s]
 17%|█▋        | 2245/13317 [23:07<1:53:14,  1.63it/s]
                                                      
{'loss': 0.3193, 'grad_norm': 0.5536683201789856, 'learning_rate': 9.857494577472359e-06, 'epoch': 0.51}

 17%|█▋        | 2245/13317 [23:08<1:53:14,  1.63it/s]
 17%|█▋        | 2246/13317 [23:08<1:53:10,  1.63it/s]
                                                      
{'loss': 0.2799, 'grad_norm': 0.6587669253349304, 'learning_rate': 9.857183732379136e-06, 'epoch': 0.51}

 17%|█▋        | 2246/13317 [23:08<1:53:10,  1.63it/s]
 17%|█▋        | 2247/13317 [23:09<1:53:17,  1.63it/s]
                                                      
{'loss': 0.2673, 'grad_norm': 0.5278927087783813, 'learning_rate': 9.856872553545962e-06, 'epoch': 0.51}

 17%|█▋        | 2247/13317 [23:09<1:53:17,  1.63it/s]
 17%|█▋        | 2248/13317 [23:09<1:53:17,  1.63it/s]
                                                      
{'loss': 0.2968, 'grad_norm': 0.6353583931922913, 'learning_rate': 9.856561040994218e-06, 'epoch': 0.51}

 17%|█▋        | 2248/13317 [23:09<1:53:17,  1.63it/s]
 17%|█▋        | 2249/13317 [23:10<1:53:19,  1.63it/s]
                                                      
{'loss': 0.223, 'grad_norm': 0.47413358092308044, 'learning_rate': 9.856249194745309e-06, 'epoch': 0.51}

 17%|█▋        | 2249/13317 [23:10<1:53:19,  1.63it/s]
 17%|█▋        | 2250/13317 [23:10<1:53:17,  1.63it/s]
                                                      
{'loss': 0.2446, 'grad_norm': 0.5056502819061279, 'learning_rate': 9.85593701482066e-06, 'epoch': 0.51}

 17%|█▋        | 2250/13317 [23:11<1:53:17,  1.63it/s]
 17%|█▋        | 2251/13317 [23:11<1:53:13,  1.63it/s]
                                                      
{'loss': 0.3031, 'grad_norm': 0.6154353022575378, 'learning_rate': 9.855624501241723e-06, 'epoch': 0.51}

 17%|█▋        | 2251/13317 [23:11<1:53:13,  1.63it/s]
 17%|█▋        | 2252/13317 [23:12<1:53:09,  1.63it/s]
                                                      
{'loss': 0.2176, 'grad_norm': 0.49970194697380066, 'learning_rate': 9.855311654029972e-06, 'epoch': 0.51}

 17%|█▋        | 2252/13317 [23:12<1:53:09,  1.63it/s]
 17%|█▋        | 2253/13317 [23:12<1:53:09,  1.63it/s]
                                                      
{'loss': 0.217, 'grad_norm': 0.5084220767021179, 'learning_rate': 9.8549984732069e-06, 'epoch': 0.51}

 17%|█▋        | 2253/13317 [23:12<1:53:09,  1.63it/s]
 17%|█▋        | 2254/13317 [23:13<1:53:10,  1.63it/s]
                                                      
{'loss': 0.2543, 'grad_norm': 0.5614343285560608, 'learning_rate': 9.854684958794029e-06, 'epoch': 0.51}

 17%|█▋        | 2254/13317 [23:13<1:53:10,  1.63it/s]
 17%|█▋        | 2255/13317 [23:14<1:53:11,  1.63it/s]
                                                      
{'loss': 0.2292, 'grad_norm': 0.4833372235298157, 'learning_rate': 9.854371110812896e-06, 'epoch': 0.51}

 17%|█▋        | 2255/13317 [23:14<1:53:11,  1.63it/s]
 17%|█▋        | 2256/13317 [23:14<1:53:15,  1.63it/s]
                                                      
{'loss': 0.304, 'grad_norm': 0.5764563679695129, 'learning_rate': 9.854056929285072e-06, 'epoch': 0.51}

 17%|█▋        | 2256/13317 [23:14<1:53:15,  1.63it/s]
 17%|█▋        | 2257/13317 [23:15<1:53:10,  1.63it/s]
                                                      
{'loss': 0.2868, 'grad_norm': 0.5771557688713074, 'learning_rate': 9.85374241423214e-06, 'epoch': 0.51}

 17%|█▋        | 2257/13317 [23:15<1:53:10,  1.63it/s]
 17%|█▋        | 2258/13317 [23:15<1:53:08,  1.63it/s]
                                                      
{'loss': 0.243, 'grad_norm': 0.5017721056938171, 'learning_rate': 9.853427565675712e-06, 'epoch': 0.51}

 17%|█▋        | 2258/13317 [23:16<1:53:08,  1.63it/s]
 17%|█▋        | 2259/13317 [23:16<1:53:12,  1.63it/s]
                                                      
{'loss': 0.3997, 'grad_norm': 0.6711179614067078, 'learning_rate': 9.853112383637422e-06, 'epoch': 0.51}

 17%|█▋        | 2259/13317 [23:16<1:53:12,  1.63it/s]
 17%|█▋        | 2260/13317 [23:17<1:53:08,  1.63it/s]
                                                      
{'loss': 0.2947, 'grad_norm': 0.5401800274848938, 'learning_rate': 9.852796868138925e-06, 'epoch': 0.51}

 17%|█▋        | 2260/13317 [23:17<1:53:08,  1.63it/s]
 17%|█▋        | 2261/13317 [23:17<1:53:02,  1.63it/s]
                                                      
{'loss': 0.3015, 'grad_norm': 0.6066696643829346, 'learning_rate': 9.8524810192019e-06, 'epoch': 0.51}

 17%|█▋        | 2261/13317 [23:17<1:53:02,  1.63it/s]
 17%|█▋        | 2262/13317 [23:18<1:53:01,  1.63it/s]
                                                      
{'loss': 0.2195, 'grad_norm': 0.4731000065803528, 'learning_rate': 9.852164836848052e-06, 'epoch': 0.51}

 17%|█▋        | 2262/13317 [23:18<1:53:01,  1.63it/s]
 17%|█▋        | 2263/13317 [23:18<1:52:59,  1.63it/s]
                                                      
{'loss': 0.2358, 'grad_norm': 0.5258533954620361, 'learning_rate': 9.851848321099103e-06, 'epoch': 0.51}

 17%|█▋        | 2263/13317 [23:19<1:52:59,  1.63it/s]
 17%|█▋        | 2264/13317 [23:19<1:53:10,  1.63it/s]
                                                      
{'loss': 0.323, 'grad_norm': 0.7608445286750793, 'learning_rate': 9.851531471976803e-06, 'epoch': 0.51}

 17%|█▋        | 2264/13317 [23:19<1:53:10,  1.63it/s]
 17%|█▋        | 2265/13317 [23:20<1:53:05,  1.63it/s]
                                                      
{'loss': 0.2606, 'grad_norm': 0.6363921761512756, 'learning_rate': 9.851214289502922e-06, 'epoch': 0.51}

 17%|█▋        | 2265/13317 [23:20<1:53:05,  1.63it/s]
 17%|█▋        | 2266/13317 [23:20<1:52:59,  1.63it/s]
                                                      
{'loss': 0.2687, 'grad_norm': 0.5686508417129517, 'learning_rate': 9.850896773699252e-06, 'epoch': 0.51}

 17%|█▋        | 2266/13317 [23:20<1:52:59,  1.63it/s]
 17%|█▋        | 2267/13317 [23:21<1:52:56,  1.63it/s]
                                                      
{'loss': 0.3391, 'grad_norm': 0.6715508103370667, 'learning_rate': 9.850578924587614e-06, 'epoch': 0.51}

 17%|█▋        | 2267/13317 [23:21<1:52:56,  1.63it/s]
 17%|█▋        | 2268/13317 [23:22<1:52:58,  1.63it/s]
                                                      
{'loss': 0.3411, 'grad_norm': 0.6343123316764832, 'learning_rate': 9.850260742189843e-06, 'epoch': 0.51}

 17%|█▋        | 2268/13317 [23:22<1:52:58,  1.63it/s]
 17%|█▋        | 2269/13317 [23:22<1:53:07,  1.63it/s]
                                                      
{'loss': 0.2416, 'grad_norm': 0.6808313727378845, 'learning_rate': 9.849942226527805e-06, 'epoch': 0.51}

 17%|█▋        | 2269/13317 [23:22<1:53:07,  1.63it/s]
 17%|█▋        | 2270/13317 [23:23<1:53:03,  1.63it/s]
                                                      
{'loss': 0.2742, 'grad_norm': 0.5917943716049194, 'learning_rate': 9.849623377623384e-06, 'epoch': 0.51}

 17%|█▋        | 2270/13317 [23:23<1:53:03,  1.63it/s]
 17%|█▋        | 2271/13317 [23:23<1:53:00,  1.63it/s]
                                                      
{'loss': 0.449, 'grad_norm': 0.7489339113235474, 'learning_rate': 9.849304195498487e-06, 'epoch': 0.51}

 17%|█▋        | 2271/13317 [23:24<1:53:00,  1.63it/s]
 17%|█▋        | 2272/13317 [23:24<1:52:56,  1.63it/s]
                                                      
{'loss': 0.2671, 'grad_norm': 0.5829532742500305, 'learning_rate': 9.848984680175049e-06, 'epoch': 0.51}

 17%|█▋        | 2272/13317 [23:24<1:52:56,  1.63it/s]
 17%|█▋        | 2273/13317 [23:25<1:52:55,  1.63it/s]
                                                      
{'loss': 0.2805, 'grad_norm': 0.5433275103569031, 'learning_rate': 9.84866483167502e-06, 'epoch': 0.51}

 17%|█▋        | 2273/13317 [23:25<1:52:55,  1.63it/s]
 17%|█▋        | 2274/13317 [23:25<1:52:56,  1.63it/s]
                                                      
{'loss': 0.268, 'grad_norm': 0.5580660700798035, 'learning_rate': 9.848344650020379e-06, 'epoch': 0.51}

 17%|█▋        | 2274/13317 [23:25<1:52:56,  1.63it/s]
 17%|█▋        | 2275/13317 [23:26<1:52:53,  1.63it/s]
                                                      
{'loss': 0.2734, 'grad_norm': 0.5742812156677246, 'learning_rate': 9.848024135233123e-06, 'epoch': 0.51}

 17%|█▋        | 2275/13317 [23:26<1:52:53,  1.63it/s]
 17%|█▋        | 2276/13317 [23:26<1:52:50,  1.63it/s]
                                                      
{'loss': 0.3313, 'grad_norm': 0.6519685983657837, 'learning_rate': 9.847703287335279e-06, 'epoch': 0.51}

 17%|█▋        | 2276/13317 [23:27<1:52:50,  1.63it/s]
 17%|█▋        | 2277/13317 [23:27<1:52:53,  1.63it/s]
                                                      
{'loss': 0.2273, 'grad_norm': 0.6075624823570251, 'learning_rate': 9.847382106348892e-06, 'epoch': 0.51}

 17%|█▋        | 2277/13317 [23:27<1:52:53,  1.63it/s]
 17%|█▋        | 2278/13317 [23:28<1:52:55,  1.63it/s]
                                                      
{'loss': 0.2545, 'grad_norm': 0.5523224472999573, 'learning_rate': 9.84706059229603e-06, 'epoch': 0.51}

 17%|█▋        | 2278/13317 [23:28<1:52:55,  1.63it/s]
 17%|█▋        | 2279/13317 [23:28<1:52:49,  1.63it/s]
                                                      
{'loss': 0.3138, 'grad_norm': 0.6368107795715332, 'learning_rate': 9.846738745198781e-06, 'epoch': 0.51}

 17%|█▋        | 2279/13317 [23:28<1:52:49,  1.63it/s]
 17%|█▋        | 2280/13317 [23:29<1:52:53,  1.63it/s]
                                                      
{'loss': 0.2242, 'grad_norm': 0.5108243227005005, 'learning_rate': 9.846416565079264e-06, 'epoch': 0.51}

 17%|█▋        | 2280/13317 [23:29<1:52:53,  1.63it/s]
 17%|█▋        | 2281/13317 [23:30<1:52:52,  1.63it/s]
                                                      
{'loss': 0.3105, 'grad_norm': 0.48200884461402893, 'learning_rate': 9.846094051959614e-06, 'epoch': 0.51}

 17%|█▋        | 2281/13317 [23:30<1:52:52,  1.63it/s]
 17%|█▋        | 2282/13317 [23:30<1:52:48,  1.63it/s]
                                                      
{'loss': 0.2663, 'grad_norm': 0.539749026298523, 'learning_rate': 9.84577120586199e-06, 'epoch': 0.51}

 17%|█▋        | 2282/13317 [23:30<1:52:48,  1.63it/s]
 17%|█▋        | 2283/13317 [23:31<1:52:48,  1.63it/s]
                                                      
{'loss': 0.3046, 'grad_norm': 0.593388020992279, 'learning_rate': 9.845448026808578e-06, 'epoch': 0.51}

 17%|█▋        | 2283/13317 [23:31<1:52:48,  1.63it/s]
 17%|█▋        | 2284/13317 [23:31<1:52:47,  1.63it/s]
                                                      
{'loss': 0.2152, 'grad_norm': 0.5103970170021057, 'learning_rate': 9.845124514821581e-06, 'epoch': 0.51}

 17%|█▋        | 2284/13317 [23:31<1:52:47,  1.63it/s]
 17%|█▋        | 2285/13317 [23:32<1:52:45,  1.63it/s]
                                                      
{'loss': 0.2095, 'grad_norm': 0.505423367023468, 'learning_rate': 9.84480066992323e-06, 'epoch': 0.51}

 17%|█▋        | 2285/13317 [23:32<1:52:45,  1.63it/s]
 17%|█▋        | 2286/13317 [23:33<1:52:47,  1.63it/s]
                                                      
{'loss': 0.3249, 'grad_norm': 0.5934234261512756, 'learning_rate': 9.844476492135775e-06, 'epoch': 0.51}

 17%|█▋        | 2286/13317 [23:33<1:52:47,  1.63it/s]
 17%|█▋        | 2287/13317 [23:33<1:52:46,  1.63it/s]
                                                      
{'loss': 0.2664, 'grad_norm': 0.5861340761184692, 'learning_rate': 9.84415198148149e-06, 'epoch': 0.52}

 17%|█▋        | 2287/13317 [23:33<1:52:46,  1.63it/s]
 17%|█▋        | 2288/13317 [23:34<1:52:41,  1.63it/s]
                                                      
{'loss': 0.3724, 'grad_norm': 0.6166456937789917, 'learning_rate': 9.843827137982673e-06, 'epoch': 0.52}

 17%|█▋        | 2288/13317 [23:34<1:52:41,  1.63it/s]
 17%|█▋        | 2289/13317 [23:34<1:52:39,  1.63it/s]
                                                      
{'loss': 0.225, 'grad_norm': 0.5371770262718201, 'learning_rate': 9.843501961661646e-06, 'epoch': 0.52}

 17%|█▋        | 2289/13317 [23:35<1:52:39,  1.63it/s]
 17%|█▋        | 2290/13317 [23:35<1:52:40,  1.63it/s]
                                                      
{'loss': 0.2715, 'grad_norm': 0.6601698994636536, 'learning_rate': 9.84317645254075e-06, 'epoch': 0.52}

 17%|█▋        | 2290/13317 [23:35<1:52:40,  1.63it/s]
 17%|█▋        | 2291/13317 [23:36<1:52:37,  1.63it/s]
                                                      
{'loss': 0.3077, 'grad_norm': 0.6676005721092224, 'learning_rate': 9.84285061064235e-06, 'epoch': 0.52}

 17%|█▋        | 2291/13317 [23:36<1:52:37,  1.63it/s]
 17%|█▋        | 2292/13317 [23:36<1:52:37,  1.63it/s]
                                                      
{'loss': 0.3198, 'grad_norm': 0.5791602730751038, 'learning_rate': 9.842524435988839e-06, 'epoch': 0.52}

 17%|█▋        | 2292/13317 [23:36<1:52:37,  1.63it/s]
 17%|█▋        | 2293/13317 [23:37<1:52:38,  1.63it/s]
                                                      
{'loss': 0.2319, 'grad_norm': 0.5838137269020081, 'learning_rate': 9.842197928602622e-06, 'epoch': 0.52}

 17%|█▋        | 2293/13317 [23:37<1:52:38,  1.63it/s]
 17%|█▋        | 2294/13317 [23:37<1:52:35,  1.63it/s]
                                                      
{'loss': 0.2558, 'grad_norm': 0.5407522320747375, 'learning_rate': 9.84187108850614e-06, 'epoch': 0.52}

 17%|█▋        | 2294/13317 [23:38<1:52:35,  1.63it/s]
 17%|█▋        | 2295/13317 [23:38<1:52:38,  1.63it/s]
                                                      
{'loss': 0.3557, 'grad_norm': 0.7187723517417908, 'learning_rate': 9.841543915721846e-06, 'epoch': 0.52}

 17%|█▋        | 2295/13317 [23:38<1:52:38,  1.63it/s]
 17%|█▋        | 2296/13317 [23:39<1:52:38,  1.63it/s]
                                                      
{'loss': 0.2463, 'grad_norm': 0.5499375462532043, 'learning_rate': 9.841216410272223e-06, 'epoch': 0.52}

 17%|█▋        | 2296/13317 [23:39<1:52:38,  1.63it/s]
 17%|█▋        | 2297/13317 [23:39<1:52:37,  1.63it/s]
                                                      
{'loss': 0.2711, 'grad_norm': 0.5525514483451843, 'learning_rate': 9.840888572179771e-06, 'epoch': 0.52}

 17%|█▋        | 2297/13317 [23:39<1:52:37,  1.63it/s]
 17%|█▋        | 2298/13317 [23:40<1:52:37,  1.63it/s]
                                                      
{'loss': 0.2342, 'grad_norm': 0.5220012664794922, 'learning_rate': 9.840560401467021e-06, 'epoch': 0.52}

 17%|█▋        | 2298/13317 [23:40<1:52:37,  1.63it/s]
 17%|█▋        | 2299/13317 [23:41<1:52:37,  1.63it/s]
                                                      
{'loss': 0.284, 'grad_norm': 0.586678683757782, 'learning_rate': 9.840231898156518e-06, 'epoch': 0.52}

 17%|█▋        | 2299/13317 [23:41<1:52:37,  1.63it/s]
 17%|█▋        | 2300/13317 [23:41<1:52:37,  1.63it/s]
                                                      
{'loss': 0.204, 'grad_norm': 0.5181857943534851, 'learning_rate': 9.839903062270834e-06, 'epoch': 0.52}

 17%|█▋        | 2300/13317 [23:41<1:52:37,  1.63it/s]
 17%|█▋        | 2301/13317 [23:42<1:58:53,  1.54it/s]
                                                      
{'loss': 0.2318, 'grad_norm': 0.5117096304893494, 'learning_rate': 9.839573893832564e-06, 'epoch': 0.52}

 17%|█▋        | 2301/13317 [23:42<1:58:53,  1.54it/s]
 17%|█▋        | 2302/13317 [23:43<1:57:04,  1.57it/s]
                                                      
{'loss': 0.2274, 'grad_norm': 0.5183722972869873, 'learning_rate': 9.839244392864324e-06, 'epoch': 0.52}

 17%|█▋        | 2302/13317 [23:43<1:57:04,  1.57it/s]
 17%|█▋        | 2303/13317 [23:43<1:55:41,  1.59it/s]
                                                      
{'loss': 0.2884, 'grad_norm': 0.5655578970909119, 'learning_rate': 9.838914559388758e-06, 'epoch': 0.52}

 17%|█▋        | 2303/13317 [23:43<1:55:41,  1.59it/s]
 17%|█▋        | 2304/13317 [23:44<1:54:55,  1.60it/s]
                                                      
{'loss': 0.3525, 'grad_norm': 0.6479787230491638, 'learning_rate': 9.838584393428522e-06, 'epoch': 0.52}

 17%|█▋        | 2304/13317 [23:44<1:54:55,  1.60it/s]
 17%|█▋        | 2305/13317 [23:44<1:54:10,  1.61it/s]
                                                      
{'loss': 0.2699, 'grad_norm': 0.570887565612793, 'learning_rate': 9.838253895006312e-06, 'epoch': 0.52}

 17%|█▋        | 2305/13317 [23:44<1:54:10,  1.61it/s]
 17%|█▋        | 2306/13317 [23:45<1:53:40,  1.61it/s]
                                                      
{'loss': 0.2906, 'grad_norm': 0.6225704550743103, 'learning_rate': 9.837923064144828e-06, 'epoch': 0.52}

 17%|█▋        | 2306/13317 [23:45<1:53:40,  1.61it/s]
 17%|█▋        | 2307/13317 [23:46<1:53:20,  1.62it/s]
                                                      
{'loss': 0.3358, 'grad_norm': 0.6314661502838135, 'learning_rate': 9.837591900866805e-06, 'epoch': 0.52}

 17%|█▋        | 2307/13317 [23:46<1:53:20,  1.62it/s]
 17%|█▋        | 2308/13317 [23:46<1:53:02,  1.62it/s]
                                                      
{'loss': 0.2714, 'grad_norm': 0.5910045504570007, 'learning_rate': 9.837260405194997e-06, 'epoch': 0.52}

 17%|█▋        | 2308/13317 [23:46<1:53:02,  1.62it/s]
 17%|█▋        | 2309/13317 [23:47<1:52:54,  1.62it/s]
                                                      
{'loss': 0.2651, 'grad_norm': 0.5814926028251648, 'learning_rate': 9.836928577152181e-06, 'epoch': 0.52}

 17%|█▋        | 2309/13317 [23:47<1:52:54,  1.62it/s]
 17%|█▋        | 2310/13317 [23:47<1:52:52,  1.63it/s]
                                                      
{'loss': 0.2871, 'grad_norm': 0.6421974897384644, 'learning_rate': 9.83659641676116e-06, 'epoch': 0.52}

 17%|█▋        | 2310/13317 [23:48<1:52:52,  1.63it/s]
 17%|█▋        | 2311/13317 [23:48<1:52:46,  1.63it/s]
                                                      
{'loss': 0.2734, 'grad_norm': 0.5437297821044922, 'learning_rate': 9.83626392404475e-06, 'epoch': 0.52}

 17%|█▋        | 2311/13317 [23:48<1:52:46,  1.63it/s]
 17%|█▋        | 2312/13317 [23:49<1:52:40,  1.63it/s]
                                                      
{'loss': 0.2129, 'grad_norm': 0.529634416103363, 'learning_rate': 9.835931099025807e-06, 'epoch': 0.52}

 17%|█▋        | 2312/13317 [23:49<1:52:40,  1.63it/s]
 17%|█▋        | 2313/13317 [23:49<1:52:38,  1.63it/s]
                                                      
{'loss': 0.3453, 'grad_norm': 0.5899722576141357, 'learning_rate': 9.83559794172719e-06, 'epoch': 0.52}

 17%|█▋        | 2313/13317 [23:49<1:52:38,  1.63it/s]
 17%|█▋        | 2314/13317 [23:50<1:52:36,  1.63it/s]
                                                      
{'loss': 0.2573, 'grad_norm': 0.6592074036598206, 'learning_rate': 9.835264452171795e-06, 'epoch': 0.52}

 17%|█▋        | 2314/13317 [23:50<1:52:36,  1.63it/s]
 17%|█▋        | 2315/13317 [23:50<1:52:34,  1.63it/s]
                                                      
{'loss': 0.3086, 'grad_norm': 0.580508291721344, 'learning_rate': 9.834930630382536e-06, 'epoch': 0.52}

 17%|█▋        | 2315/13317 [23:51<1:52:34,  1.63it/s]
 17%|█▋        | 2316/13317 [23:51<1:52:30,  1.63it/s]
                                                      
{'loss': 0.2771, 'grad_norm': 0.6241456866264343, 'learning_rate': 9.834596476382349e-06, 'epoch': 0.52}

 17%|█▋        | 2316/13317 [23:51<1:52:30,  1.63it/s]
 17%|█▋        | 2317/13317 [23:52<1:52:30,  1.63it/s]
                                                      
{'loss': 0.2833, 'grad_norm': 0.6292261481285095, 'learning_rate': 9.834261990194195e-06, 'epoch': 0.52}

 17%|█▋        | 2317/13317 [23:52<1:52:30,  1.63it/s]
 17%|█▋        | 2318/13317 [23:52<1:52:26,  1.63it/s]
                                                      
{'loss': 0.3306, 'grad_norm': 0.6396340727806091, 'learning_rate': 9.833927171841055e-06, 'epoch': 0.52}

 17%|█▋        | 2318/13317 [23:52<1:52:26,  1.63it/s]
 17%|█▋        | 2319/13317 [23:53<1:52:26,  1.63it/s]
                                                      
{'loss': 0.2556, 'grad_norm': 0.563748300075531, 'learning_rate': 9.833592021345938e-06, 'epoch': 0.52}

 17%|█▋        | 2319/13317 [23:53<1:52:26,  1.63it/s]
 17%|█▋        | 2320/13317 [23:54<1:52:30,  1.63it/s]
                                                      
{'loss': 0.4011, 'grad_norm': 0.698057234287262, 'learning_rate': 9.83325653873187e-06, 'epoch': 0.52}

 17%|█▋        | 2320/13317 [23:54<1:52:30,  1.63it/s]
 17%|█▋        | 2321/13317 [23:54<1:52:30,  1.63it/s]
                                                      
{'loss': 0.3095, 'grad_norm': 0.5392757654190063, 'learning_rate': 9.832920724021901e-06, 'epoch': 0.52}

 17%|█▋        | 2321/13317 [23:54<1:52:30,  1.63it/s]
 17%|█▋        | 2322/13317 [23:55<1:52:28,  1.63it/s]
                                                      
{'loss': 0.2453, 'grad_norm': 0.5690225958824158, 'learning_rate': 9.832584577239107e-06, 'epoch': 0.52}

 17%|█▋        | 2322/13317 [23:55<1:52:28,  1.63it/s]
 17%|█▋        | 2323/13317 [23:55<1:52:31,  1.63it/s]
                                                      
{'loss': 0.3175, 'grad_norm': 0.6050390005111694, 'learning_rate': 9.832248098406585e-06, 'epoch': 0.52}

 17%|█▋        | 2323/13317 [23:56<1:52:31,  1.63it/s]
 17%|█▋        | 2324/13317 [23:56<1:52:28,  1.63it/s]
                                                      
{'loss': 0.3705, 'grad_norm': 0.6470702290534973, 'learning_rate': 9.831911287547453e-06, 'epoch': 0.52}

 17%|█▋        | 2324/13317 [23:56<1:52:28,  1.63it/s]
 17%|█▋        | 2325/13317 [23:57<1:52:24,  1.63it/s]
                                                      
{'loss': 0.2416, 'grad_norm': 0.5133500099182129, 'learning_rate': 9.831574144684855e-06, 'epoch': 0.52}

 17%|█▋        | 2325/13317 [23:57<1:52:24,  1.63it/s]
 17%|█▋        | 2326/13317 [23:57<1:52:19,  1.63it/s]
                                                      
{'loss': 0.261, 'grad_norm': 0.5060635805130005, 'learning_rate': 9.831236669841956e-06, 'epoch': 0.52}

 17%|█▋        | 2326/13317 [23:57<1:52:19,  1.63it/s]
 17%|█▋        | 2327/13317 [23:58<1:52:18,  1.63it/s]
                                                      
{'loss': 0.2287, 'grad_norm': 0.46299445629119873, 'learning_rate': 9.830898863041943e-06, 'epoch': 0.52}

 17%|█▋        | 2327/13317 [23:58<1:52:18,  1.63it/s]
 17%|█▋        | 2328/13317 [23:58<1:52:17,  1.63it/s]
                                                      
{'loss': 0.2697, 'grad_norm': 0.5236720442771912, 'learning_rate': 9.830560724308028e-06, 'epoch': 0.52}

 17%|█▋        | 2328/13317 [23:59<1:52:17,  1.63it/s]
 17%|█▋        | 2329/13317 [23:59<1:52:17,  1.63it/s]
                                                      
{'loss': 0.2945, 'grad_norm': 0.5198934674263, 'learning_rate': 9.830222253663445e-06, 'epoch': 0.52}

 17%|█▋        | 2329/13317 [23:59<1:52:17,  1.63it/s]
 17%|█▋        | 2330/13317 [24:00<1:52:19,  1.63it/s]
                                                      
{'loss': 0.3053, 'grad_norm': 0.6498871445655823, 'learning_rate': 9.82988345113145e-06, 'epoch': 0.52}

 17%|█▋        | 2330/13317 [24:00<1:52:19,  1.63it/s]
 18%|█▊        | 2331/13317 [24:00<1:52:16,  1.63it/s]
                                                      
{'loss': 0.2467, 'grad_norm': 0.5567374229431152, 'learning_rate': 9.829544316735323e-06, 'epoch': 0.53}

 18%|█▊        | 2331/13317 [24:00<1:52:16,  1.63it/s]
 18%|█▊        | 2332/13317 [24:01<1:52:16,  1.63it/s]
                                                      
{'loss': 0.2536, 'grad_norm': 0.5468451976776123, 'learning_rate': 9.829204850498362e-06, 'epoch': 0.53}

 18%|█▊        | 2332/13317 [24:01<1:52:16,  1.63it/s]
 18%|█▊        | 2333/13317 [24:02<1:52:16,  1.63it/s]
                                                      
{'loss': 0.3171, 'grad_norm': 0.5865020155906677, 'learning_rate': 9.828865052443899e-06, 'epoch': 0.53}

 18%|█▊        | 2333/13317 [24:02<1:52:16,  1.63it/s]
 18%|█▊        | 2334/13317 [24:02<1:52:13,  1.63it/s]
                                                      
{'loss': 0.2978, 'grad_norm': 0.5952492356300354, 'learning_rate': 9.828524922595276e-06, 'epoch': 0.53}

 18%|█▊        | 2334/13317 [24:02<1:52:13,  1.63it/s]
 18%|█▊        | 2335/13317 [24:03<1:52:18,  1.63it/s]
                                                      
{'loss': 0.3283, 'grad_norm': 0.6185967326164246, 'learning_rate': 9.828184460975867e-06, 'epoch': 0.53}

 18%|█▊        | 2335/13317 [24:03<1:52:18,  1.63it/s]
 18%|█▊        | 2336/13317 [24:03<1:52:13,  1.63it/s]
                                                      
{'loss': 0.294, 'grad_norm': 0.6721116900444031, 'learning_rate': 9.827843667609064e-06, 'epoch': 0.53}

 18%|█▊        | 2336/13317 [24:04<1:52:13,  1.63it/s]
 18%|█▊        | 2337/13317 [24:04<1:52:13,  1.63it/s]
                                                      
{'loss': 0.2666, 'grad_norm': 0.5291370749473572, 'learning_rate': 9.827502542518281e-06, 'epoch': 0.53}

 18%|█▊        | 2337/13317 [24:04<1:52:13,  1.63it/s]
 18%|█▊        | 2338/13317 [24:05<1:52:11,  1.63it/s]
                                                      
{'loss': 0.3204, 'grad_norm': 0.5960599184036255, 'learning_rate': 9.827161085726959e-06, 'epoch': 0.53}

 18%|█▊        | 2338/13317 [24:05<1:52:11,  1.63it/s]
 18%|█▊        | 2339/13317 [24:05<1:52:09,  1.63it/s]
                                                      
{'loss': 0.3082, 'grad_norm': 0.54073566198349, 'learning_rate': 9.826819297258562e-06, 'epoch': 0.53}

 18%|█▊        | 2339/13317 [24:05<1:52:09,  1.63it/s]
 18%|█▊        | 2340/13317 [24:06<1:52:11,  1.63it/s]
                                                      
{'loss': 0.2188, 'grad_norm': 0.4992220103740692, 'learning_rate': 9.82647717713657e-06, 'epoch': 0.53}

 18%|█▊        | 2340/13317 [24:06<1:52:11,  1.63it/s]
 18%|█▊        | 2341/13317 [24:06<1:52:09,  1.63it/s]
                                                      
{'loss': 0.2843, 'grad_norm': 0.6220024228096008, 'learning_rate': 9.826134725384493e-06, 'epoch': 0.53}

 18%|█▊        | 2341/13317 [24:07<1:52:09,  1.63it/s]
 18%|█▊        | 2342/13317 [24:07<1:52:09,  1.63it/s]
                                                      
{'loss': 0.2591, 'grad_norm': 0.5951791405677795, 'learning_rate': 9.825791942025861e-06, 'epoch': 0.53}

 18%|█▊        | 2342/13317 [24:07<1:52:09,  1.63it/s]
 18%|█▊        | 2343/13317 [24:08<1:52:06,  1.63it/s]
                                                      
{'loss': 0.2491, 'grad_norm': 0.5686314702033997, 'learning_rate': 9.825448827084226e-06, 'epoch': 0.53}

 18%|█▊        | 2343/13317 [24:08<1:52:06,  1.63it/s]
 18%|█▊        | 2344/13317 [24:08<1:52:09,  1.63it/s]
                                                      
{'loss': 0.3075, 'grad_norm': 0.6417514681816101, 'learning_rate': 9.825105380583164e-06, 'epoch': 0.53}

 18%|█▊        | 2344/13317 [24:08<1:52:09,  1.63it/s]
 18%|█▊        | 2345/13317 [24:09<1:52:05,  1.63it/s]
                                                      
{'loss': 0.3211, 'grad_norm': 0.7492538690567017, 'learning_rate': 9.824761602546273e-06, 'epoch': 0.53}

 18%|█▊        | 2345/13317 [24:09<1:52:05,  1.63it/s]
 18%|█▊        | 2346/13317 [24:09<1:52:02,  1.63it/s]
                                                      
{'loss': 0.2265, 'grad_norm': 0.49722397327423096, 'learning_rate': 9.824417492997175e-06, 'epoch': 0.53}

 18%|█▊        | 2346/13317 [24:10<1:52:02,  1.63it/s]
 18%|█▊        | 2347/13317 [24:10<1:52:04,  1.63it/s]
                                                      
{'loss': 0.2855, 'grad_norm': 0.5939038991928101, 'learning_rate': 9.824073051959512e-06, 'epoch': 0.53}

 18%|█▊        | 2347/13317 [24:10<1:52:04,  1.63it/s]
 18%|█▊        | 2348/13317 [24:11<1:52:04,  1.63it/s]
                                                      
{'loss': 0.3191, 'grad_norm': 0.5735130906105042, 'learning_rate': 9.823728279456954e-06, 'epoch': 0.53}

 18%|█▊        | 2348/13317 [24:11<1:52:04,  1.63it/s]
 18%|█▊        | 2349/13317 [24:11<1:52:01,  1.63it/s]
                                                      
{'loss': 0.2955, 'grad_norm': 0.6225186586380005, 'learning_rate': 9.823383175513189e-06, 'epoch': 0.53}

 18%|█▊        | 2349/13317 [24:11<1:52:01,  1.63it/s]
 18%|█▊        | 2350/13317 [24:12<1:52:05,  1.63it/s]
                                                      
{'loss': 0.2426, 'grad_norm': 0.5337162017822266, 'learning_rate': 9.823037740151928e-06, 'epoch': 0.53}

 18%|█▊        | 2350/13317 [24:12<1:52:05,  1.63it/s]
 18%|█▊        | 2351/13317 [24:13<1:52:07,  1.63it/s]
                                                      
{'loss': 0.2649, 'grad_norm': 0.6021965742111206, 'learning_rate': 9.822691973396907e-06, 'epoch': 0.53}

 18%|█▊        | 2351/13317 [24:13<1:52:07,  1.63it/s]
 18%|█▊        | 2352/13317 [24:13<1:52:03,  1.63it/s]
                                                      
{'loss': 0.3752, 'grad_norm': 0.6077548861503601, 'learning_rate': 9.822345875271884e-06, 'epoch': 0.53}

 18%|█▊        | 2352/13317 [24:13<1:52:03,  1.63it/s]
 18%|█▊        | 2353/13317 [24:14<1:52:02,  1.63it/s]
                                                      
{'loss': 0.18, 'grad_norm': 0.5630260109901428, 'learning_rate': 9.821999445800639e-06, 'epoch': 0.53}

 18%|█▊        | 2353/13317 [24:14<1:52:02,  1.63it/s]
 18%|█▊        | 2354/13317 [24:14<1:52:02,  1.63it/s]
                                                      
{'loss': 0.309, 'grad_norm': 0.590865433216095, 'learning_rate': 9.821652685006978e-06, 'epoch': 0.53}

 18%|█▊        | 2354/13317 [24:15<1:52:02,  1.63it/s]
 18%|█▊        | 2355/13317 [24:15<1:52:03,  1.63it/s]
                                                      
{'loss': 0.2181, 'grad_norm': 0.4693679213523865, 'learning_rate': 9.821305592914722e-06, 'epoch': 0.53}

 18%|█▊        | 2355/13317 [24:15<1:52:03,  1.63it/s]
 18%|█▊        | 2356/13317 [24:16<1:52:03,  1.63it/s]
                                                      
{'loss': 0.2695, 'grad_norm': 0.5847597122192383, 'learning_rate': 9.820958169547725e-06, 'epoch': 0.53}

 18%|█▊        | 2356/13317 [24:16<1:52:03,  1.63it/s]
 18%|█▊        | 2357/13317 [24:16<1:52:02,  1.63it/s]
                                                      
{'loss': 0.2479, 'grad_norm': 0.6225796341896057, 'learning_rate': 9.820610414929856e-06, 'epoch': 0.53}

 18%|█▊        | 2357/13317 [24:16<1:52:02,  1.63it/s]
 18%|█▊        | 2358/13317 [24:17<1:52:00,  1.63it/s]
                                                      
{'loss': 0.4215, 'grad_norm': 0.7728519439697266, 'learning_rate': 9.820262329085011e-06, 'epoch': 0.53}

 18%|█▊        | 2358/13317 [24:17<1:52:00,  1.63it/s]
 18%|█▊        | 2359/13317 [24:17<1:51:59,  1.63it/s]
                                                      
{'loss': 0.1738, 'grad_norm': 0.4338122010231018, 'learning_rate': 9.819913912037106e-06, 'epoch': 0.53}

 18%|█▊        | 2359/13317 [24:18<1:51:59,  1.63it/s]
 18%|█▊        | 2360/13317 [24:18<1:52:02,  1.63it/s]
                                                      
{'loss': 0.3337, 'grad_norm': 0.6449947953224182, 'learning_rate': 9.819565163810081e-06, 'epoch': 0.53}

 18%|█▊        | 2360/13317 [24:18<1:52:02,  1.63it/s]
 18%|█▊        | 2361/13317 [24:19<1:52:02,  1.63it/s]
                                                      
{'loss': 0.2678, 'grad_norm': 0.5572062134742737, 'learning_rate': 9.819216084427898e-06, 'epoch': 0.53}

 18%|█▊        | 2361/13317 [24:19<1:52:02,  1.63it/s]
 18%|█▊        | 2362/13317 [24:19<1:55:00,  1.59it/s]
                                                      
{'loss': 0.3763, 'grad_norm': 0.6217794418334961, 'learning_rate': 9.818866673914545e-06, 'epoch': 0.53}

 18%|█▊        | 2362/13317 [24:19<1:55:00,  1.59it/s]
 18%|█▊        | 2363/13317 [24:20<1:54:11,  1.60it/s]
                                                      
{'loss': 0.2419, 'grad_norm': 0.5042833685874939, 'learning_rate': 9.818516932294026e-06, 'epoch': 0.53}

 18%|█▊        | 2363/13317 [24:20<1:54:11,  1.60it/s]
 18%|█▊        | 2364/13317 [24:21<1:53:44,  1.60it/s]
                                                      
{'loss': 0.2351, 'grad_norm': 0.5353940725326538, 'learning_rate': 9.818166859590377e-06, 'epoch': 0.53}

 18%|█▊        | 2364/13317 [24:21<1:53:44,  1.60it/s]
 18%|█▊        | 2365/13317 [24:21<1:53:13,  1.61it/s]
                                                      
{'loss': 0.332, 'grad_norm': 0.6748570203781128, 'learning_rate': 9.817816455827648e-06, 'epoch': 0.53}

 18%|█▊        | 2365/13317 [24:21<1:53:13,  1.61it/s]
 18%|█▊        | 2366/13317 [24:22<1:52:50,  1.62it/s]
                                                      
{'loss': 0.3075, 'grad_norm': 0.6581902503967285, 'learning_rate': 9.817465721029916e-06, 'epoch': 0.53}

 18%|█▊        | 2366/13317 [24:22<1:52:50,  1.62it/s]
 18%|█▊        | 2367/13317 [24:22<1:52:42,  1.62it/s]
                                                      
{'loss': 0.2484, 'grad_norm': 0.5894600749015808, 'learning_rate': 9.817114655221282e-06, 'epoch': 0.53}

 18%|█▊        | 2367/13317 [24:23<1:52:42,  1.62it/s]
 18%|█▊        | 2368/13317 [24:23<1:52:25,  1.62it/s]
                                                      
{'loss': 0.2364, 'grad_norm': 0.5006355047225952, 'learning_rate': 9.816763258425867e-06, 'epoch': 0.53}

 18%|█▊        | 2368/13317 [24:23<1:52:25,  1.62it/s]
 18%|█▊        | 2369/13317 [24:24<1:52:21,  1.62it/s]
                                                      
{'loss': 0.2351, 'grad_norm': 0.5470225811004639, 'learning_rate': 9.816411530667815e-06, 'epoch': 0.53}

 18%|█▊        | 2369/13317 [24:24<1:52:21,  1.62it/s]
 18%|█▊        | 2370/13317 [24:24<1:52:09,  1.63it/s]
                                                      
{'loss': 0.2687, 'grad_norm': 0.5852634906768799, 'learning_rate': 9.816059471971293e-06, 'epoch': 0.53}

 18%|█▊        | 2370/13317 [24:24<1:52:09,  1.63it/s]
 18%|█▊        | 2371/13317 [24:25<1:52:05,  1.63it/s]
                                                      
{'loss': 0.2483, 'grad_norm': 0.5335555076599121, 'learning_rate': 9.815707082360492e-06, 'epoch': 0.53}

 18%|█▊        | 2371/13317 [24:25<1:52:05,  1.63it/s]
 18%|█▊        | 2372/13317 [24:26<1:52:08,  1.63it/s]
                                                      
{'loss': 0.2993, 'grad_norm': 0.6153882145881653, 'learning_rate': 9.815354361859626e-06, 'epoch': 0.53}

 18%|█▊        | 2372/13317 [24:26<1:52:08,  1.63it/s]
 18%|█▊        | 2373/13317 [24:26<1:52:01,  1.63it/s]
                                                      
{'loss': 0.2832, 'grad_norm': 0.5509562492370605, 'learning_rate': 9.815001310492928e-06, 'epoch': 0.53}

 18%|█▊        | 2373/13317 [24:26<1:52:01,  1.63it/s]
 18%|█▊        | 2374/13317 [24:27<1:51:58,  1.63it/s]
                                                      
{'loss': 0.2102, 'grad_norm': 0.5336902737617493, 'learning_rate': 9.81464792828466e-06, 'epoch': 0.53}

 18%|█▊        | 2374/13317 [24:27<1:51:58,  1.63it/s]
 18%|█▊        | 2375/13317 [24:27<1:51:56,  1.63it/s]
                                                      
{'loss': 0.3167, 'grad_norm': 0.6100742816925049, 'learning_rate': 9.8142942152591e-06, 'epoch': 0.54}

 18%|█▊        | 2375/13317 [24:27<1:51:56,  1.63it/s]
 18%|█▊        | 2376/13317 [24:28<1:51:53,  1.63it/s]
                                                      
{'loss': 0.2463, 'grad_norm': 0.5255952477455139, 'learning_rate': 9.813940171440553e-06, 'epoch': 0.54}

 18%|█▊        | 2376/13317 [24:28<1:51:53,  1.63it/s]
 18%|█▊        | 2377/13317 [24:29<1:51:51,  1.63it/s]
                                                      
{'loss': 0.2915, 'grad_norm': 0.6314389705657959, 'learning_rate': 9.813585796853346e-06, 'epoch': 0.54}

 18%|█▊        | 2377/13317 [24:29<1:51:51,  1.63it/s]
 18%|█▊        | 2378/13317 [24:29<1:51:50,  1.63it/s]
                                                      
{'loss': 0.3054, 'grad_norm': 0.6610992550849915, 'learning_rate': 9.813231091521826e-06, 'epoch': 0.54}

 18%|█▊        | 2378/13317 [24:29<1:51:50,  1.63it/s]
 18%|█▊        | 2379/13317 [24:30<1:51:48,  1.63it/s]
                                                      
{'loss': 0.2175, 'grad_norm': 0.5478627681732178, 'learning_rate': 9.812876055470368e-06, 'epoch': 0.54}

 18%|█▊        | 2379/13317 [24:30<1:51:48,  1.63it/s]
 18%|█▊        | 2380/13317 [24:30<1:51:51,  1.63it/s]
                                                      
{'loss': 0.276, 'grad_norm': 0.613308846950531, 'learning_rate': 9.812520688723366e-06, 'epoch': 0.54}

 18%|█▊        | 2380/13317 [24:31<1:51:51,  1.63it/s]
 18%|█▊        | 2381/13317 [24:31<1:51:49,  1.63it/s]
                                                      
{'loss': 0.3203, 'grad_norm': 0.5771105885505676, 'learning_rate': 9.812164991305236e-06, 'epoch': 0.54}

 18%|█▊        | 2381/13317 [24:31<1:51:49,  1.63it/s]
 18%|█▊        | 2382/13317 [24:32<1:51:47,  1.63it/s]
                                                      
{'loss': 0.2265, 'grad_norm': 0.49581676721572876, 'learning_rate': 9.81180896324042e-06, 'epoch': 0.54}

 18%|█▊        | 2382/13317 [24:32<1:51:47,  1.63it/s]
 18%|█▊        | 2383/13317 [24:32<1:51:45,  1.63it/s]
                                                      
{'loss': 0.3136, 'grad_norm': 0.5877315402030945, 'learning_rate': 9.811452604553377e-06, 'epoch': 0.54}

 18%|█▊        | 2383/13317 [24:32<1:51:45,  1.63it/s]
 18%|█▊        | 2384/13317 [24:33<1:51:45,  1.63it/s]
                                                      
{'loss': 0.1949, 'grad_norm': 0.5363995432853699, 'learning_rate': 9.811095915268598e-06, 'epoch': 0.54}

 18%|█▊        | 2384/13317 [24:33<1:51:45,  1.63it/s]
 18%|█▊        | 2385/13317 [24:33<1:51:47,  1.63it/s]
                                                      
{'loss': 0.2267, 'grad_norm': 0.5825826525688171, 'learning_rate': 9.81073889541059e-06, 'epoch': 0.54}

 18%|█▊        | 2385/13317 [24:34<1:51:47,  1.63it/s]
 18%|█▊        | 2386/13317 [24:34<1:51:46,  1.63it/s]
                                                      
{'loss': 0.2913, 'grad_norm': 0.5971139669418335, 'learning_rate': 9.810381545003882e-06, 'epoch': 0.54}

 18%|█▊        | 2386/13317 [24:34<1:51:46,  1.63it/s]
 18%|█▊        | 2387/13317 [24:35<1:51:46,  1.63it/s]
                                                      
{'loss': 0.2201, 'grad_norm': 0.4928883910179138, 'learning_rate': 9.810023864073026e-06, 'epoch': 0.54}

 18%|█▊        | 2387/13317 [24:35<1:51:46,  1.63it/s]
 18%|█▊        | 2388/13317 [24:35<1:51:44,  1.63it/s]
                                                      
{'loss': 0.281, 'grad_norm': 0.5808761715888977, 'learning_rate': 9.809665852642605e-06, 'epoch': 0.54}

 18%|█▊        | 2388/13317 [24:35<1:51:44,  1.63it/s]
 18%|█▊        | 2389/13317 [24:36<1:51:44,  1.63it/s]
                                                      
{'loss': 0.3376, 'grad_norm': 0.6386665105819702, 'learning_rate': 9.809307510737211e-06, 'epoch': 0.54}

 18%|█▊        | 2389/13317 [24:36<1:51:44,  1.63it/s]
 18%|█▊        | 2390/13317 [24:37<1:51:45,  1.63it/s]
                                                      
{'loss': 0.3445, 'grad_norm': 0.5212363004684448, 'learning_rate': 9.808948838381473e-06, 'epoch': 0.54}

 18%|█▊        | 2390/13317 [24:37<1:51:45,  1.63it/s]
 18%|█▊        | 2391/13317 [24:37<1:51:51,  1.63it/s]
                                                      
{'loss': 0.4138, 'grad_norm': 0.6337407827377319, 'learning_rate': 9.80858983560003e-06, 'epoch': 0.54}

 18%|█▊        | 2391/13317 [24:37<1:51:51,  1.63it/s]
 18%|█▊        | 2392/13317 [24:38<1:51:48,  1.63it/s]
                                                      
{'loss': 0.2453, 'grad_norm': 0.5074365735054016, 'learning_rate': 9.808230502417551e-06, 'epoch': 0.54}

 18%|█▊        | 2392/13317 [24:38<1:51:48,  1.63it/s]
 18%|█▊        | 2393/13317 [24:38<1:51:44,  1.63it/s]
                                                      
{'loss': 0.2937, 'grad_norm': 0.6207624673843384, 'learning_rate': 9.807870838858726e-06, 'epoch': 0.54}

 18%|█▊        | 2393/13317 [24:39<1:51:44,  1.63it/s]
 18%|█▊        | 2394/13317 [24:39<1:51:42,  1.63it/s]
                                                      
{'loss': 0.2608, 'grad_norm': 0.6316824555397034, 'learning_rate': 9.807510844948267e-06, 'epoch': 0.54}

 18%|█▊        | 2394/13317 [24:39<1:51:42,  1.63it/s]
 18%|█▊        | 2395/13317 [24:40<1:51:38,  1.63it/s]
                                                      
{'loss': 0.2824, 'grad_norm': 0.5450043082237244, 'learning_rate': 9.807150520710912e-06, 'epoch': 0.54}

 18%|█▊        | 2395/13317 [24:40<1:51:38,  1.63it/s]
 18%|█▊        | 2396/13317 [24:40<1:51:37,  1.63it/s]
                                                      
{'loss': 0.2633, 'grad_norm': 0.6558758616447449, 'learning_rate': 9.806789866171417e-06, 'epoch': 0.54}

 18%|█▊        | 2396/13317 [24:40<1:51:37,  1.63it/s]
 18%|█▊        | 2397/13317 [24:41<1:51:36,  1.63it/s]
                                                      
{'loss': 0.2266, 'grad_norm': 0.5026245713233948, 'learning_rate': 9.806428881354561e-06, 'epoch': 0.54}

 18%|█▊        | 2397/13317 [24:41<1:51:36,  1.63it/s]
 18%|█▊        | 2398/13317 [24:41<1:51:35,  1.63it/s]
                                                      
{'loss': 0.2754, 'grad_norm': 0.5485198497772217, 'learning_rate': 9.806067566285153e-06, 'epoch': 0.54}

 18%|█▊        | 2398/13317 [24:42<1:51:35,  1.63it/s]
 18%|█▊        | 2399/13317 [24:42<1:51:34,  1.63it/s]
                                                      
{'loss': 0.2686, 'grad_norm': 0.6724951267242432, 'learning_rate': 9.805705920988014e-06, 'epoch': 0.54}

 18%|█▊        | 2399/13317 [24:42<1:51:34,  1.63it/s]
 18%|█▊        | 2400/13317 [24:43<1:51:34,  1.63it/s]
                                                      
{'loss': 0.3569, 'grad_norm': 0.6314237117767334, 'learning_rate': 9.805343945487995e-06, 'epoch': 0.54}

 18%|█▊        | 2400/13317 [24:43<1:51:34,  1.63it/s]
 18%|█▊        | 2401/13317 [24:43<1:51:34,  1.63it/s]
                                                      
{'loss': 0.2411, 'grad_norm': 0.5400872826576233, 'learning_rate': 9.804981639809965e-06, 'epoch': 0.54}

 18%|█▊        | 2401/13317 [24:43<1:51:34,  1.63it/s]
 18%|█▊        | 2402/13317 [24:44<1:51:33,  1.63it/s]
                                                      
{'loss': 0.3121, 'grad_norm': 0.593433678150177, 'learning_rate': 9.804619003978822e-06, 'epoch': 0.54}

 18%|█▊        | 2402/13317 [24:44<1:51:33,  1.63it/s]
 18%|█▊        | 2403/13317 [24:45<1:51:32,  1.63it/s]
                                                      
{'loss': 0.3148, 'grad_norm': 0.6430307030677795, 'learning_rate': 9.804256038019482e-06, 'epoch': 0.54}

 18%|█▊        | 2403/13317 [24:45<1:51:32,  1.63it/s]
 18%|█▊        | 2404/13317 [24:45<1:51:33,  1.63it/s]
                                                      
{'loss': 0.297, 'grad_norm': 0.6332643628120422, 'learning_rate': 9.803892741956883e-06, 'epoch': 0.54}

 18%|█▊        | 2404/13317 [24:45<1:51:33,  1.63it/s]
 18%|█▊        | 2405/13317 [24:46<1:51:34,  1.63it/s]
                                                      
{'loss': 0.297, 'grad_norm': 0.5291303396224976, 'learning_rate': 9.803529115815988e-06, 'epoch': 0.54}

 18%|█▊        | 2405/13317 [24:46<1:51:34,  1.63it/s]
 18%|█▊        | 2406/13317 [24:46<1:51:35,  1.63it/s]
                                                      
{'loss': 0.2596, 'grad_norm': 0.5063179135322571, 'learning_rate': 9.803165159621784e-06, 'epoch': 0.54}

 18%|█▊        | 2406/13317 [24:47<1:51:35,  1.63it/s]
 18%|█▊        | 2407/13317 [24:47<1:51:35,  1.63it/s]
                                                      
{'loss': 0.275, 'grad_norm': 0.5585975646972656, 'learning_rate': 9.802800873399274e-06, 'epoch': 0.54}

 18%|█▊        | 2407/13317 [24:47<1:51:35,  1.63it/s]
 18%|█▊        | 2408/13317 [24:48<1:51:36,  1.63it/s]
                                                      
{'loss': 0.2423, 'grad_norm': 0.4969949424266815, 'learning_rate': 9.802436257173494e-06, 'epoch': 0.54}

 18%|█▊        | 2408/13317 [24:48<1:51:36,  1.63it/s]
 18%|█▊        | 2409/13317 [24:48<1:51:33,  1.63it/s]
                                                      
{'loss': 0.2415, 'grad_norm': 0.5961881875991821, 'learning_rate': 9.802071310969493e-06, 'epoch': 0.54}

 18%|█▊        | 2409/13317 [24:48<1:51:33,  1.63it/s]
 18%|█▊        | 2410/13317 [24:49<1:51:37,  1.63it/s]
                                                      
{'loss': 0.3234, 'grad_norm': 0.6543909907341003, 'learning_rate': 9.801706034812345e-06, 'epoch': 0.54}

 18%|█▊        | 2410/13317 [24:49<1:51:37,  1.63it/s]
 18%|█▊        | 2411/13317 [24:49<1:51:35,  1.63it/s]
                                                      
{'loss': 0.2893, 'grad_norm': 0.5307403802871704, 'learning_rate': 9.801340428727156e-06, 'epoch': 0.54}

 18%|█▊        | 2411/13317 [24:50<1:51:35,  1.63it/s]
 18%|█▊        | 2412/13317 [24:50<1:51:32,  1.63it/s]
                                                      
{'loss': 0.3083, 'grad_norm': 0.6901949644088745, 'learning_rate': 9.800974492739039e-06, 'epoch': 0.54}

 18%|█▊        | 2412/13317 [24:50<1:51:32,  1.63it/s]
 18%|█▊        | 2413/13317 [24:51<1:51:33,  1.63it/s]
                                                      
{'loss': 0.3027, 'grad_norm': 0.5811121463775635, 'learning_rate': 9.800608226873143e-06, 'epoch': 0.54}

 18%|█▊        | 2413/13317 [24:51<1:51:33,  1.63it/s]
 18%|█▊        | 2414/13317 [24:51<1:51:31,  1.63it/s]
                                                      
{'loss': 0.2304, 'grad_norm': 0.5529397130012512, 'learning_rate': 9.800241631154629e-06, 'epoch': 0.54}

 18%|█▊        | 2414/13317 [24:51<1:51:31,  1.63it/s]
 18%|█▊        | 2415/13317 [24:52<1:51:29,  1.63it/s]
                                                      
{'loss': 0.2065, 'grad_norm': 0.4657454788684845, 'learning_rate': 9.799874705608692e-06, 'epoch': 0.54}

 18%|█▊        | 2415/13317 [24:52<1:51:29,  1.63it/s]
 18%|█▊        | 2416/13317 [24:53<1:51:27,  1.63it/s]
                                                      
{'loss': 0.2511, 'grad_norm': 0.5582900643348694, 'learning_rate': 9.799507450260539e-06, 'epoch': 0.54}

 18%|█▊        | 2416/13317 [24:53<1:51:27,  1.63it/s]
 18%|█▊        | 2417/13317 [24:53<1:51:28,  1.63it/s]
                                                      
{'loss': 0.2862, 'grad_norm': 0.6103206872940063, 'learning_rate': 9.799139865135408e-06, 'epoch': 0.54}

 18%|█▊        | 2417/13317 [24:53<1:51:28,  1.63it/s]
 18%|█▊        | 2418/13317 [24:54<1:51:28,  1.63it/s]
                                                      
{'loss': 0.2234, 'grad_norm': 0.5003435611724854, 'learning_rate': 9.798771950258553e-06, 'epoch': 0.54}

 18%|█▊        | 2418/13317 [24:54<1:51:28,  1.63it/s]
 18%|█▊        | 2419/13317 [24:54<1:51:24,  1.63it/s]
                                                      
{'loss': 0.1872, 'grad_norm': 0.5610058903694153, 'learning_rate': 9.798403705655253e-06, 'epoch': 0.54}

 18%|█▊        | 2419/13317 [24:54<1:51:24,  1.63it/s]
 18%|█▊        | 2420/13317 [24:55<1:51:22,  1.63it/s]
                                                      
{'loss': 0.2987, 'grad_norm': 0.6844501495361328, 'learning_rate': 9.798035131350814e-06, 'epoch': 0.55}

 18%|█▊        | 2420/13317 [24:55<1:51:22,  1.63it/s]
 18%|█▊        | 2421/13317 [24:56<1:51:19,  1.63it/s]
                                                      
{'loss': 0.2435, 'grad_norm': 0.5352789759635925, 'learning_rate': 9.79766622737056e-06, 'epoch': 0.55}

 18%|█▊        | 2421/13317 [24:56<1:51:19,  1.63it/s]
 18%|█▊        | 2422/13317 [24:56<1:51:16,  1.63it/s]
                                                      
{'loss': 0.2343, 'grad_norm': 0.5529120564460754, 'learning_rate': 9.797296993739837e-06, 'epoch': 0.55}

 18%|█▊        | 2422/13317 [24:56<1:51:16,  1.63it/s]
 18%|█▊        | 2423/13317 [24:57<1:51:21,  1.63it/s]
                                                      
{'loss': 0.3013, 'grad_norm': 0.5915618538856506, 'learning_rate': 9.796927430484016e-06, 'epoch': 0.55}

 18%|█▊        | 2423/13317 [24:57<1:51:21,  1.63it/s]
 18%|█▊        | 2424/13317 [24:57<1:51:23,  1.63it/s]
                                                      
{'loss': 0.3076, 'grad_norm': 0.5392407774925232, 'learning_rate': 9.796557537628489e-06, 'epoch': 0.55}

 18%|█▊        | 2424/13317 [24:58<1:51:23,  1.63it/s]
 18%|█▊        | 2425/13317 [24:58<1:51:22,  1.63it/s]
                                                      
{'loss': 0.2604, 'grad_norm': 0.5722216367721558, 'learning_rate': 9.796187315198673e-06, 'epoch': 0.55}

 18%|█▊        | 2425/13317 [24:58<1:51:22,  1.63it/s]
 18%|█▊        | 2426/13317 [24:59<1:51:21,  1.63it/s]
                                                      
{'loss': 0.1853, 'grad_norm': 0.4899540841579437, 'learning_rate': 9.795816763220005e-06, 'epoch': 0.55}

 18%|█▊        | 2426/13317 [24:59<1:51:21,  1.63it/s]
 18%|█▊        | 2427/13317 [24:59<1:51:21,  1.63it/s]
                                                      
{'loss': 0.2831, 'grad_norm': 0.6260392665863037, 'learning_rate': 9.795445881717946e-06, 'epoch': 0.55}

 18%|█▊        | 2427/13317 [24:59<1:51:21,  1.63it/s]
 18%|█▊        | 2428/13317 [25:00<1:51:18,  1.63it/s]
                                                      
{'loss': 0.2434, 'grad_norm': 0.5070266127586365, 'learning_rate': 9.795074670717982e-06, 'epoch': 0.55}

 18%|█▊        | 2428/13317 [25:00<1:51:18,  1.63it/s]
 18%|█▊        | 2429/13317 [25:00<1:51:19,  1.63it/s]
                                                      
{'loss': 0.2548, 'grad_norm': 0.5681193470954895, 'learning_rate': 9.794703130245615e-06, 'epoch': 0.55}

 18%|█▊        | 2429/13317 [25:01<1:51:19,  1.63it/s]
 18%|█▊        | 2430/13317 [25:01<1:51:16,  1.63it/s]
                                                      
{'loss': 0.368, 'grad_norm': 0.5764867663383484, 'learning_rate': 9.794331260326378e-06, 'epoch': 0.55}

 18%|█▊        | 2430/13317 [25:01<1:51:16,  1.63it/s]
 18%|█▊        | 2431/13317 [25:02<1:51:11,  1.63it/s]
                                                      
{'loss': 0.3086, 'grad_norm': 0.5988187789916992, 'learning_rate': 9.793959060985817e-06, 'epoch': 0.55}

 18%|█▊        | 2431/13317 [25:02<1:51:11,  1.63it/s]
 18%|█▊        | 2432/13317 [25:02<1:51:11,  1.63it/s]
                                                      
{'loss': 0.2354, 'grad_norm': 0.598723292350769, 'learning_rate': 9.793586532249513e-06, 'epoch': 0.55}

 18%|█▊        | 2432/13317 [25:02<1:51:11,  1.63it/s]
 18%|█▊        | 2433/13317 [25:03<1:51:13,  1.63it/s]
                                                      
{'loss': 0.2431, 'grad_norm': 0.5628484487533569, 'learning_rate': 9.793213674143055e-06, 'epoch': 0.55}

 18%|█▊        | 2433/13317 [25:03<1:51:13,  1.63it/s]
 18%|█▊        | 2434/13317 [25:04<1:51:11,  1.63it/s]
                                                      
{'loss': 0.3256, 'grad_norm': 0.6626503467559814, 'learning_rate': 9.79284048669207e-06, 'epoch': 0.55}

 18%|█▊        | 2434/13317 [25:04<1:51:11,  1.63it/s]
 18%|█▊        | 2435/13317 [25:04<1:51:15,  1.63it/s]
                                                      
{'loss': 0.3085, 'grad_norm': 0.5772507786750793, 'learning_rate': 9.792466969922193e-06, 'epoch': 0.55}

 18%|█▊        | 2435/13317 [25:04<1:51:15,  1.63it/s]
 18%|█▊        | 2436/13317 [25:05<1:51:14,  1.63it/s]
                                                      
{'loss': 0.3496, 'grad_norm': 0.70481276512146, 'learning_rate': 9.792093123859095e-06, 'epoch': 0.55}

 18%|█▊        | 2436/13317 [25:05<1:51:14,  1.63it/s]
 18%|█▊        | 2437/13317 [25:05<1:51:15,  1.63it/s]
                                                      
{'loss': 0.2461, 'grad_norm': 0.6033056974411011, 'learning_rate': 9.791718948528457e-06, 'epoch': 0.55}

 18%|█▊        | 2437/13317 [25:06<1:51:15,  1.63it/s]
 18%|█▊        | 2438/13317 [25:06<1:51:13,  1.63it/s]
                                                      
{'loss': 0.2264, 'grad_norm': 0.5069847106933594, 'learning_rate': 9.791344443955995e-06, 'epoch': 0.55}

 18%|█▊        | 2438/13317 [25:06<1:51:13,  1.63it/s]
 18%|█▊        | 2439/13317 [25:07<1:51:10,  1.63it/s]
                                                      
{'loss': 0.2846, 'grad_norm': 0.7294721007347107, 'learning_rate': 9.790969610167434e-06, 'epoch': 0.55}

 18%|█▊        | 2439/13317 [25:07<1:51:10,  1.63it/s]
 18%|█▊        | 2440/13317 [25:07<1:51:04,  1.63it/s]
                                                      
{'loss': 0.2454, 'grad_norm': 0.5647783279418945, 'learning_rate': 9.790594447188537e-06, 'epoch': 0.55}

 18%|█▊        | 2440/13317 [25:07<1:51:04,  1.63it/s]
 18%|█▊        | 2441/13317 [25:08<1:51:07,  1.63it/s]
                                                      
{'loss': 0.2644, 'grad_norm': 0.5621007084846497, 'learning_rate': 9.790218955045075e-06, 'epoch': 0.55}

 18%|█▊        | 2441/13317 [25:08<1:51:07,  1.63it/s]
 18%|█▊        | 2442/13317 [25:08<1:51:05,  1.63it/s]
                                                      
{'loss': 0.2493, 'grad_norm': 0.6164500117301941, 'learning_rate': 9.789843133762852e-06, 'epoch': 0.55}

 18%|█▊        | 2442/13317 [25:09<1:51:05,  1.63it/s]
 18%|█▊        | 2443/13317 [25:09<1:51:07,  1.63it/s]
                                                      
{'loss': 0.3437, 'grad_norm': 0.6644439697265625, 'learning_rate': 9.78946698336769e-06, 'epoch': 0.55}

 18%|█▊        | 2443/13317 [25:09<1:51:07,  1.63it/s]
 18%|█▊        | 2444/13317 [25:10<1:51:08,  1.63it/s]
                                                      
{'loss': 0.2973, 'grad_norm': 0.5860273838043213, 'learning_rate': 9.789090503885435e-06, 'epoch': 0.55}

 18%|█▊        | 2444/13317 [25:10<1:51:08,  1.63it/s]
 18%|█▊        | 2445/13317 [25:10<1:51:07,  1.63it/s]
                                                      
{'loss': 0.359, 'grad_norm': 0.6260866522789001, 'learning_rate': 9.788713695341954e-06, 'epoch': 0.55}

 18%|█▊        | 2445/13317 [25:10<1:51:07,  1.63it/s]
 18%|█▊        | 2446/13317 [25:11<1:51:07,  1.63it/s]
                                                      
{'loss': 0.2683, 'grad_norm': 0.5605093836784363, 'learning_rate': 9.78833655776314e-06, 'epoch': 0.55}

 18%|█▊        | 2446/13317 [25:11<1:51:07,  1.63it/s]
 18%|█▊        | 2447/13317 [25:12<1:51:08,  1.63it/s]
                                                      
{'loss': 0.3335, 'grad_norm': 0.6489800214767456, 'learning_rate': 9.787959091174904e-06, 'epoch': 0.55}

 18%|█▊        | 2447/13317 [25:12<1:51:08,  1.63it/s]
 18%|█▊        | 2448/13317 [25:12<1:51:05,  1.63it/s]
                                                      
{'loss': 0.1934, 'grad_norm': 0.48849549889564514, 'learning_rate': 9.787581295603181e-06, 'epoch': 0.55}

 18%|█▊        | 2448/13317 [25:12<1:51:05,  1.63it/s]
 18%|█▊        | 2449/13317 [25:13<1:51:11,  1.63it/s]
                                                      
{'loss': 0.2182, 'grad_norm': 0.5230580568313599, 'learning_rate': 9.787203171073933e-06, 'epoch': 0.55}

 18%|█▊        | 2449/13317 [25:13<1:51:11,  1.63it/s]
 18%|█▊        | 2450/13317 [25:13<1:51:10,  1.63it/s]
                                                      
{'loss': 0.2625, 'grad_norm': 0.5476917624473572, 'learning_rate': 9.78682471761314e-06, 'epoch': 0.55}

 18%|█▊        | 2450/13317 [25:13<1:51:10,  1.63it/s]
 18%|█▊        | 2451/13317 [25:14<1:51:14,  1.63it/s]
                                                      
{'loss': 0.1865, 'grad_norm': 0.4048476219177246, 'learning_rate': 9.786445935246802e-06, 'epoch': 0.55}

 18%|█▊        | 2451/13317 [25:14<1:51:14,  1.63it/s]
 18%|█▊        | 2452/13317 [25:15<1:51:14,  1.63it/s]
                                                      
{'loss': 0.3019, 'grad_norm': 0.5959759950637817, 'learning_rate': 9.786066824000952e-06, 'epoch': 0.55}

 18%|█▊        | 2452/13317 [25:15<1:51:14,  1.63it/s]
 18%|█▊        | 2453/13317 [25:15<1:51:09,  1.63it/s]
                                                      
{'loss': 0.242, 'grad_norm': 0.509423553943634, 'learning_rate': 9.785687383901633e-06, 'epoch': 0.55}

 18%|█▊        | 2453/13317 [25:15<1:51:09,  1.63it/s]
 18%|█▊        | 2454/13317 [25:16<1:51:05,  1.63it/s]
                                                      
{'loss': 0.3276, 'grad_norm': 0.7274518013000488, 'learning_rate': 9.785307614974922e-06, 'epoch': 0.55}

 18%|█▊        | 2454/13317 [25:16<1:51:05,  1.63it/s]
 18%|█▊        | 2455/13317 [25:16<1:51:03,  1.63it/s]
                                                      
{'loss': 0.2665, 'grad_norm': 0.6279539465904236, 'learning_rate': 9.784927517246907e-06, 'epoch': 0.55}

 18%|█▊        | 2455/13317 [25:17<1:51:03,  1.63it/s]
 18%|█▊        | 2456/13317 [25:17<1:51:00,  1.63it/s]
                                                      
{'loss': 0.2905, 'grad_norm': 0.638610303401947, 'learning_rate': 9.78454709074371e-06, 'epoch': 0.55}

 18%|█▊        | 2456/13317 [25:17<1:51:00,  1.63it/s]
 18%|█▊        | 2457/13317 [25:18<1:51:01,  1.63it/s]
                                                      
{'loss': 0.2509, 'grad_norm': 0.5501888394355774, 'learning_rate': 9.784166335491467e-06, 'epoch': 0.55}

 18%|█▊        | 2457/13317 [25:18<1:51:01,  1.63it/s]
 18%|█▊        | 2458/13317 [25:18<1:51:04,  1.63it/s]
                                                      
{'loss': 0.2929, 'grad_norm': 0.587063729763031, 'learning_rate': 9.783785251516342e-06, 'epoch': 0.55}

 18%|█▊        | 2458/13317 [25:18<1:51:04,  1.63it/s]
 18%|█▊        | 2459/13317 [25:19<1:51:03,  1.63it/s]
                                                      
{'loss': 0.2859, 'grad_norm': 0.5315777659416199, 'learning_rate': 9.783403838844519e-06, 'epoch': 0.55}

 18%|█▊        | 2459/13317 [25:19<1:51:03,  1.63it/s]
 18%|█▊        | 2460/13317 [25:19<1:51:02,  1.63it/s]
                                                      
{'loss': 0.2831, 'grad_norm': 0.5203699469566345, 'learning_rate': 9.783022097502204e-06, 'epoch': 0.55}

 18%|█▊        | 2460/13317 [25:20<1:51:02,  1.63it/s]
 18%|█▊        | 2461/13317 [25:20<1:51:03,  1.63it/s]
                                                      
{'loss': 0.2726, 'grad_norm': 0.5992633700370789, 'learning_rate': 9.78264002751563e-06, 'epoch': 0.55}

 18%|█▊        | 2461/13317 [25:20<1:51:03,  1.63it/s]
 18%|█▊        | 2462/13317 [25:21<1:50:56,  1.63it/s]
                                                      
{'loss': 0.2986, 'grad_norm': 0.5992740392684937, 'learning_rate': 9.782257628911042e-06, 'epoch': 0.55}

 18%|█▊        | 2462/13317 [25:21<1:50:56,  1.63it/s]
 18%|█▊        | 2463/13317 [25:21<1:50:56,  1.63it/s]
                                                      
{'loss': 0.2911, 'grad_norm': 0.5407317280769348, 'learning_rate': 9.781874901714724e-06, 'epoch': 0.55}

 18%|█▊        | 2463/13317 [25:21<1:50:56,  1.63it/s]
 19%|█▊        | 2464/13317 [25:22<1:50:58,  1.63it/s]
                                                      
{'loss': 0.3247, 'grad_norm': 0.6018262505531311, 'learning_rate': 9.781491845952967e-06, 'epoch': 0.56}

 19%|█▊        | 2464/13317 [25:22<1:50:58,  1.63it/s]
 19%|█▊        | 2465/13317 [25:23<1:50:57,  1.63it/s]
                                                      
{'loss': 0.2535, 'grad_norm': 0.5655821561813354, 'learning_rate': 9.781108461652093e-06, 'epoch': 0.56}

 19%|█▊        | 2465/13317 [25:23<1:50:57,  1.63it/s]
 19%|█▊        | 2466/13317 [25:23<1:50:58,  1.63it/s]
                                                      
{'loss': 0.3513, 'grad_norm': 0.6462979316711426, 'learning_rate': 9.780724748838444e-06, 'epoch': 0.56}

 19%|█▊        | 2466/13317 [25:23<1:50:58,  1.63it/s]
 19%|█▊        | 2467/13317 [25:24<1:50:56,  1.63it/s]
                                                      
{'loss': 0.2804, 'grad_norm': 0.5589683055877686, 'learning_rate': 9.780340707538386e-06, 'epoch': 0.56}

 19%|█▊        | 2467/13317 [25:24<1:50:56,  1.63it/s]
 19%|█▊        | 2468/13317 [25:24<1:50:52,  1.63it/s]
                                                      
{'loss': 0.3816, 'grad_norm': 0.6036646366119385, 'learning_rate': 9.779956337778307e-06, 'epoch': 0.56}

 19%|█▊        | 2468/13317 [25:25<1:50:52,  1.63it/s]
 19%|█▊        | 2469/13317 [25:25<1:50:54,  1.63it/s]
                                                      
{'loss': 0.3275, 'grad_norm': 0.562882661819458, 'learning_rate': 9.779571639584615e-06, 'epoch': 0.56}

 19%|█▊        | 2469/13317 [25:25<1:50:54,  1.63it/s]
 19%|█▊        | 2470/13317 [25:26<1:50:52,  1.63it/s]
                                                      
{'loss': 0.263, 'grad_norm': 0.577944278717041, 'learning_rate': 9.779186612983745e-06, 'epoch': 0.56}

 19%|█▊        | 2470/13317 [25:26<1:50:52,  1.63it/s]
 19%|█▊        | 2471/13317 [25:26<1:50:53,  1.63it/s]
                                                      
{'loss': 0.2232, 'grad_norm': 0.5113176107406616, 'learning_rate': 9.778801258002153e-06, 'epoch': 0.56}

 19%|█▊        | 2471/13317 [25:26<1:50:53,  1.63it/s]
 19%|█▊        | 2472/13317 [25:27<1:50:50,  1.63it/s]
                                                      
{'loss': 0.3128, 'grad_norm': 0.5709730982780457, 'learning_rate': 9.778415574666315e-06, 'epoch': 0.56}

 19%|█▊        | 2472/13317 [25:27<1:50:50,  1.63it/s]
 19%|█▊        | 2473/13317 [25:27<1:50:57,  1.63it/s]
                                                      
{'loss': 0.2349, 'grad_norm': 0.4925057888031006, 'learning_rate': 9.778029563002733e-06, 'epoch': 0.56}

 19%|█▊        | 2473/13317 [25:28<1:50:57,  1.63it/s]
 19%|█▊        | 2474/13317 [25:28<1:50:53,  1.63it/s]
                                                      
{'loss': 0.2793, 'grad_norm': 0.589152455329895, 'learning_rate': 9.77764322303793e-06, 'epoch': 0.56}

 19%|█▊        | 2474/13317 [25:28<1:50:53,  1.63it/s]
 19%|█▊        | 2475/13317 [25:29<1:50:54,  1.63it/s]
                                                      
{'loss': 0.3015, 'grad_norm': 0.5304453372955322, 'learning_rate': 9.777256554798449e-06, 'epoch': 0.56}

 19%|█▊        | 2475/13317 [25:29<1:50:54,  1.63it/s]
 19%|█▊        | 2476/13317 [25:29<1:50:51,  1.63it/s]
                                                      
{'loss': 0.2125, 'grad_norm': 0.5183631777763367, 'learning_rate': 9.77686955831086e-06, 'epoch': 0.56}

 19%|█▊        | 2476/13317 [25:29<1:50:51,  1.63it/s]
 19%|█▊        | 2477/13317 [25:30<1:50:52,  1.63it/s]
                                                      
{'loss': 0.2551, 'grad_norm': 0.6092011332511902, 'learning_rate': 9.776482233601756e-06, 'epoch': 0.56}

 19%|█▊        | 2477/13317 [25:30<1:50:52,  1.63it/s]
 19%|█▊        | 2478/13317 [25:31<1:50:48,  1.63it/s]
                                                      
{'loss': 0.3364, 'grad_norm': 0.6077736616134644, 'learning_rate': 9.776094580697748e-06, 'epoch': 0.56}

 19%|█▊        | 2478/13317 [25:31<1:50:48,  1.63it/s]
 19%|█▊        | 2479/13317 [25:31<1:50:48,  1.63it/s]
                                                      
{'loss': 0.3011, 'grad_norm': 0.6466400623321533, 'learning_rate': 9.775706599625472e-06, 'epoch': 0.56}

 19%|█▊        | 2479/13317 [25:31<1:50:48,  1.63it/s]
 19%|█▊        | 2480/13317 [25:32<1:50:56,  1.63it/s]
                                                      
{'loss': 0.2399, 'grad_norm': 0.5299805998802185, 'learning_rate': 9.775318290411585e-06, 'epoch': 0.56}

 19%|█▊        | 2480/13317 [25:32<1:50:56,  1.63it/s]
 19%|█▊        | 2481/13317 [25:32<1:50:52,  1.63it/s]
                                                      
{'loss': 0.2779, 'grad_norm': 0.6597146987915039, 'learning_rate': 9.774929653082771e-06, 'epoch': 0.56}

 19%|█▊        | 2481/13317 [25:33<1:50:52,  1.63it/s]
 19%|█▊        | 2482/13317 [25:33<1:50:49,  1.63it/s]
                                                      
{'loss': 0.2532, 'grad_norm': 0.5940714478492737, 'learning_rate': 9.774540687665733e-06, 'epoch': 0.56}

 19%|█▊        | 2482/13317 [25:33<1:50:49,  1.63it/s]
 19%|█▊        | 2483/13317 [25:34<1:50:47,  1.63it/s]
                                                      
{'loss': 0.2278, 'grad_norm': 0.6210017800331116, 'learning_rate': 9.774151394187195e-06, 'epoch': 0.56}

 19%|█▊        | 2483/13317 [25:34<1:50:47,  1.63it/s]
 19%|█▊        | 2484/13317 [25:34<1:50:50,  1.63it/s]
                                                      
{'loss': 0.3233, 'grad_norm': 0.6590368151664734, 'learning_rate': 9.773761772673908e-06, 'epoch': 0.56}

 19%|█▊        | 2484/13317 [25:34<1:50:50,  1.63it/s]
 19%|█▊        | 2485/13317 [25:35<1:50:50,  1.63it/s]
                                                      
{'loss': 0.2959, 'grad_norm': 0.6703782677650452, 'learning_rate': 9.77337182315264e-06, 'epoch': 0.56}

 19%|█▊        | 2485/13317 [25:35<1:50:50,  1.63it/s]
 19%|█▊        | 2486/13317 [25:35<1:50:47,  1.63it/s]
                                                      
{'loss': 0.3228, 'grad_norm': 0.686154842376709, 'learning_rate': 9.772981545650187e-06, 'epoch': 0.56}

 19%|█▊        | 2486/13317 [25:36<1:50:47,  1.63it/s]
 19%|█▊        | 2487/13317 [25:36<1:50:49,  1.63it/s]
                                                      
{'loss': 0.2682, 'grad_norm': 0.6281701922416687, 'learning_rate': 9.772590940193366e-06, 'epoch': 0.56}

 19%|█▊        | 2487/13317 [25:36<1:50:49,  1.63it/s]
 19%|█▊        | 2488/13317 [25:37<1:50:52,  1.63it/s]
                                                      
{'loss': 0.3057, 'grad_norm': 0.6054776310920715, 'learning_rate': 9.772200006809011e-06, 'epoch': 0.56}

 19%|█▊        | 2488/13317 [25:37<1:50:52,  1.63it/s]
 19%|█▊        | 2489/13317 [25:37<1:50:50,  1.63it/s]
                                                      
{'loss': 0.2625, 'grad_norm': 0.567385196685791, 'learning_rate': 9.77180874552399e-06, 'epoch': 0.56}

 19%|█▊        | 2489/13317 [25:37<1:50:50,  1.63it/s]
 19%|█▊        | 2490/13317 [25:38<1:50:50,  1.63it/s]
                                                      
{'loss': 0.1927, 'grad_norm': 0.47122707962989807, 'learning_rate': 9.771417156365183e-06, 'epoch': 0.56}

 19%|█▊        | 2490/13317 [25:38<1:50:50,  1.63it/s]
 19%|█▊        | 2491/13317 [25:39<1:50:49,  1.63it/s]
                                                      
{'loss': 0.2557, 'grad_norm': 0.5994471311569214, 'learning_rate': 9.771025239359497e-06, 'epoch': 0.56}

 19%|█▊        | 2491/13317 [25:39<1:50:49,  1.63it/s]
 19%|█▊        | 2492/13317 [25:39<1:50:51,  1.63it/s]
                                                      
{'loss': 0.2625, 'grad_norm': 0.49469253420829773, 'learning_rate': 9.77063299453386e-06, 'epoch': 0.56}

 19%|█▊        | 2492/13317 [25:39<1:50:51,  1.63it/s]
 19%|█▊        | 2493/13317 [25:40<1:50:47,  1.63it/s]
                                                      
{'loss': 0.3184, 'grad_norm': 0.642974853515625, 'learning_rate': 9.770240421915222e-06, 'epoch': 0.56}

 19%|█▊        | 2493/13317 [25:40<1:50:47,  1.63it/s]
 19%|█▊        | 2494/13317 [25:40<1:50:46,  1.63it/s]
                                                      
{'loss': 0.314, 'grad_norm': 0.5688338875770569, 'learning_rate': 9.769847521530562e-06, 'epoch': 0.56}

 19%|█▊        | 2494/13317 [25:40<1:50:46,  1.63it/s]
 19%|█▊        | 2495/13317 [25:41<1:50:47,  1.63it/s]
                                                      
{'loss': 0.1917, 'grad_norm': 0.49490630626678467, 'learning_rate': 9.769454293406871e-06, 'epoch': 0.56}

 19%|█▊        | 2495/13317 [25:41<1:50:47,  1.63it/s]
 19%|█▊        | 2496/13317 [25:42<1:50:39,  1.63it/s]
                                                      
{'loss': 0.2839, 'grad_norm': 0.5658356547355652, 'learning_rate': 9.769060737571172e-06, 'epoch': 0.56}

 19%|█▊        | 2496/13317 [25:42<1:50:39,  1.63it/s]
 19%|█▉        | 2497/13317 [25:42<1:50:40,  1.63it/s]
                                                      
{'loss': 0.2902, 'grad_norm': 0.6035402417182922, 'learning_rate': 9.768666854050504e-06, 'epoch': 0.56}

 19%|█▉        | 2497/13317 [25:42<1:50:40,  1.63it/s]
 19%|█▉        | 2498/13317 [25:43<1:50:37,  1.63it/s]
                                                      
{'loss': 0.3875, 'grad_norm': 0.6006999611854553, 'learning_rate': 9.768272642871931e-06, 'epoch': 0.56}

 19%|█▉        | 2498/13317 [25:43<1:50:37,  1.63it/s]
 19%|█▉        | 2499/13317 [25:43<1:50:48,  1.63it/s]
                                                      
{'loss': 0.2828, 'grad_norm': 0.5630525350570679, 'learning_rate': 9.767878104062541e-06, 'epoch': 0.56}

 19%|█▉        | 2499/13317 [25:44<1:50:48,  1.63it/s]
 19%|█▉        | 2500/13317 [25:44<1:50:44,  1.63it/s]
                                                      
{'loss': 0.2878, 'grad_norm': 0.5892601013183594, 'learning_rate': 9.767483237649441e-06, 'epoch': 0.56}

 19%|█▉        | 2500/13317 [25:44<1:50:44,  1.63it/s]

  0%|          | 0/13 [00:00<?, ?it/s]

 15%|█▌        | 2/13 [00:00<00:00, 13.37it/s]

 31%|███       | 4/13 [00:00<00:01,  8.40it/s]

 38%|███▊      | 5/13 [00:00<00:01,  7.80it/s]

 46%|████▌     | 6/13 [00:00<00:00,  7.43it/s]

 54%|█████▍    | 7/13 [00:00<00:00,  7.18it/s]

 62%|██████▏   | 8/13 [00:01<00:00,  7.02it/s]

 69%|██████▉   | 9/13 [00:01<00:00,  6.91it/s]

 77%|███████▋  | 10/13 [00:01<00:00,  6.84it/s]

 85%|████████▍ | 11/13 [00:01<00:00,  6.79it/s]

 92%|█████████▏| 12/13 [00:01<00:00,  6.75it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]
                                                      

                                               
{'eval_loss': 0.7167089581489563, 'eval_runtime': 1.9516, 'eval_samples_per_second': 102.478, 'eval_steps_per_second': 6.661, 'epoch': 0.56}

 19%|█▉        | 2500/13317 [25:46<1:50:44,  1.63it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]

                                               
 19%|█▉        | 2501/13317 [25:47<3:36:31,  1.20s/it]
                                                      
{'loss': 0.2442, 'grad_norm': 0.6445679068565369, 'learning_rate': 9.767088043659767e-06, 'epoch': 0.56}

 19%|█▉        | 2501/13317 [25:47<3:36:31,  1.20s/it]
 19%|█▉        | 2502/13317 [25:47<3:04:46,  1.03s/it]
                                                      
{'loss': 0.3166, 'grad_norm': 0.6192626953125, 'learning_rate': 9.766692522120667e-06, 'epoch': 0.56}

 19%|█▉        | 2502/13317 [25:47<3:04:46,  1.03s/it]
 19%|█▉        | 2503/13317 [25:48<2:42:28,  1.11it/s]
                                                      
{'loss': 0.2666, 'grad_norm': 0.5286032557487488, 'learning_rate': 9.76629667305932e-06, 'epoch': 0.56}

 19%|█▉        | 2503/13317 [25:48<2:42:28,  1.11it/s]
 19%|█▉        | 2504/13317 [25:48<2:27:07,  1.22it/s]
                                                      
{'loss': 0.1902, 'grad_norm': 0.428637832403183, 'learning_rate': 9.765900496502925e-06, 'epoch': 0.56}

 19%|█▉        | 2504/13317 [25:49<2:27:07,  1.22it/s]
 19%|█▉        | 2505/13317 [25:49<2:16:07,  1.32it/s]
                                                      
{'loss': 0.2288, 'grad_norm': 0.5025879144668579, 'learning_rate': 9.765503992478704e-06, 'epoch': 0.56}

 19%|█▉        | 2505/13317 [25:49<2:16:07,  1.32it/s]
 19%|█▉        | 2506/13317 [25:50<2:08:25,  1.40it/s]
                                                      
{'loss': 0.3666, 'grad_norm': 0.6245514154434204, 'learning_rate': 9.765107161013902e-06, 'epoch': 0.56}

 19%|█▉        | 2506/13317 [25:50<2:08:25,  1.40it/s]
 19%|█▉        | 2507/13317 [25:50<2:03:00,  1.46it/s]
                                                      
{'loss': 0.2504, 'grad_norm': 0.5606364607810974, 'learning_rate': 9.764710002135784e-06, 'epoch': 0.56}

 19%|█▉        | 2507/13317 [25:50<2:03:00,  1.46it/s]
 19%|█▉        | 2508/13317 [25:51<1:59:13,  1.51it/s]
                                                      
{'loss': 0.3555, 'grad_norm': 0.719491720199585, 'learning_rate': 9.764312515871639e-06, 'epoch': 0.56}

 19%|█▉        | 2508/13317 [25:51<1:59:13,  1.51it/s]
 19%|█▉        | 2509/13317 [25:52<1:56:46,  1.54it/s]
                                                      
{'loss': 0.2463, 'grad_norm': 0.5850000977516174, 'learning_rate': 9.763914702248779e-06, 'epoch': 0.57}

 19%|█▉        | 2509/13317 [25:52<1:56:46,  1.54it/s]
 19%|█▉        | 2510/13317 [25:52<1:54:56,  1.57it/s]
                                                      
{'loss': 0.2106, 'grad_norm': 0.44962480664253235, 'learning_rate': 9.763516561294537e-06, 'epoch': 0.57}

 19%|█▉        | 2510/13317 [25:52<1:54:56,  1.57it/s]
 19%|█▉        | 2511/13317 [25:53<1:53:34,  1.59it/s]
                                                      
{'loss': 0.293, 'grad_norm': 0.5603370666503906, 'learning_rate': 9.76311809303627e-06, 'epoch': 0.57}

 19%|█▉        | 2511/13317 [25:53<1:53:34,  1.59it/s]
 19%|█▉        | 2512/13317 [25:53<1:52:40,  1.60it/s]
                                                      
{'loss': 0.2625, 'grad_norm': 0.5484066605567932, 'learning_rate': 9.762719297501357e-06, 'epoch': 0.57}

 19%|█▉        | 2512/13317 [25:54<1:52:40,  1.60it/s]
 19%|█▉        | 2513/13317 [25:54<1:52:00,  1.61it/s]
                                                      
{'loss': 0.2872, 'grad_norm': 0.6037606596946716, 'learning_rate': 9.7623201747172e-06, 'epoch': 0.57}

 19%|█▉        | 2513/13317 [25:54<1:52:00,  1.61it/s]
 19%|█▉        | 2514/13317 [25:55<1:51:42,  1.61it/s]
                                                      
{'loss': 0.2531, 'grad_norm': 0.5081239938735962, 'learning_rate': 9.761920724711225e-06, 'epoch': 0.57}

 19%|█▉        | 2514/13317 [25:55<1:51:42,  1.61it/s]
 19%|█▉        | 2515/13317 [25:55<1:51:17,  1.62it/s]
                                                      
{'loss': 0.2859, 'grad_norm': 0.4979511797428131, 'learning_rate': 9.761520947510875e-06, 'epoch': 0.57}

 19%|█▉        | 2515/13317 [25:55<1:51:17,  1.62it/s]
 19%|█▉        | 2516/13317 [25:56<1:51:03,  1.62it/s]
                                                      
{'loss': 0.2407, 'grad_norm': 0.5401044487953186, 'learning_rate': 9.76112084314362e-06, 'epoch': 0.57}

 19%|█▉        | 2516/13317 [25:56<1:51:03,  1.62it/s]
 19%|█▉        | 2517/13317 [25:56<1:50:53,  1.62it/s]
                                                      
{'loss': 0.2261, 'grad_norm': 0.6898863911628723, 'learning_rate': 9.760720411636951e-06, 'epoch': 0.57}

 19%|█▉        | 2517/13317 [25:57<1:50:53,  1.62it/s]
 19%|█▉        | 2518/13317 [25:57<1:50:46,  1.62it/s]
                                                      
{'loss': 0.2656, 'grad_norm': 0.5604248642921448, 'learning_rate': 9.760319653018383e-06, 'epoch': 0.57}

 19%|█▉        | 2518/13317 [25:57<1:50:46,  1.62it/s]
 19%|█▉        | 2519/13317 [25:58<1:50:40,  1.63it/s]
                                                      
{'loss': 0.2086, 'grad_norm': 0.47286340594291687, 'learning_rate': 9.759918567315452e-06, 'epoch': 0.57}

 19%|█▉        | 2519/13317 [25:58<1:50:40,  1.63it/s]
 19%|█▉        | 2520/13317 [25:58<1:50:33,  1.63it/s]
                                                      
{'loss': 0.235, 'grad_norm': 0.5440824031829834, 'learning_rate': 9.759517154555715e-06, 'epoch': 0.57}

 19%|█▉        | 2520/13317 [25:58<1:50:33,  1.63it/s]
 19%|█▉        | 2521/13317 [25:59<1:50:33,  1.63it/s]
                                                      
{'loss': 0.2676, 'grad_norm': 0.5274617075920105, 'learning_rate': 9.759115414766756e-06, 'epoch': 0.57}

 19%|█▉        | 2521/13317 [25:59<1:50:33,  1.63it/s]
 19%|█▉        | 2522/13317 [26:00<1:50:34,  1.63it/s]
                                                      
{'loss': 0.2147, 'grad_norm': 0.5260202884674072, 'learning_rate': 9.758713347976179e-06, 'epoch': 0.57}

 19%|█▉        | 2522/13317 [26:00<1:50:34,  1.63it/s]
 19%|█▉        | 2523/13317 [26:00<1:50:30,  1.63it/s]
                                                      
{'loss': 0.2683, 'grad_norm': 0.499332070350647, 'learning_rate': 9.758310954211607e-06, 'epoch': 0.57}

 19%|█▉        | 2523/13317 [26:00<1:50:30,  1.63it/s]
 19%|█▉        | 2524/13317 [26:01<1:50:32,  1.63it/s]
                                                      
{'loss': 0.2076, 'grad_norm': 0.5543277859687805, 'learning_rate': 9.75790823350069e-06, 'epoch': 0.57}

 19%|█▉        | 2524/13317 [26:01<1:50:32,  1.63it/s]
 19%|█▉        | 2525/13317 [26:01<1:50:27,  1.63it/s]
                                                      
{'loss': 0.3008, 'grad_norm': 0.5405012965202332, 'learning_rate': 9.757505185871102e-06, 'epoch': 0.57}

 19%|█▉        | 2525/13317 [26:01<1:50:27,  1.63it/s]
 19%|█▉        | 2526/13317 [26:02<1:50:21,  1.63it/s]
                                                      
{'loss': 0.1668, 'grad_norm': 0.4404791295528412, 'learning_rate': 9.757101811350534e-06, 'epoch': 0.57}

 19%|█▉        | 2526/13317 [26:02<1:50:21,  1.63it/s]
 19%|█▉        | 2527/13317 [26:03<1:50:31,  1.63it/s]
                                                      
{'loss': 0.35, 'grad_norm': 0.7105923295021057, 'learning_rate': 9.756698109966702e-06, 'epoch': 0.57}

 19%|█▉        | 2527/13317 [26:03<1:50:31,  1.63it/s]
 19%|█▉        | 2528/13317 [26:03<1:50:25,  1.63it/s]
                                                      
{'loss': 0.3452, 'grad_norm': 0.6499127745628357, 'learning_rate': 9.756294081747345e-06, 'epoch': 0.57}

 19%|█▉        | 2528/13317 [26:03<1:50:25,  1.63it/s]
 19%|█▉        | 2529/13317 [26:04<1:50:19,  1.63it/s]
                                                      
{'loss': 0.1824, 'grad_norm': 0.4662121832370758, 'learning_rate': 9.755889726720224e-06, 'epoch': 0.57}

 19%|█▉        | 2529/13317 [26:04<1:50:19,  1.63it/s]
 19%|█▉        | 2530/13317 [26:04<1:50:21,  1.63it/s]
                                                      
{'loss': 0.256, 'grad_norm': 0.5433530807495117, 'learning_rate': 9.755485044913123e-06, 'epoch': 0.57}

 19%|█▉        | 2530/13317 [26:05<1:50:21,  1.63it/s]
 19%|█▉        | 2531/13317 [26:05<1:50:22,  1.63it/s]
                                                      
{'loss': 0.1927, 'grad_norm': 0.46796974539756775, 'learning_rate': 9.755080036353847e-06, 'epoch': 0.57}

 19%|█▉        | 2531/13317 [26:05<1:50:22,  1.63it/s]
 19%|█▉        | 2532/13317 [26:06<1:50:30,  1.63it/s]
                                                      
{'loss': 0.2812, 'grad_norm': 0.627766489982605, 'learning_rate': 9.754674701070226e-06, 'epoch': 0.57}

 19%|█▉        | 2532/13317 [26:06<1:50:30,  1.63it/s]
 19%|█▉        | 2533/13317 [26:06<1:50:27,  1.63it/s]
                                                      
{'loss': 0.3125, 'grad_norm': 0.8093714714050293, 'learning_rate': 9.754269039090109e-06, 'epoch': 0.57}

 19%|█▉        | 2533/13317 [26:06<1:50:27,  1.63it/s]
 19%|█▉        | 2534/13317 [26:07<1:50:25,  1.63it/s]
                                                      
{'loss': 0.2907, 'grad_norm': 0.595430850982666, 'learning_rate': 9.75386305044137e-06, 'epoch': 0.57}

 19%|█▉        | 2534/13317 [26:07<1:50:25,  1.63it/s]
 19%|█▉        | 2535/13317 [26:08<1:50:19,  1.63it/s]
                                                      
{'loss': 0.2441, 'grad_norm': 0.5652787685394287, 'learning_rate': 9.753456735151905e-06, 'epoch': 0.57}

 19%|█▉        | 2535/13317 [26:08<1:50:19,  1.63it/s]
 19%|█▉        | 2536/13317 [26:08<1:50:10,  1.63it/s]
                                                      
{'loss': 0.2591, 'grad_norm': 0.6242323517799377, 'learning_rate': 9.753050093249632e-06, 'epoch': 0.57}

 19%|█▉        | 2536/13317 [26:08<1:50:10,  1.63it/s]
 19%|█▉        | 2537/13317 [26:09<1:50:12,  1.63it/s]
                                                      
{'loss': 0.3279, 'grad_norm': 0.6131358742713928, 'learning_rate': 9.752643124762491e-06, 'epoch': 0.57}

 19%|█▉        | 2537/13317 [26:09<1:50:12,  1.63it/s]
 19%|█▉        | 2538/13317 [26:09<1:50:12,  1.63it/s]
                                                      
{'loss': 0.2728, 'grad_norm': 0.5650448203086853, 'learning_rate': 9.752235829718446e-06, 'epoch': 0.57}

 19%|█▉        | 2538/13317 [26:09<1:50:12,  1.63it/s]
 19%|█▉        | 2539/13317 [26:10<1:50:08,  1.63it/s]
                                                      
{'loss': 0.2621, 'grad_norm': 0.6125362515449524, 'learning_rate': 9.751828208145482e-06, 'epoch': 0.57}

 19%|█▉        | 2539/13317 [26:10<1:50:08,  1.63it/s]
 19%|█▉        | 2540/13317 [26:11<1:50:08,  1.63it/s]
                                                      
{'loss': 0.224, 'grad_norm': 0.5447441935539246, 'learning_rate': 9.751420260071606e-06, 'epoch': 0.57}

 19%|█▉        | 2540/13317 [26:11<1:50:08,  1.63it/s]
 19%|█▉        | 2541/13317 [26:11<1:50:06,  1.63it/s]
                                                      
{'loss': 0.1839, 'grad_norm': 0.4494311213493347, 'learning_rate': 9.75101198552485e-06, 'epoch': 0.57}

 19%|█▉        | 2541/13317 [26:11<1:50:06,  1.63it/s]
 19%|█▉        | 2542/13317 [26:12<1:50:06,  1.63it/s]
                                                      
{'loss': 0.2471, 'grad_norm': 0.515454113483429, 'learning_rate': 9.750603384533267e-06, 'epoch': 0.57}

 19%|█▉        | 2542/13317 [26:12<1:50:06,  1.63it/s]
 19%|█▉        | 2543/13317 [26:12<1:50:05,  1.63it/s]
                                                      
{'loss': 0.3431, 'grad_norm': 0.5571158528327942, 'learning_rate': 9.75019445712493e-06, 'epoch': 0.57}

 19%|█▉        | 2543/13317 [26:13<1:50:05,  1.63it/s]
 19%|█▉        | 2544/13317 [26:13<1:50:05,  1.63it/s]
                                                      
{'loss': 0.2467, 'grad_norm': 0.5438475608825684, 'learning_rate': 9.74978520332794e-06, 'epoch': 0.57}

 19%|█▉        | 2544/13317 [26:13<1:50:05,  1.63it/s]
 19%|█▉        | 2545/13317 [26:14<1:50:03,  1.63it/s]
                                                      
{'loss': 0.3292, 'grad_norm': 0.6658268570899963, 'learning_rate': 9.749375623170412e-06, 'epoch': 0.57}

 19%|█▉        | 2545/13317 [26:14<1:50:03,  1.63it/s]
 19%|█▉        | 2546/13317 [26:14<1:50:03,  1.63it/s]
                                                      
{'loss': 0.3131, 'grad_norm': 0.6213856339454651, 'learning_rate': 9.748965716680493e-06, 'epoch': 0.57}

 19%|█▉        | 2546/13317 [26:14<1:50:03,  1.63it/s]
 19%|█▉        | 2547/13317 [26:15<1:50:05,  1.63it/s]
                                                      
{'loss': 0.2425, 'grad_norm': 0.5672732591629028, 'learning_rate': 9.748555483886347e-06, 'epoch': 0.57}

 19%|█▉        | 2547/13317 [26:15<1:50:05,  1.63it/s]
 19%|█▉        | 2548/13317 [26:15<1:50:05,  1.63it/s]
                                                      
{'loss': 0.3026, 'grad_norm': 0.5654504895210266, 'learning_rate': 9.748144924816158e-06, 'epoch': 0.57}

 19%|█▉        | 2548/13317 [26:16<1:50:05,  1.63it/s]
 19%|█▉        | 2549/13317 [26:16<1:50:06,  1.63it/s]
                                                      
{'loss': 0.2839, 'grad_norm': 0.596011757850647, 'learning_rate': 9.747734039498142e-06, 'epoch': 0.57}

 19%|█▉        | 2549/13317 [26:16<1:50:06,  1.63it/s]
 19%|█▉        | 2550/13317 [26:17<1:50:06,  1.63it/s]
                                                      
{'loss': 0.2236, 'grad_norm': 0.573558509349823, 'learning_rate': 9.747322827960526e-06, 'epoch': 0.57}

 19%|█▉        | 2550/13317 [26:17<1:50:06,  1.63it/s]
 19%|█▉        | 2551/13317 [26:17<1:50:03,  1.63it/s]
                                                      
{'loss': 0.2564, 'grad_norm': 0.5118023157119751, 'learning_rate': 9.746911290231567e-06, 'epoch': 0.57}

 19%|█▉        | 2551/13317 [26:17<1:50:03,  1.63it/s]
 19%|█▉        | 2552/13317 [26:18<1:50:05,  1.63it/s]
                                                      
{'loss': 0.2275, 'grad_norm': 0.5026276111602783, 'learning_rate': 9.746499426339539e-06, 'epoch': 0.57}

 19%|█▉        | 2552/13317 [26:18<1:50:05,  1.63it/s]
 19%|█▉        | 2553/13317 [26:19<1:50:05,  1.63it/s]
                                                      
{'loss': 0.264, 'grad_norm': 0.5547671914100647, 'learning_rate': 9.746087236312747e-06, 'epoch': 0.58}

 19%|█▉        | 2553/13317 [26:19<1:50:05,  1.63it/s]
 19%|█▉        | 2554/13317 [26:19<1:50:03,  1.63it/s]
                                                      
{'loss': 0.2423, 'grad_norm': 0.5693719387054443, 'learning_rate': 9.745674720179507e-06, 'epoch': 0.58}

 19%|█▉        | 2554/13317 [26:19<1:50:03,  1.63it/s]
 19%|█▉        | 2555/13317 [26:20<1:50:01,  1.63it/s]
                                                      
{'loss': 0.3247, 'grad_norm': 0.5673713684082031, 'learning_rate': 9.745261877968166e-06, 'epoch': 0.58}

 19%|█▉        | 2555/13317 [26:20<1:50:01,  1.63it/s]
 19%|█▉        | 2556/13317 [26:20<1:50:01,  1.63it/s]
                                                      
{'loss': 0.262, 'grad_norm': 0.5392853617668152, 'learning_rate': 9.74484870970709e-06, 'epoch': 0.58}

 19%|█▉        | 2556/13317 [26:21<1:50:01,  1.63it/s]
 19%|█▉        | 2557/13317 [26:21<1:49:59,  1.63it/s]
                                                      
{'loss': 0.2444, 'grad_norm': 0.611835777759552, 'learning_rate': 9.74443521542467e-06, 'epoch': 0.58}

 19%|█▉        | 2557/13317 [26:21<1:49:59,  1.63it/s]
 19%|█▉        | 2558/13317 [26:22<1:49:58,  1.63it/s]
                                                      
{'loss': 0.2129, 'grad_norm': 0.5026295185089111, 'learning_rate': 9.744021395149316e-06, 'epoch': 0.58}

 19%|█▉        | 2558/13317 [26:22<1:49:58,  1.63it/s]
 19%|█▉        | 2559/13317 [26:22<1:49:57,  1.63it/s]
                                                      
{'loss': 0.2252, 'grad_norm': 0.5071903467178345, 'learning_rate': 9.743607248909461e-06, 'epoch': 0.58}

 19%|█▉        | 2559/13317 [26:22<1:49:57,  1.63it/s]
 19%|█▉        | 2560/13317 [26:23<1:49:55,  1.63it/s]
                                                      
{'loss': 0.2411, 'grad_norm': 0.5450606346130371, 'learning_rate': 9.743192776733561e-06, 'epoch': 0.58}

 19%|█▉        | 2560/13317 [26:23<1:49:55,  1.63it/s]
 19%|█▉        | 2561/13317 [26:23<1:49:57,  1.63it/s]
                                                      
{'loss': 0.2773, 'grad_norm': 0.5809766054153442, 'learning_rate': 9.742777978650095e-06, 'epoch': 0.58}

 19%|█▉        | 2561/13317 [26:24<1:49:57,  1.63it/s]
 19%|█▉        | 2562/13317 [26:24<1:49:55,  1.63it/s]
                                                      
{'loss': 0.2334, 'grad_norm': 0.5144176483154297, 'learning_rate': 9.742362854687567e-06, 'epoch': 0.58}

 19%|█▉        | 2562/13317 [26:24<1:49:55,  1.63it/s]
 19%|█▉        | 2563/13317 [26:25<1:49:54,  1.63it/s]
                                                      
{'loss': 0.3419, 'grad_norm': 0.6435290575027466, 'learning_rate': 9.741947404874497e-06, 'epoch': 0.58}

 19%|█▉        | 2563/13317 [26:25<1:49:54,  1.63it/s]
 19%|█▉        | 2564/13317 [26:25<1:49:52,  1.63it/s]
                                                      
{'loss': 0.2591, 'grad_norm': 0.5452629327774048, 'learning_rate': 9.741531629239431e-06, 'epoch': 0.58}

 19%|█▉        | 2564/13317 [26:25<1:49:52,  1.63it/s]
 19%|█▉        | 2565/13317 [26:26<1:49:54,  1.63it/s]
                                                      
{'loss': 0.2253, 'grad_norm': 0.4903252422809601, 'learning_rate': 9.741115527810937e-06, 'epoch': 0.58}

 19%|█▉        | 2565/13317 [26:26<1:49:54,  1.63it/s]
 19%|█▉        | 2566/13317 [26:27<1:49:51,  1.63it/s]
                                                      
{'loss': 0.277, 'grad_norm': 0.6220775246620178, 'learning_rate': 9.74069910061761e-06, 'epoch': 0.58}

 19%|█▉        | 2566/13317 [26:27<1:49:51,  1.63it/s]
 19%|█▉        | 2567/13317 [26:27<1:49:46,  1.63it/s]
                                                      
{'loss': 0.2636, 'grad_norm': 0.5264109969139099, 'learning_rate': 9.740282347688056e-06, 'epoch': 0.58}

 19%|█▉        | 2567/13317 [26:27<1:49:46,  1.63it/s]
 19%|█▉        | 2568/13317 [26:28<1:49:48,  1.63it/s]
                                                      
{'loss': 0.3513, 'grad_norm': 0.7893503308296204, 'learning_rate': 9.739865269050914e-06, 'epoch': 0.58}

 19%|█▉        | 2568/13317 [26:28<1:49:48,  1.63it/s]
 19%|█▉        | 2569/13317 [26:28<1:49:48,  1.63it/s]
                                                      
{'loss': 0.2627, 'grad_norm': 0.5368137955665588, 'learning_rate': 9.739447864734844e-06, 'epoch': 0.58}

 19%|█▉        | 2569/13317 [26:28<1:49:48,  1.63it/s]
 19%|█▉        | 2570/13317 [26:29<1:49:50,  1.63it/s]
                                                      
{'loss': 0.2756, 'grad_norm': 0.6551600694656372, 'learning_rate': 9.739030134768522e-06, 'epoch': 0.58}

 19%|█▉        | 2570/13317 [26:29<1:49:50,  1.63it/s]
 19%|█▉        | 2571/13317 [26:30<1:49:49,  1.63it/s]
                                                      
{'loss': 0.2227, 'grad_norm': 0.6163540482521057, 'learning_rate': 9.738612079180651e-06, 'epoch': 0.58}

 19%|█▉        | 2571/13317 [26:30<1:49:49,  1.63it/s]
 19%|█▉        | 2572/13317 [26:30<1:49:47,  1.63it/s]
                                                      
{'loss': 0.2565, 'grad_norm': 0.5437400937080383, 'learning_rate': 9.738193697999959e-06, 'epoch': 0.58}

 19%|█▉        | 2572/13317 [26:30<1:49:47,  1.63it/s]
 19%|█▉        | 2573/13317 [26:31<1:49:45,  1.63it/s]
                                                      
{'loss': 0.2688, 'grad_norm': 0.5695928335189819, 'learning_rate': 9.737774991255191e-06, 'epoch': 0.58}

 19%|█▉        | 2573/13317 [26:31<1:49:45,  1.63it/s]
 19%|█▉        | 2574/13317 [26:31<1:49:48,  1.63it/s]
                                                      
{'loss': 0.2689, 'grad_norm': 0.6159367561340332, 'learning_rate': 9.737355958975115e-06, 'epoch': 0.58}

 19%|█▉        | 2574/13317 [26:32<1:49:48,  1.63it/s]
 19%|█▉        | 2575/13317 [26:32<1:49:51,  1.63it/s]
                                                      
{'loss': 0.2535, 'grad_norm': 0.6172235012054443, 'learning_rate': 9.736936601188528e-06, 'epoch': 0.58}

 19%|█▉        | 2575/13317 [26:32<1:49:51,  1.63it/s]
 19%|█▉        | 2576/13317 [26:33<1:49:52,  1.63it/s]
                                                      
{'loss': 0.3304, 'grad_norm': 0.5496262311935425, 'learning_rate': 9.736516917924238e-06, 'epoch': 0.58}

 19%|█▉        | 2576/13317 [26:33<1:49:52,  1.63it/s]
 19%|█▉        | 2577/13317 [26:33<1:49:58,  1.63it/s]
                                                      
{'loss': 0.1999, 'grad_norm': 0.4760874807834625, 'learning_rate': 9.736096909211085e-06, 'epoch': 0.58}

 19%|█▉        | 2577/13317 [26:33<1:49:58,  1.63it/s]
 19%|█▉        | 2578/13317 [26:34<1:49:55,  1.63it/s]
                                                      
{'loss': 0.3269, 'grad_norm': 0.544866144657135, 'learning_rate': 9.735676575077929e-06, 'epoch': 0.58}

 19%|█▉        | 2578/13317 [26:34<1:49:55,  1.63it/s]
 19%|█▉        | 2579/13317 [26:34<1:49:58,  1.63it/s]
                                                      
{'loss': 0.2647, 'grad_norm': 0.5463356971740723, 'learning_rate': 9.735255915553647e-06, 'epoch': 0.58}

 19%|█▉        | 2579/13317 [26:35<1:49:58,  1.63it/s]
 19%|█▉        | 2580/13317 [26:35<1:49:52,  1.63it/s]
                                                      
{'loss': 0.3203, 'grad_norm': 0.6042847633361816, 'learning_rate': 9.73483493066715e-06, 'epoch': 0.58}

 19%|█▉        | 2580/13317 [26:35<1:49:52,  1.63it/s]
 19%|█▉        | 2581/13317 [26:36<1:49:50,  1.63it/s]
                                                      
{'loss': 0.2084, 'grad_norm': 0.4959774613380432, 'learning_rate': 9.734413620447358e-06, 'epoch': 0.58}

 19%|█▉        | 2581/13317 [26:36<1:49:50,  1.63it/s]
 19%|█▉        | 2582/13317 [26:36<1:49:47,  1.63it/s]
                                                      
{'loss': 0.2116, 'grad_norm': 0.558099627494812, 'learning_rate': 9.733991984923222e-06, 'epoch': 0.58}

 19%|█▉        | 2582/13317 [26:36<1:49:47,  1.63it/s]
 19%|█▉        | 2583/13317 [26:37<1:49:46,  1.63it/s]
                                                      
{'loss': 0.2867, 'grad_norm': 0.5576102137565613, 'learning_rate': 9.73357002412371e-06, 'epoch': 0.58}

 19%|█▉        | 2583/13317 [26:37<1:49:46,  1.63it/s]
 19%|█▉        | 2584/13317 [26:38<1:49:45,  1.63it/s]
                                                      
{'loss': 0.2758, 'grad_norm': 0.46005192399024963, 'learning_rate': 9.733147738077822e-06, 'epoch': 0.58}

 19%|█▉        | 2584/13317 [26:38<1:49:45,  1.63it/s]
 19%|█▉        | 2585/13317 [26:38<1:49:48,  1.63it/s]
                                                      
{'loss': 0.2381, 'grad_norm': 0.5505496263504028, 'learning_rate': 9.732725126814567e-06, 'epoch': 0.58}

 19%|█▉        | 2585/13317 [26:38<1:49:48,  1.63it/s]
 19%|█▉        | 2586/13317 [26:39<1:49:50,  1.63it/s]
                                                      
{'loss': 0.234, 'grad_norm': 0.5354155898094177, 'learning_rate': 9.732302190362983e-06, 'epoch': 0.58}

 19%|█▉        | 2586/13317 [26:39<1:49:50,  1.63it/s]
 19%|█▉        | 2587/13317 [26:39<1:49:45,  1.63it/s]
                                                      
{'loss': 0.197, 'grad_norm': 0.47493410110473633, 'learning_rate': 9.731878928752134e-06, 'epoch': 0.58}

 19%|█▉        | 2587/13317 [26:40<1:49:45,  1.63it/s]
 19%|█▉        | 2588/13317 [26:40<1:49:45,  1.63it/s]
                                                      
{'loss': 0.3182, 'grad_norm': 0.6227079033851624, 'learning_rate': 9.7314553420111e-06, 'epoch': 0.58}

 19%|█▉        | 2588/13317 [26:40<1:49:45,  1.63it/s]
 19%|█▉        | 2589/13317 [26:41<1:49:42,  1.63it/s]
                                                      
{'loss': 0.2847, 'grad_norm': 0.6121814846992493, 'learning_rate': 9.731031430168986e-06, 'epoch': 0.58}

 19%|█▉        | 2589/13317 [26:41<1:49:42,  1.63it/s]
 19%|█▉        | 2590/13317 [26:41<1:49:41,  1.63it/s]
                                                      
{'loss': 0.2127, 'grad_norm': 0.5123658180236816, 'learning_rate': 9.730607193254921e-06, 'epoch': 0.58}

 19%|█▉        | 2590/13317 [26:41<1:49:41,  1.63it/s]
 19%|█▉        | 2591/13317 [26:42<1:49:36,  1.63it/s]
                                                      
{'loss': 0.256, 'grad_norm': 0.5030328035354614, 'learning_rate': 9.730182631298053e-06, 'epoch': 0.58}

 19%|█▉        | 2591/13317 [26:42<1:49:36,  1.63it/s]
 19%|█▉        | 2592/13317 [26:42<1:49:39,  1.63it/s]
                                                      
{'loss': 0.2675, 'grad_norm': 0.5677589774131775, 'learning_rate': 9.729757744327555e-06, 'epoch': 0.58}

 19%|█▉        | 2592/13317 [26:43<1:49:39,  1.63it/s]
 19%|█▉        | 2593/13317 [26:43<1:49:34,  1.63it/s]
                                                      
{'loss': 0.2939, 'grad_norm': 0.6011199951171875, 'learning_rate': 9.729332532372619e-06, 'epoch': 0.58}

 19%|█▉        | 2593/13317 [26:43<1:49:34,  1.63it/s]
 19%|█▉        | 2594/13317 [26:44<1:49:33,  1.63it/s]
                                                      
{'loss': 0.2674, 'grad_norm': 0.587906539440155, 'learning_rate': 9.728906995462463e-06, 'epoch': 0.58}

 19%|█▉        | 2594/13317 [26:44<1:49:33,  1.63it/s]
 19%|█▉        | 2595/13317 [26:44<1:49:34,  1.63it/s]
                                                      
{'loss': 0.2476, 'grad_norm': 0.5968888998031616, 'learning_rate': 9.728481133626329e-06, 'epoch': 0.58}

 19%|█▉        | 2595/13317 [26:44<1:49:34,  1.63it/s]
 19%|█▉        | 2596/13317 [26:45<1:49:34,  1.63it/s]
                                                      
{'loss': 0.311, 'grad_norm': 0.7162779569625854, 'learning_rate': 9.728054946893471e-06, 'epoch': 0.58}

 19%|█▉        | 2596/13317 [26:45<1:49:34,  1.63it/s]
 20%|█▉        | 2597/13317 [26:46<1:49:33,  1.63it/s]
                                                      
{'loss': 0.2347, 'grad_norm': 0.5739919543266296, 'learning_rate': 9.727628435293179e-06, 'epoch': 0.59}

 20%|█▉        | 2597/13317 [26:46<1:49:33,  1.63it/s]
 20%|█▉        | 2598/13317 [26:46<1:49:31,  1.63it/s]
                                                      
{'loss': 0.2449, 'grad_norm': 0.5592507719993591, 'learning_rate': 9.727201598854757e-06, 'epoch': 0.59}

 20%|█▉        | 2598/13317 [26:46<1:49:31,  1.63it/s]
 20%|█▉        | 2599/13317 [26:47<1:49:33,  1.63it/s]
                                                      
{'loss': 0.2869, 'grad_norm': 0.5718072652816772, 'learning_rate': 9.726774437607531e-06, 'epoch': 0.59}

 20%|█▉        | 2599/13317 [26:47<1:49:33,  1.63it/s]
 20%|█▉        | 2600/13317 [26:47<1:49:35,  1.63it/s]
                                                      
{'loss': 0.3554, 'grad_norm': 0.6228532195091248, 'learning_rate': 9.726346951580855e-06, 'epoch': 0.59}

 20%|█▉        | 2600/13317 [26:48<1:49:35,  1.63it/s]
 20%|█▉        | 2601/13317 [26:48<1:49:29,  1.63it/s]
                                                      
{'loss': 0.2032, 'grad_norm': 0.5355802178382874, 'learning_rate': 9.7259191408041e-06, 'epoch': 0.59}

 20%|█▉        | 2601/13317 [26:48<1:49:29,  1.63it/s]
 20%|█▉        | 2602/13317 [26:49<1:49:30,  1.63it/s]
                                                      
{'loss': 0.2134, 'grad_norm': 0.552807092666626, 'learning_rate': 9.725491005306659e-06, 'epoch': 0.59}

 20%|█▉        | 2602/13317 [26:49<1:49:30,  1.63it/s]
 20%|█▉        | 2603/13317 [26:49<1:49:28,  1.63it/s]
                                                      
{'loss': 0.3025, 'grad_norm': 0.6223698258399963, 'learning_rate': 9.725062545117954e-06, 'epoch': 0.59}

 20%|█▉        | 2603/13317 [26:49<1:49:28,  1.63it/s]
 20%|█▉        | 2604/13317 [26:50<1:49:27,  1.63it/s]
                                                      
{'loss': 0.2745, 'grad_norm': 0.5811429619789124, 'learning_rate': 9.72463376026742e-06, 'epoch': 0.59}

 20%|█▉        | 2604/13317 [26:50<1:49:27,  1.63it/s]
 20%|█▉        | 2605/13317 [26:50<1:49:31,  1.63it/s]
                                                      
{'loss': 0.3158, 'grad_norm': 0.5977892279624939, 'learning_rate': 9.724204650784524e-06, 'epoch': 0.59}

 20%|█▉        | 2605/13317 [26:51<1:49:31,  1.63it/s]
 20%|█▉        | 2606/13317 [26:51<1:49:30,  1.63it/s]
                                                      
{'loss': 0.2617, 'grad_norm': 0.5173847079277039, 'learning_rate': 9.723775216698746e-06, 'epoch': 0.59}

 20%|█▉        | 2606/13317 [26:51<1:49:30,  1.63it/s]
 20%|█▉        | 2607/13317 [26:52<1:49:59,  1.62it/s]
                                                      
{'loss': 0.2776, 'grad_norm': 0.5474826097488403, 'learning_rate': 9.723345458039595e-06, 'epoch': 0.59}

 20%|█▉        | 2607/13317 [26:52<1:49:59,  1.62it/s]
 20%|█▉        | 2608/13317 [26:52<1:49:51,  1.62it/s]
                                                      
{'loss': 0.3259, 'grad_norm': 0.5995017886161804, 'learning_rate': 9.7229153748366e-06, 'epoch': 0.59}

 20%|█▉        | 2608/13317 [26:52<1:49:51,  1.62it/s]
 20%|█▉        | 2609/13317 [26:53<1:49:40,  1.63it/s]
                                                      
{'loss': 0.2605, 'grad_norm': 0.5393586158752441, 'learning_rate': 9.72248496711931e-06, 'epoch': 0.59}

 20%|█▉        | 2609/13317 [26:53<1:49:40,  1.63it/s]
 20%|█▉        | 2610/13317 [26:54<1:49:35,  1.63it/s]
                                                      
{'loss': 0.2783, 'grad_norm': 0.5501528382301331, 'learning_rate': 9.722054234917302e-06, 'epoch': 0.59}

 20%|█▉        | 2610/13317 [26:54<1:49:35,  1.63it/s]
 20%|█▉        | 2611/13317 [26:54<1:49:43,  1.63it/s]
                                                      
{'loss': 0.2824, 'grad_norm': 0.5530862808227539, 'learning_rate': 9.72162317826017e-06, 'epoch': 0.59}

 20%|█▉        | 2611/13317 [26:54<1:49:43,  1.63it/s]
 20%|█▉        | 2612/13317 [26:55<1:49:45,  1.63it/s]
                                                      
{'loss': 0.2311, 'grad_norm': 0.6055856943130493, 'learning_rate': 9.721191797177531e-06, 'epoch': 0.59}

 20%|█▉        | 2612/13317 [26:55<1:49:45,  1.63it/s]
 20%|█▉        | 2613/13317 [26:55<1:49:38,  1.63it/s]
                                                      
{'loss': 0.2835, 'grad_norm': 0.5678766369819641, 'learning_rate': 9.720760091699029e-06, 'epoch': 0.59}

 20%|█▉        | 2613/13317 [26:55<1:49:38,  1.63it/s]
 20%|█▉        | 2614/13317 [26:56<1:49:36,  1.63it/s]
                                                      
{'loss': 0.1836, 'grad_norm': 0.4747050404548645, 'learning_rate': 9.720328061854322e-06, 'epoch': 0.59}

 20%|█▉        | 2614/13317 [26:56<1:49:36,  1.63it/s]
 20%|█▉        | 2615/13317 [26:57<1:49:32,  1.63it/s]
                                                      
{'loss': 0.287, 'grad_norm': 0.6062408089637756, 'learning_rate': 9.7198957076731e-06, 'epoch': 0.59}

 20%|█▉        | 2615/13317 [26:57<1:49:32,  1.63it/s]
 20%|█▉        | 2616/13317 [26:57<1:49:59,  1.62it/s]
                                                      
{'loss': 0.3, 'grad_norm': 0.7021021842956543, 'learning_rate': 9.719463029185067e-06, 'epoch': 0.59}

 20%|█▉        | 2616/13317 [26:57<1:49:59,  1.62it/s]
 20%|█▉        | 2617/13317 [26:58<1:49:52,  1.62it/s]
                                                      
{'loss': 0.3041, 'grad_norm': 0.6157855987548828, 'learning_rate': 9.719030026419954e-06, 'epoch': 0.59}

 20%|█▉        | 2617/13317 [26:58<1:49:52,  1.62it/s]
 20%|█▉        | 2618/13317 [26:58<1:49:46,  1.62it/s]
                                                      
{'loss': 0.2166, 'grad_norm': 0.5092508792877197, 'learning_rate': 9.71859669940751e-06, 'epoch': 0.59}

 20%|█▉        | 2618/13317 [26:59<1:49:46,  1.62it/s]
 20%|█▉        | 2619/13317 [26:59<1:49:38,  1.63it/s]
                                                      
{'loss': 0.3111, 'grad_norm': 0.6640470623970032, 'learning_rate': 9.718163048177513e-06, 'epoch': 0.59}

 20%|█▉        | 2619/13317 [26:59<1:49:38,  1.63it/s]
 20%|█▉        | 2620/13317 [27:00<1:49:37,  1.63it/s]
                                                      
{'loss': 0.2665, 'grad_norm': 0.6384447813034058, 'learning_rate': 9.717729072759757e-06, 'epoch': 0.59}

 20%|█▉        | 2620/13317 [27:00<1:49:37,  1.63it/s]
 20%|█▉        | 2621/13317 [27:00<1:49:34,  1.63it/s]
                                                      
{'loss': 0.2217, 'grad_norm': 0.4974691867828369, 'learning_rate': 9.717294773184062e-06, 'epoch': 0.59}

 20%|█▉        | 2621/13317 [27:00<1:49:34,  1.63it/s]
 20%|█▉        | 2622/13317 [27:01<1:49:38,  1.63it/s]
                                                      
{'loss': 0.314, 'grad_norm': 0.7269319891929626, 'learning_rate': 9.716860149480269e-06, 'epoch': 0.59}

 20%|█▉        | 2622/13317 [27:01<1:49:38,  1.63it/s]
 20%|█▉        | 2623/13317 [27:02<1:49:33,  1.63it/s]
                                                      
{'loss': 0.3098, 'grad_norm': 0.6539899706840515, 'learning_rate': 9.71642520167824e-06, 'epoch': 0.59}

 20%|█▉        | 2623/13317 [27:02<1:49:33,  1.63it/s]
 20%|█▉        | 2624/13317 [27:02<1:49:35,  1.63it/s]
                                                      
{'loss': 0.2568, 'grad_norm': 0.5466057062149048, 'learning_rate': 9.715989929807863e-06, 'epoch': 0.59}

 20%|█▉        | 2624/13317 [27:02<1:49:35,  1.63it/s]
 20%|█▉        | 2625/13317 [27:03<1:49:29,  1.63it/s]
                                                      
{'loss': 0.2351, 'grad_norm': 0.6079336404800415, 'learning_rate': 9.715554333899043e-06, 'epoch': 0.59}

 20%|█▉        | 2625/13317 [27:03<1:49:29,  1.63it/s]
 20%|█▉        | 2626/13317 [27:03<1:49:27,  1.63it/s]
                                                      
{'loss': 0.2304, 'grad_norm': 0.5705031156539917, 'learning_rate': 9.71511841398171e-06, 'epoch': 0.59}

 20%|█▉        | 2626/13317 [27:03<1:49:27,  1.63it/s]
 20%|█▉        | 2627/13317 [27:04<1:49:27,  1.63it/s]
                                                      
{'loss': 0.2314, 'grad_norm': 0.4497797191143036, 'learning_rate': 9.714682170085821e-06, 'epoch': 0.59}

 20%|█▉        | 2627/13317 [27:04<1:49:27,  1.63it/s]
 20%|█▉        | 2628/13317 [27:05<1:49:26,  1.63it/s]
                                                      
{'loss': 0.2817, 'grad_norm': 0.5664573907852173, 'learning_rate': 9.714245602241343e-06, 'epoch': 0.59}

 20%|█▉        | 2628/13317 [27:05<1:49:26,  1.63it/s]
 20%|█▉        | 2629/13317 [27:05<1:49:32,  1.63it/s]
                                                      
{'loss': 0.2449, 'grad_norm': 0.5475367307662964, 'learning_rate': 9.71380871047828e-06, 'epoch': 0.59}

 20%|█▉        | 2629/13317 [27:05<1:49:32,  1.63it/s]
 20%|█▉        | 2630/13317 [27:06<1:49:27,  1.63it/s]
                                                      
{'loss': 0.2099, 'grad_norm': 0.47525835037231445, 'learning_rate': 9.713371494826646e-06, 'epoch': 0.59}

 20%|█▉        | 2630/13317 [27:06<1:49:27,  1.63it/s]
 20%|█▉        | 2631/13317 [27:06<1:49:22,  1.63it/s]
                                                      
{'loss': 0.1926, 'grad_norm': 0.4653555154800415, 'learning_rate': 9.712933955316486e-06, 'epoch': 0.59}

 20%|█▉        | 2631/13317 [27:07<1:49:22,  1.63it/s]
 20%|█▉        | 2632/13317 [27:07<1:49:17,  1.63it/s]
                                                      
{'loss': 0.2234, 'grad_norm': 0.4964256286621094, 'learning_rate': 9.71249609197786e-06, 'epoch': 0.59}

 20%|█▉        | 2632/13317 [27:07<1:49:17,  1.63it/s]
 20%|█▉        | 2633/13317 [27:08<1:49:14,  1.63it/s]
                                                      
{'loss': 0.2618, 'grad_norm': 0.5600623488426208, 'learning_rate': 9.712057904840858e-06, 'epoch': 0.59}

 20%|█▉        | 2633/13317 [27:08<1:49:14,  1.63it/s]
 20%|█▉        | 2634/13317 [27:08<1:49:23,  1.63it/s]
                                                      
{'loss': 0.2864, 'grad_norm': 0.6466591954231262, 'learning_rate': 9.711619393935583e-06, 'epoch': 0.59}

 20%|█▉        | 2634/13317 [27:08<1:49:23,  1.63it/s]
 20%|█▉        | 2635/13317 [27:09<1:49:16,  1.63it/s]
                                                      
{'loss': 0.282, 'grad_norm': 0.6835272908210754, 'learning_rate': 9.71118055929217e-06, 'epoch': 0.59}

 20%|█▉        | 2635/13317 [27:09<1:49:16,  1.63it/s]
 20%|█▉        | 2636/13317 [27:09<1:49:13,  1.63it/s]
                                                      
{'loss': 0.4044, 'grad_norm': 0.6476162672042847, 'learning_rate': 9.71074140094077e-06, 'epoch': 0.59}

 20%|█▉        | 2636/13317 [27:10<1:49:13,  1.63it/s]
 20%|█▉        | 2637/13317 [27:10<1:49:11,  1.63it/s]
                                                      
{'loss': 0.2932, 'grad_norm': 0.6164636611938477, 'learning_rate': 9.710301918911557e-06, 'epoch': 0.59}

 20%|█▉        | 2637/13317 [27:10<1:49:11,  1.63it/s]
 20%|█▉        | 2638/13317 [27:11<1:49:11,  1.63it/s]
                                                      
{'loss': 0.2414, 'grad_norm': 0.5036197900772095, 'learning_rate': 9.70986211323473e-06, 'epoch': 0.59}

 20%|█▉        | 2638/13317 [27:11<1:49:11,  1.63it/s]
 20%|█▉        | 2639/13317 [27:11<1:49:12,  1.63it/s]
                                                      
{'loss': 0.2815, 'grad_norm': 0.5663371682167053, 'learning_rate': 9.709421983940505e-06, 'epoch': 0.59}

 20%|█▉        | 2639/13317 [27:11<1:49:12,  1.63it/s]
 20%|█▉        | 2640/13317 [27:12<1:49:11,  1.63it/s]
                                                      
{'loss': 0.2364, 'grad_norm': 0.5630803108215332, 'learning_rate': 9.708981531059127e-06, 'epoch': 0.59}

 20%|█▉        | 2640/13317 [27:12<1:49:11,  1.63it/s]
 20%|█▉        | 2641/13317 [27:13<1:49:14,  1.63it/s]
                                                      
{'loss': 0.271, 'grad_norm': 0.5944691896438599, 'learning_rate': 9.708540754620856e-06, 'epoch': 0.59}

 20%|█▉        | 2641/13317 [27:13<1:49:14,  1.63it/s]
 20%|█▉        | 2642/13317 [27:13<1:49:09,  1.63it/s]
                                                      
{'loss': 0.3851, 'grad_norm': 0.6215130686759949, 'learning_rate': 9.708099654655983e-06, 'epoch': 0.6}

 20%|█▉        | 2642/13317 [27:13<1:49:09,  1.63it/s]
 20%|█▉        | 2643/13317 [27:14<1:49:07,  1.63it/s]
                                                      
{'loss': 0.2311, 'grad_norm': 0.5460854172706604, 'learning_rate': 9.707658231194813e-06, 'epoch': 0.6}

 20%|█▉        | 2643/13317 [27:14<1:49:07,  1.63it/s]
 20%|█▉        | 2644/13317 [27:14<1:49:11,  1.63it/s]
                                                      
{'loss': 0.2951, 'grad_norm': 0.5032547116279602, 'learning_rate': 9.707216484267677e-06, 'epoch': 0.6}

 20%|█▉        | 2644/13317 [27:15<1:49:11,  1.63it/s]
 20%|█▉        | 2645/13317 [27:15<1:49:09,  1.63it/s]
                                                      
{'loss': 0.3247, 'grad_norm': 0.5580963492393494, 'learning_rate': 9.706774413904927e-06, 'epoch': 0.6}

 20%|█▉        | 2645/13317 [27:15<1:49:09,  1.63it/s]
 20%|█▉        | 2646/13317 [27:16<1:49:06,  1.63it/s]
                                                      
{'loss': 0.1945, 'grad_norm': 0.45621517300605774, 'learning_rate': 9.70633202013694e-06, 'epoch': 0.6}

 20%|█▉        | 2646/13317 [27:16<1:49:06,  1.63it/s]
 20%|█▉        | 2647/13317 [27:16<1:49:07,  1.63it/s]
                                                      
{'loss': 0.2746, 'grad_norm': 0.5562261939048767, 'learning_rate': 9.705889302994108e-06, 'epoch': 0.6}

 20%|█▉        | 2647/13317 [27:16<1:49:07,  1.63it/s]
 20%|█▉        | 2648/13317 [27:17<1:49:07,  1.63it/s]
                                                      
{'loss': 0.2333, 'grad_norm': 0.5356273651123047, 'learning_rate': 9.705446262506858e-06, 'epoch': 0.6}

 20%|█▉        | 2648/13317 [27:17<1:49:07,  1.63it/s]
 20%|█▉        | 2649/13317 [27:17<1:49:07,  1.63it/s]
                                                      
{'loss': 0.2229, 'grad_norm': 0.5262927412986755, 'learning_rate': 9.705002898705626e-06, 'epoch': 0.6}

 20%|█▉        | 2649/13317 [27:18<1:49:07,  1.63it/s]
 20%|█▉        | 2650/13317 [27:18<1:49:08,  1.63it/s]
                                                      
{'loss': 0.2916, 'grad_norm': 0.5550562739372253, 'learning_rate': 9.70455921162088e-06, 'epoch': 0.6}

 20%|█▉        | 2650/13317 [27:18<1:49:08,  1.63it/s]
 20%|█▉        | 2651/13317 [27:19<1:49:08,  1.63it/s]
                                                      
{'loss': 0.1714, 'grad_norm': 0.46707531809806824, 'learning_rate': 9.704115201283103e-06, 'epoch': 0.6}

 20%|█▉        | 2651/13317 [27:19<1:49:08,  1.63it/s]
 20%|█▉        | 2652/13317 [27:19<1:49:04,  1.63it/s]
                                                      
{'loss': 0.2333, 'grad_norm': 0.47364917397499084, 'learning_rate': 9.7036708677228e-06, 'epoch': 0.6}

 20%|█▉        | 2652/13317 [27:19<1:49:04,  1.63it/s]
 20%|█▉        | 2653/13317 [27:20<1:49:03,  1.63it/s]
                                                      
{'loss': 0.1775, 'grad_norm': 0.47584080696105957, 'learning_rate': 9.70322621097051e-06, 'epoch': 0.6}

 20%|█▉        | 2653/13317 [27:20<1:49:03,  1.63it/s]
 20%|█▉        | 2654/13317 [27:21<1:49:01,  1.63it/s]
                                                      
{'loss': 0.2523, 'grad_norm': 0.5120347738265991, 'learning_rate': 9.702781231056777e-06, 'epoch': 0.6}

 20%|█▉        | 2654/13317 [27:21<1:49:01,  1.63it/s]
 20%|█▉        | 2655/13317 [27:21<1:49:00,  1.63it/s]
                                                      
{'loss': 0.2354, 'grad_norm': 0.47187283635139465, 'learning_rate': 9.70233592801218e-06, 'epoch': 0.6}

 20%|█▉        | 2655/13317 [27:21<1:49:00,  1.63it/s]
 20%|█▉        | 2656/13317 [27:22<1:48:58,  1.63it/s]
                                                      
{'loss': 0.3053, 'grad_norm': 0.6074573397636414, 'learning_rate': 9.701890301867317e-06, 'epoch': 0.6}

 20%|█▉        | 2656/13317 [27:22<1:48:58,  1.63it/s]
 20%|█▉        | 2657/13317 [27:22<1:48:59,  1.63it/s]
                                                      
{'loss': 0.2936, 'grad_norm': 0.6738703846931458, 'learning_rate': 9.701444352652805e-06, 'epoch': 0.6}

 20%|█▉        | 2657/13317 [27:23<1:48:59,  1.63it/s]
 20%|█▉        | 2658/13317 [27:23<1:48:58,  1.63it/s]
                                                      
{'loss': 0.2102, 'grad_norm': 0.5359985828399658, 'learning_rate': 9.700998080399287e-06, 'epoch': 0.6}

 20%|█▉        | 2658/13317 [27:23<1:48:58,  1.63it/s]
 20%|█▉        | 2659/13317 [27:24<1:48:57,  1.63it/s]
                                                      
{'loss': 0.206, 'grad_norm': 0.5216466784477234, 'learning_rate': 9.700551485137425e-06, 'epoch': 0.6}

 20%|█▉        | 2659/13317 [27:24<1:48:57,  1.63it/s]
 20%|█▉        | 2660/13317 [27:24<1:48:57,  1.63it/s]
                                                      
{'loss': 0.241, 'grad_norm': 0.5155110955238342, 'learning_rate': 9.700104566897906e-06, 'epoch': 0.6}

 20%|█▉        | 2660/13317 [27:24<1:48:57,  1.63it/s]
 20%|█▉        | 2661/13317 [27:25<1:49:04,  1.63it/s]
                                                      
{'loss': 0.3641, 'grad_norm': 0.6389501690864563, 'learning_rate': 9.699657325711438e-06, 'epoch': 0.6}

 20%|█▉        | 2661/13317 [27:25<1:49:04,  1.63it/s]
 20%|█▉        | 2662/13317 [27:25<1:48:57,  1.63it/s]
                                                      
{'loss': 0.3394, 'grad_norm': 0.6121448278427124, 'learning_rate': 9.69920976160875e-06, 'epoch': 0.6}

 20%|█▉        | 2662/13317 [27:26<1:48:57,  1.63it/s]
 20%|█▉        | 2663/13317 [27:26<1:48:58,  1.63it/s]
                                                      
{'loss': 0.277, 'grad_norm': 0.6274598240852356, 'learning_rate': 9.698761874620596e-06, 'epoch': 0.6}

 20%|█▉        | 2663/13317 [27:26<1:48:58,  1.63it/s]
 20%|██        | 2664/13317 [27:27<1:48:55,  1.63it/s]
                                                      
{'loss': 0.2768, 'grad_norm': 0.6076623797416687, 'learning_rate': 9.698313664777751e-06, 'epoch': 0.6}

 20%|██        | 2664/13317 [27:27<1:48:55,  1.63it/s]
 20%|██        | 2665/13317 [27:27<1:48:53,  1.63it/s]
                                                      
{'loss': 0.2696, 'grad_norm': 0.5779857039451599, 'learning_rate': 9.69786513211101e-06, 'epoch': 0.6}

 20%|██        | 2665/13317 [27:27<1:48:53,  1.63it/s]
 20%|██        | 2666/13317 [27:28<1:48:52,  1.63it/s]
                                                      
{'loss': 0.2478, 'grad_norm': 0.5408627986907959, 'learning_rate': 9.697416276651193e-06, 'epoch': 0.6}

 20%|██        | 2666/13317 [27:28<1:48:52,  1.63it/s]
 20%|██        | 2667/13317 [27:29<1:48:52,  1.63it/s]
                                                      
{'loss': 0.2897, 'grad_norm': 0.6871560215950012, 'learning_rate': 9.69696709842914e-06, 'epoch': 0.6}

 20%|██        | 2667/13317 [27:29<1:48:52,  1.63it/s]
 20%|██        | 2668/13317 [27:29<1:48:50,  1.63it/s]
                                                      
{'loss': 0.1883, 'grad_norm': 0.4992251992225647, 'learning_rate': 9.696517597475714e-06, 'epoch': 0.6}

 20%|██        | 2668/13317 [27:29<1:48:50,  1.63it/s]
 20%|██        | 2669/13317 [27:30<1:48:50,  1.63it/s]
                                                      
{'loss': 0.2833, 'grad_norm': 0.600949227809906, 'learning_rate': 9.696067773821804e-06, 'epoch': 0.6}

 20%|██        | 2669/13317 [27:30<1:48:50,  1.63it/s]
 20%|██        | 2670/13317 [27:30<1:48:48,  1.63it/s]
                                                      
{'loss': 0.2334, 'grad_norm': 0.5412989854812622, 'learning_rate': 9.695617627498313e-06, 'epoch': 0.6}

 20%|██        | 2670/13317 [27:30<1:48:48,  1.63it/s]
 20%|██        | 2671/13317 [27:31<1:48:47,  1.63it/s]
                                                      
{'loss': 0.2636, 'grad_norm': 0.5272619724273682, 'learning_rate': 9.695167158536175e-06, 'epoch': 0.6}

 20%|██        | 2671/13317 [27:31<1:48:47,  1.63it/s]
 20%|██        | 2672/13317 [27:32<1:48:46,  1.63it/s]
                                                      
{'loss': 0.2995, 'grad_norm': 0.5214086174964905, 'learning_rate': 9.69471636696634e-06, 'epoch': 0.6}

 20%|██        | 2672/13317 [27:32<1:48:46,  1.63it/s]
 20%|██        | 2673/13317 [27:32<1:48:45,  1.63it/s]
                                                      
{'loss': 0.2653, 'grad_norm': 0.43811509013175964, 'learning_rate': 9.69426525281978e-06, 'epoch': 0.6}

 20%|██        | 2673/13317 [27:32<1:48:45,  1.63it/s]
 20%|██        | 2674/13317 [27:33<1:48:42,  1.63it/s]
                                                      
{'loss': 0.2305, 'grad_norm': 0.46280306577682495, 'learning_rate': 9.693813816127494e-06, 'epoch': 0.6}

 20%|██        | 2674/13317 [27:33<1:48:42,  1.63it/s]
 20%|██        | 2675/13317 [27:33<1:48:43,  1.63it/s]
                                                      
{'loss': 0.3895, 'grad_norm': 0.6367475390434265, 'learning_rate': 9.693362056920501e-06, 'epoch': 0.6}

 20%|██        | 2675/13317 [27:34<1:48:43,  1.63it/s]
 20%|██        | 2676/13317 [27:34<1:48:48,  1.63it/s]
                                                      
{'loss': 0.2584, 'grad_norm': 0.5819411873817444, 'learning_rate': 9.69290997522984e-06, 'epoch': 0.6}

 20%|██        | 2676/13317 [27:34<1:48:48,  1.63it/s]
 20%|██        | 2677/13317 [27:35<1:48:47,  1.63it/s]
                                                      
{'loss': 0.3072, 'grad_norm': 0.6106085777282715, 'learning_rate': 9.692457571086574e-06, 'epoch': 0.6}

 20%|██        | 2677/13317 [27:35<1:48:47,  1.63it/s]
 20%|██        | 2678/13317 [27:35<1:48:47,  1.63it/s]
                                                      
{'loss': 0.2989, 'grad_norm': 0.6173418760299683, 'learning_rate': 9.69200484452179e-06, 'epoch': 0.6}

 20%|██        | 2678/13317 [27:35<1:48:47,  1.63it/s]
 20%|██        | 2679/13317 [27:36<1:48:46,  1.63it/s]
                                                      
{'loss': 0.24, 'grad_norm': 0.550738513469696, 'learning_rate': 9.691551795566592e-06, 'epoch': 0.6}

 20%|██        | 2679/13317 [27:36<1:48:46,  1.63it/s]
 20%|██        | 2680/13317 [27:36<1:48:44,  1.63it/s]
                                                      
{'loss': 0.2406, 'grad_norm': 0.4988444745540619, 'learning_rate': 9.69109842425211e-06, 'epoch': 0.6}

 20%|██        | 2680/13317 [27:37<1:48:44,  1.63it/s]
 20%|██        | 2681/13317 [27:37<1:48:42,  1.63it/s]
                                                      
{'loss': 0.2765, 'grad_norm': 0.5213693380355835, 'learning_rate': 9.690644730609497e-06, 'epoch': 0.6}

 20%|██        | 2681/13317 [27:37<1:48:42,  1.63it/s]
 20%|██        | 2682/13317 [27:38<1:48:40,  1.63it/s]
                                                      
{'loss': 0.2846, 'grad_norm': 0.6870158910751343, 'learning_rate': 9.690190714669926e-06, 'epoch': 0.6}

 20%|██        | 2682/13317 [27:38<1:48:40,  1.63it/s]
 20%|██        | 2683/13317 [27:38<1:48:39,  1.63it/s]
                                                      
{'loss': 0.2247, 'grad_norm': 0.4786563217639923, 'learning_rate': 9.689736376464593e-06, 'epoch': 0.6}

 20%|██        | 2683/13317 [27:38<1:48:39,  1.63it/s]
 20%|██        | 2684/13317 [27:39<1:48:39,  1.63it/s]
                                                      
{'loss': 0.2982, 'grad_norm': 0.5710258483886719, 'learning_rate': 9.689281716024713e-06, 'epoch': 0.6}

 20%|██        | 2684/13317 [27:39<1:48:39,  1.63it/s]
 20%|██        | 2685/13317 [27:40<1:48:39,  1.63it/s]
                                                      
{'loss': 0.2099, 'grad_norm': 0.49270254373550415, 'learning_rate': 9.68882673338153e-06, 'epoch': 0.6}

 20%|██        | 2685/13317 [27:40<1:48:39,  1.63it/s]
 20%|██        | 2686/13317 [27:40<1:48:37,  1.63it/s]
                                                      
{'loss': 0.229, 'grad_norm': 0.5374448895454407, 'learning_rate': 9.688371428566304e-06, 'epoch': 0.61}

 20%|██        | 2686/13317 [27:40<1:48:37,  1.63it/s]
 20%|██        | 2687/13317 [27:41<1:48:40,  1.63it/s]
                                                      
{'loss': 0.1943, 'grad_norm': 0.4733146131038666, 'learning_rate': 9.68791580161032e-06, 'epoch': 0.61}

 20%|██        | 2687/13317 [27:41<1:48:40,  1.63it/s]
 20%|██        | 2688/13317 [27:41<1:48:33,  1.63it/s]
                                                      
{'loss': 0.302, 'grad_norm': 0.580886960029602, 'learning_rate': 9.687459852544883e-06, 'epoch': 0.61}

 20%|██        | 2688/13317 [27:42<1:48:33,  1.63it/s]
 20%|██        | 2689/13317 [27:42<1:48:38,  1.63it/s]
                                                      
{'loss': 0.2578, 'grad_norm': 0.5413209795951843, 'learning_rate': 9.687003581401322e-06, 'epoch': 0.61}

 20%|██        | 2689/13317 [27:42<1:48:38,  1.63it/s]
 20%|██        | 2690/13317 [27:43<1:48:39,  1.63it/s]
                                                      
{'loss': 0.3248, 'grad_norm': 0.6875874400138855, 'learning_rate': 9.68654698821099e-06, 'epoch': 0.61}

 20%|██        | 2690/13317 [27:43<1:48:39,  1.63it/s]
 20%|██        | 2691/13317 [27:43<1:48:37,  1.63it/s]
                                                      
{'loss': 0.3111, 'grad_norm': 0.7026335000991821, 'learning_rate': 9.686090073005256e-06, 'epoch': 0.61}

 20%|██        | 2691/13317 [27:43<1:48:37,  1.63it/s]
 20%|██        | 2692/13317 [27:44<1:48:35,  1.63it/s]
                                                      
{'loss': 0.2282, 'grad_norm': 0.555158793926239, 'learning_rate': 9.685632835815519e-06, 'epoch': 0.61}

 20%|██        | 2692/13317 [27:44<1:48:35,  1.63it/s]
 20%|██        | 2693/13317 [27:44<1:48:36,  1.63it/s]
                                                      
{'loss': 0.2498, 'grad_norm': 0.538512110710144, 'learning_rate': 9.685175276673192e-06, 'epoch': 0.61}

 20%|██        | 2693/13317 [27:45<1:48:36,  1.63it/s]
 20%|██        | 2694/13317 [27:45<1:48:37,  1.63it/s]
                                                      
{'loss': 0.2257, 'grad_norm': 0.5548421740531921, 'learning_rate': 9.684717395609717e-06, 'epoch': 0.61}

 20%|██        | 2694/13317 [27:45<1:48:37,  1.63it/s]
 20%|██        | 2695/13317 [27:46<1:48:35,  1.63it/s]
                                                      
{'loss': 0.2675, 'grad_norm': 0.4919334352016449, 'learning_rate': 9.684259192656554e-06, 'epoch': 0.61}

 20%|██        | 2695/13317 [27:46<1:48:35,  1.63it/s]
 20%|██        | 2696/13317 [27:46<1:48:34,  1.63it/s]
                                                      
{'loss': 0.2922, 'grad_norm': 0.5935472846031189, 'learning_rate': 9.683800667845188e-06, 'epoch': 0.61}

 20%|██        | 2696/13317 [27:46<1:48:34,  1.63it/s]
 20%|██        | 2697/13317 [27:47<1:48:31,  1.63it/s]
                                                      
{'loss': 0.2227, 'grad_norm': 0.5690497756004333, 'learning_rate': 9.68334182120712e-06, 'epoch': 0.61}

 20%|██        | 2697/13317 [27:47<1:48:31,  1.63it/s]
 20%|██        | 2698/13317 [27:48<1:48:30,  1.63it/s]
                                                      
{'loss': 0.2575, 'grad_norm': 0.5473765134811401, 'learning_rate': 9.682882652773883e-06, 'epoch': 0.61}

 20%|██        | 2698/13317 [27:48<1:48:30,  1.63it/s]
 20%|██        | 2699/13317 [27:48<1:48:31,  1.63it/s]
                                                      
{'loss': 0.2718, 'grad_norm': 0.5733469128608704, 'learning_rate': 9.682423162577026e-06, 'epoch': 0.61}

 20%|██        | 2699/13317 [27:48<1:48:31,  1.63it/s]
 20%|██        | 2700/13317 [27:49<1:48:28,  1.63it/s]
                                                      
{'loss': 0.2513, 'grad_norm': 0.5666300654411316, 'learning_rate': 9.681963350648118e-06, 'epoch': 0.61}

 20%|██        | 2700/13317 [27:49<1:48:28,  1.63it/s]
 20%|██        | 2701/13317 [27:49<1:48:30,  1.63it/s]
                                                      
{'loss': 0.3013, 'grad_norm': 0.5807902812957764, 'learning_rate': 9.681503217018754e-06, 'epoch': 0.61}

 20%|██        | 2701/13317 [27:49<1:48:30,  1.63it/s]
 20%|██        | 2702/13317 [27:50<1:48:31,  1.63it/s]
                                                      
{'loss': 0.257, 'grad_norm': 0.5116370916366577, 'learning_rate': 9.681042761720552e-06, 'epoch': 0.61}

 20%|██        | 2702/13317 [27:50<1:48:31,  1.63it/s]
 20%|██        | 2703/13317 [27:51<1:48:32,  1.63it/s]
                                                      
{'loss': 0.321, 'grad_norm': 0.6076593399047852, 'learning_rate': 9.680581984785146e-06, 'epoch': 0.61}

 20%|██        | 2703/13317 [27:51<1:48:32,  1.63it/s]
 20%|██        | 2704/13317 [27:51<1:48:30,  1.63it/s]
                                                      
{'loss': 0.1954, 'grad_norm': 0.4659391939640045, 'learning_rate': 9.6801208862442e-06, 'epoch': 0.61}

 20%|██        | 2704/13317 [27:51<1:48:30,  1.63it/s]
 20%|██        | 2705/13317 [27:52<1:48:29,  1.63it/s]
                                                      
{'loss': 0.2809, 'grad_norm': 0.5548368692398071, 'learning_rate': 9.679659466129396e-06, 'epoch': 0.61}

 20%|██        | 2705/13317 [27:52<1:48:29,  1.63it/s]
 20%|██        | 2706/13317 [27:52<1:48:29,  1.63it/s]
                                                      
{'loss': 0.2424, 'grad_norm': 0.4899936616420746, 'learning_rate': 9.679197724472437e-06, 'epoch': 0.61}

 20%|██        | 2706/13317 [27:53<1:48:29,  1.63it/s]
 20%|██        | 2707/13317 [27:53<1:48:28,  1.63it/s]
                                                      
{'loss': 0.2698, 'grad_norm': 0.5546513199806213, 'learning_rate': 9.67873566130505e-06, 'epoch': 0.61}

 20%|██        | 2707/13317 [27:53<1:48:28,  1.63it/s]
 20%|██        | 2708/13317 [27:54<1:48:28,  1.63it/s]
                                                      
{'loss': 0.2429, 'grad_norm': 0.5139153003692627, 'learning_rate': 9.678273276658985e-06, 'epoch': 0.61}

 20%|██        | 2708/13317 [27:54<1:48:28,  1.63it/s]
 20%|██        | 2709/13317 [27:54<1:48:25,  1.63it/s]
                                                      
{'loss': 0.3479, 'grad_norm': 0.6624578833580017, 'learning_rate': 9.677810570566011e-06, 'epoch': 0.61}

 20%|██        | 2709/13317 [27:54<1:48:25,  1.63it/s]
 20%|██        | 2710/13317 [27:55<1:48:25,  1.63it/s]
                                                      
{'loss': 0.2473, 'grad_norm': 0.5383546352386475, 'learning_rate': 9.67734754305792e-06, 'epoch': 0.61}

 20%|██        | 2710/13317 [27:55<1:48:25,  1.63it/s]
 20%|██        | 2711/13317 [27:55<1:48:22,  1.63it/s]
                                                      
{'loss': 0.2998, 'grad_norm': 0.6268560886383057, 'learning_rate': 9.67688419416653e-06, 'epoch': 0.61}

 20%|██        | 2711/13317 [27:56<1:48:22,  1.63it/s]
 20%|██        | 2712/13317 [27:56<1:48:22,  1.63it/s]
                                                      
{'loss': 0.3528, 'grad_norm': 0.5554570555686951, 'learning_rate': 9.676420523923675e-06, 'epoch': 0.61}

 20%|██        | 2712/13317 [27:56<1:48:22,  1.63it/s]
 20%|██        | 2713/13317 [27:57<1:48:21,  1.63it/s]
                                                      
{'loss': 0.3505, 'grad_norm': 0.5118503570556641, 'learning_rate': 9.675956532361215e-06, 'epoch': 0.61}

 20%|██        | 2713/13317 [27:57<1:48:21,  1.63it/s]
 20%|██        | 2714/13317 [27:57<1:48:19,  1.63it/s]
                                                      
{'loss': 0.3226, 'grad_norm': 0.6550145149230957, 'learning_rate': 9.675492219511033e-06, 'epoch': 0.61}

 20%|██        | 2714/13317 [27:57<1:48:19,  1.63it/s]
 20%|██        | 2715/13317 [27:58<1:48:20,  1.63it/s]
                                                      
{'loss': 0.2776, 'grad_norm': 0.5366286635398865, 'learning_rate': 9.675027585405031e-06, 'epoch': 0.61}

 20%|██        | 2715/13317 [27:58<1:48:20,  1.63it/s]
 20%|██        | 2716/13317 [27:59<1:48:17,  1.63it/s]
                                                      
{'loss': 0.2629, 'grad_norm': 0.5932427048683167, 'learning_rate': 9.67456263007513e-06, 'epoch': 0.61}

 20%|██        | 2716/13317 [27:59<1:48:17,  1.63it/s]
 20%|██        | 2717/13317 [27:59<1:48:19,  1.63it/s]
                                                      
{'loss': 0.2391, 'grad_norm': 0.4699084162712097, 'learning_rate': 9.674097353553286e-06, 'epoch': 0.61}

 20%|██        | 2717/13317 [27:59<1:48:19,  1.63it/s]
 20%|██        | 2718/13317 [28:00<1:48:21,  1.63it/s]
                                                      
{'loss': 0.2437, 'grad_norm': 0.5477875471115112, 'learning_rate': 9.673631755871461e-06, 'epoch': 0.61}

 20%|██        | 2718/13317 [28:00<1:48:21,  1.63it/s]
 20%|██        | 2719/13317 [28:00<1:48:22,  1.63it/s]
                                                      
{'loss': 0.2516, 'grad_norm': 0.4691133499145508, 'learning_rate': 9.673165837061652e-06, 'epoch': 0.61}

 20%|██        | 2719/13317 [28:01<1:48:22,  1.63it/s]
 20%|██        | 2720/13317 [28:01<1:48:18,  1.63it/s]
                                                      
{'loss': 0.2001, 'grad_norm': 0.46168890595436096, 'learning_rate': 9.672699597155865e-06, 'epoch': 0.61}

 20%|██        | 2720/13317 [28:01<1:48:18,  1.63it/s]
 20%|██        | 2721/13317 [28:02<1:48:17,  1.63it/s]
                                                      
{'loss': 0.3124, 'grad_norm': 0.5423812866210938, 'learning_rate': 9.672233036186144e-06, 'epoch': 0.61}

 20%|██        | 2721/13317 [28:02<1:48:17,  1.63it/s]
 20%|██        | 2722/13317 [28:02<1:48:17,  1.63it/s]
                                                      
{'loss': 0.251, 'grad_norm': 0.5326955914497375, 'learning_rate': 9.671766154184542e-06, 'epoch': 0.61}

 20%|██        | 2722/13317 [28:02<1:48:17,  1.63it/s]
 20%|██        | 2723/13317 [28:03<1:48:19,  1.63it/s]
                                                      
{'loss': 0.222, 'grad_norm': 0.4776618480682373, 'learning_rate': 9.671298951183139e-06, 'epoch': 0.61}

 20%|██        | 2723/13317 [28:03<1:48:19,  1.63it/s]
 20%|██        | 2724/13317 [28:03<1:48:20,  1.63it/s]
                                                      
{'loss': 0.2396, 'grad_norm': 0.5315869450569153, 'learning_rate': 9.670831427214037e-06, 'epoch': 0.61}

 20%|██        | 2724/13317 [28:04<1:48:20,  1.63it/s]
 20%|██        | 2725/13317 [28:04<1:48:43,  1.62it/s]
                                                      
{'loss': 0.2834, 'grad_norm': 0.5754846930503845, 'learning_rate': 9.670363582309363e-06, 'epoch': 0.61}

 20%|██        | 2725/13317 [28:04<1:48:43,  1.62it/s]
 20%|██        | 2726/13317 [28:05<1:48:34,  1.63it/s]
                                                      
{'loss': 0.3014, 'grad_norm': 0.6324006915092468, 'learning_rate': 9.669895416501259e-06, 'epoch': 0.61}

 20%|██        | 2726/13317 [28:05<1:48:34,  1.63it/s]
 20%|██        | 2727/13317 [28:05<1:48:29,  1.63it/s]
                                                      
{'loss': 0.3109, 'grad_norm': 0.5322806239128113, 'learning_rate': 9.669426929821894e-06, 'epoch': 0.61}

 20%|██        | 2727/13317 [28:05<1:48:29,  1.63it/s]
 20%|██        | 2728/13317 [28:06<1:48:23,  1.63it/s]
                                                      
{'loss': 0.2607, 'grad_norm': 0.5209359526634216, 'learning_rate': 9.668958122303458e-06, 'epoch': 0.61}

 20%|██        | 2728/13317 [28:06<1:48:23,  1.63it/s]
 20%|██        | 2729/13317 [28:07<1:48:22,  1.63it/s]
                                                      
{'loss': 0.2368, 'grad_norm': 0.5383862257003784, 'learning_rate': 9.668488993978164e-06, 'epoch': 0.61}

 20%|██        | 2729/13317 [28:07<1:48:22,  1.63it/s]
 21%|██        | 2730/13317 [28:07<1:48:23,  1.63it/s]
                                                      
{'loss': 0.1755, 'grad_norm': 0.4151071012020111, 'learning_rate': 9.668019544878245e-06, 'epoch': 0.62}

 21%|██        | 2730/13317 [28:07<1:48:23,  1.63it/s]
 21%|██        | 2731/13317 [28:08<1:48:17,  1.63it/s]
                                                      
{'loss': 0.23, 'grad_norm': 0.5576832294464111, 'learning_rate': 9.667549775035957e-06, 'epoch': 0.62}

 21%|██        | 2731/13317 [28:08<1:48:17,  1.63it/s]
 21%|██        | 2732/13317 [28:08<1:48:13,  1.63it/s]
                                                      
{'loss': 0.3087, 'grad_norm': 0.5442100167274475, 'learning_rate': 9.66707968448358e-06, 'epoch': 0.62}

 21%|██        | 2732/13317 [28:09<1:48:13,  1.63it/s]
 21%|██        | 2733/13317 [28:09<1:48:13,  1.63it/s]
                                                      
{'loss': 0.2803, 'grad_norm': 0.5992059707641602, 'learning_rate': 9.666609273253411e-06, 'epoch': 0.62}

 21%|██        | 2733/13317 [28:09<1:48:13,  1.63it/s]
 21%|██        | 2734/13317 [28:10<1:48:21,  1.63it/s]
                                                      
{'loss': 0.2775, 'grad_norm': 0.5915324091911316, 'learning_rate': 9.666138541377775e-06, 'epoch': 0.62}

 21%|██        | 2734/13317 [28:10<1:48:21,  1.63it/s]
 21%|██        | 2735/13317 [28:10<1:48:21,  1.63it/s]
                                                      
{'loss': 0.3242, 'grad_norm': 0.5999455451965332, 'learning_rate': 9.665667488889016e-06, 'epoch': 0.62}

 21%|██        | 2735/13317 [28:10<1:48:21,  1.63it/s]
 21%|██        | 2736/13317 [28:11<1:48:18,  1.63it/s]
                                                      
{'loss': 0.2902, 'grad_norm': 0.5716148614883423, 'learning_rate': 9.6651961158195e-06, 'epoch': 0.62}

 21%|██        | 2736/13317 [28:11<1:48:18,  1.63it/s]
 21%|██        | 2737/13317 [28:11<1:48:13,  1.63it/s]
                                                      
{'loss': 0.2452, 'grad_norm': 0.5574982762336731, 'learning_rate': 9.664724422201612e-06, 'epoch': 0.62}

 21%|██        | 2737/13317 [28:12<1:48:13,  1.63it/s]
 21%|██        | 2738/13317 [28:12<1:48:12,  1.63it/s]
                                                      
{'loss': 0.2791, 'grad_norm': 0.6307194232940674, 'learning_rate': 9.664252408067769e-06, 'epoch': 0.62}

 21%|██        | 2738/13317 [28:12<1:48:12,  1.63it/s]
 21%|██        | 2739/13317 [28:13<1:48:10,  1.63it/s]
                                                      
{'loss': 0.2205, 'grad_norm': 0.5324397087097168, 'learning_rate': 9.663780073450397e-06, 'epoch': 0.62}

 21%|██        | 2739/13317 [28:13<1:48:10,  1.63it/s]
 21%|██        | 2740/13317 [28:13<1:48:10,  1.63it/s]
                                                      
{'loss': 0.3059, 'grad_norm': 0.6620994210243225, 'learning_rate': 9.663307418381955e-06, 'epoch': 0.62}

 21%|██        | 2740/13317 [28:13<1:48:10,  1.63it/s]
 21%|██        | 2741/13317 [28:14<1:48:06,  1.63it/s]
                                                      
{'loss': 0.2565, 'grad_norm': 0.6361128687858582, 'learning_rate': 9.662834442894915e-06, 'epoch': 0.62}

 21%|██        | 2741/13317 [28:14<1:48:06,  1.63it/s]
 21%|██        | 2742/13317 [28:15<1:48:05,  1.63it/s]
                                                      
{'loss': 0.3428, 'grad_norm': 0.6085056066513062, 'learning_rate': 9.66236114702178e-06, 'epoch': 0.62}

 21%|██        | 2742/13317 [28:15<1:48:05,  1.63it/s]
 21%|██        | 2743/13317 [28:15<1:48:07,  1.63it/s]
                                                      
{'loss': 0.2937, 'grad_norm': 0.603468120098114, 'learning_rate': 9.661887530795068e-06, 'epoch': 0.62}

 21%|██        | 2743/13317 [28:15<1:48:07,  1.63it/s]
 21%|██        | 2744/13317 [28:16<1:48:06,  1.63it/s]
                                                      
{'loss': 0.3268, 'grad_norm': 0.6630398035049438, 'learning_rate': 9.66141359424732e-06, 'epoch': 0.62}

 21%|██        | 2744/13317 [28:16<1:48:06,  1.63it/s]
 21%|██        | 2745/13317 [28:16<1:48:05,  1.63it/s]
                                                      
{'loss': 0.3263, 'grad_norm': 0.628219723701477, 'learning_rate': 9.660939337411105e-06, 'epoch': 0.62}

 21%|██        | 2745/13317 [28:16<1:48:05,  1.63it/s]
 21%|██        | 2746/13317 [28:17<1:48:06,  1.63it/s]
                                                      
{'loss': 0.3727, 'grad_norm': 0.6032661199569702, 'learning_rate': 9.660464760319005e-06, 'epoch': 0.62}

 21%|██        | 2746/13317 [28:17<1:48:06,  1.63it/s]
 21%|██        | 2747/13317 [28:18<1:48:06,  1.63it/s]
                                                      
{'loss': 0.3158, 'grad_norm': 0.7920635342597961, 'learning_rate': 9.65998986300363e-06, 'epoch': 0.62}

 21%|██        | 2747/13317 [28:18<1:48:06,  1.63it/s]
 21%|██        | 2748/13317 [28:18<1:48:09,  1.63it/s]
                                                      
{'loss': 0.2789, 'grad_norm': 0.5247315168380737, 'learning_rate': 9.659514645497612e-06, 'epoch': 0.62}

 21%|██        | 2748/13317 [28:18<1:48:09,  1.63it/s]
 21%|██        | 2749/13317 [28:19<1:48:08,  1.63it/s]
                                                      
{'loss': 0.3714, 'grad_norm': 0.6559271216392517, 'learning_rate': 9.659039107833601e-06, 'epoch': 0.62}

 21%|██        | 2749/13317 [28:19<1:48:08,  1.63it/s]
 21%|██        | 2750/13317 [28:19<1:48:06,  1.63it/s]
                                                      
{'loss': 0.2698, 'grad_norm': 0.6047731637954712, 'learning_rate': 9.658563250044273e-06, 'epoch': 0.62}

 21%|██        | 2750/13317 [28:20<1:48:06,  1.63it/s]
 21%|██        | 2751/13317 [28:20<1:48:06,  1.63it/s]
                                                      
{'loss': 0.3145, 'grad_norm': 0.6220943331718445, 'learning_rate': 9.658087072162324e-06, 'epoch': 0.62}

 21%|██        | 2751/13317 [28:20<1:48:06,  1.63it/s]
 21%|██        | 2752/13317 [28:21<1:48:01,  1.63it/s]
                                                      
{'loss': 0.3779, 'grad_norm': 0.6116410493850708, 'learning_rate': 9.657610574220473e-06, 'epoch': 0.62}

 21%|██        | 2752/13317 [28:21<1:48:01,  1.63it/s]
 21%|██        | 2753/13317 [28:21<1:48:00,  1.63it/s]
                                                      
{'loss': 0.2517, 'grad_norm': 0.5547530055046082, 'learning_rate': 9.65713375625146e-06, 'epoch': 0.62}

 21%|██        | 2753/13317 [28:21<1:48:00,  1.63it/s]
 21%|██        | 2754/13317 [28:22<1:47:59,  1.63it/s]
                                                      
{'loss': 0.2295, 'grad_norm': 0.5522585511207581, 'learning_rate': 9.656656618288046e-06, 'epoch': 0.62}

 21%|██        | 2754/13317 [28:22<1:47:59,  1.63it/s]
 21%|██        | 2755/13317 [28:22<1:47:59,  1.63it/s]
                                                      
{'loss': 0.1942, 'grad_norm': 0.4754961431026459, 'learning_rate': 9.656179160363017e-06, 'epoch': 0.62}

 21%|██        | 2755/13317 [28:23<1:47:59,  1.63it/s]
 21%|██        | 2756/13317 [28:23<1:47:56,  1.63it/s]
                                                      
{'loss': 0.3016, 'grad_norm': 0.5516514182090759, 'learning_rate': 9.655701382509182e-06, 'epoch': 0.62}

 21%|██        | 2756/13317 [28:23<1:47:56,  1.63it/s]
 21%|██        | 2757/13317 [28:24<1:47:56,  1.63it/s]
                                                      
{'loss': 0.2469, 'grad_norm': 0.4979008734226227, 'learning_rate': 9.655223284759363e-06, 'epoch': 0.62}

 21%|██        | 2757/13317 [28:24<1:47:56,  1.63it/s]
 21%|██        | 2758/13317 [28:24<1:47:54,  1.63it/s]
                                                      
{'loss': 0.2205, 'grad_norm': 0.508577823638916, 'learning_rate': 9.654744867146417e-06, 'epoch': 0.62}

 21%|██        | 2758/13317 [28:24<1:47:54,  1.63it/s]
 21%|██        | 2759/13317 [28:25<1:47:55,  1.63it/s]
                                                      
{'loss': 0.2996, 'grad_norm': 0.5685909390449524, 'learning_rate': 9.654266129703211e-06, 'epoch': 0.62}

 21%|██        | 2759/13317 [28:25<1:47:55,  1.63it/s]
 21%|██        | 2760/13317 [28:26<1:47:55,  1.63it/s]
                                                      
{'loss': 0.2233, 'grad_norm': 0.48626115918159485, 'learning_rate': 9.653787072462644e-06, 'epoch': 0.62}

 21%|██        | 2760/13317 [28:26<1:47:55,  1.63it/s]
 21%|██        | 2761/13317 [28:26<1:47:57,  1.63it/s]
                                                      
{'loss': 0.2921, 'grad_norm': 0.7466758489608765, 'learning_rate': 9.653307695457628e-06, 'epoch': 0.62}

 21%|██        | 2761/13317 [28:26<1:47:57,  1.63it/s]
 21%|██        | 2762/13317 [28:27<1:47:58,  1.63it/s]
                                                      
{'loss': 0.279, 'grad_norm': 0.5368276238441467, 'learning_rate': 9.652827998721106e-06, 'epoch': 0.62}

 21%|██        | 2762/13317 [28:27<1:47:58,  1.63it/s]
 21%|██        | 2763/13317 [28:27<1:47:58,  1.63it/s]
                                                      
{'loss': 0.2823, 'grad_norm': 0.5382733345031738, 'learning_rate': 9.652347982286033e-06, 'epoch': 0.62}

 21%|██        | 2763/13317 [28:28<1:47:58,  1.63it/s]
 21%|██        | 2764/13317 [28:28<1:47:55,  1.63it/s]
                                                      
{'loss': 0.2982, 'grad_norm': 0.6833263635635376, 'learning_rate': 9.651867646185394e-06, 'epoch': 0.62}

 21%|██        | 2764/13317 [28:28<1:47:55,  1.63it/s]
 21%|██        | 2765/13317 [28:29<1:47:51,  1.63it/s]
                                                      
{'loss': 0.2981, 'grad_norm': 0.6760875582695007, 'learning_rate': 9.651386990452194e-06, 'epoch': 0.62}

 21%|██        | 2765/13317 [28:29<1:47:51,  1.63it/s]
 21%|██        | 2766/13317 [28:29<1:47:51,  1.63it/s]
                                                      
{'loss': 0.1905, 'grad_norm': 0.4523831605911255, 'learning_rate': 9.650906015119456e-06, 'epoch': 0.62}

 21%|██        | 2766/13317 [28:29<1:47:51,  1.63it/s]
 21%|██        | 2767/13317 [28:30<1:47:50,  1.63it/s]
                                                      
{'loss': 0.1799, 'grad_norm': 0.43607836961746216, 'learning_rate': 9.650424720220233e-06, 'epoch': 0.62}

 21%|██        | 2767/13317 [28:30<1:47:50,  1.63it/s]
 21%|██        | 2768/13317 [28:30<1:47:50,  1.63it/s]
                                                      
{'loss': 0.2095, 'grad_norm': 0.5242013335227966, 'learning_rate': 9.649943105787591e-06, 'epoch': 0.62}

 21%|██        | 2768/13317 [28:31<1:47:50,  1.63it/s]
 21%|██        | 2769/13317 [28:31<1:47:50,  1.63it/s]
                                                      
{'loss': 0.3394, 'grad_norm': 0.5484368205070496, 'learning_rate': 9.649461171854624e-06, 'epoch': 0.62}

 21%|██        | 2769/13317 [28:31<1:47:50,  1.63it/s]
 21%|██        | 2770/13317 [28:32<1:47:50,  1.63it/s]
                                                      
{'loss': 0.2337, 'grad_norm': 0.5413293838500977, 'learning_rate': 9.648978918454444e-06, 'epoch': 0.62}

 21%|██        | 2770/13317 [28:32<1:47:50,  1.63it/s]
 21%|██        | 2771/13317 [28:32<1:47:50,  1.63it/s]
                                                      
{'loss': 0.3095, 'grad_norm': 0.5736846923828125, 'learning_rate': 9.648496345620188e-06, 'epoch': 0.62}

 21%|██        | 2771/13317 [28:32<1:47:50,  1.63it/s]
 21%|██        | 2772/13317 [28:33<1:47:46,  1.63it/s]
                                                      
{'loss': 0.2939, 'grad_norm': 0.6242747902870178, 'learning_rate': 9.648013453385014e-06, 'epoch': 0.62}

 21%|██        | 2772/13317 [28:33<1:47:46,  1.63it/s]
 21%|██        | 2773/13317 [28:34<1:47:44,  1.63it/s]
                                                      
{'loss': 0.2902, 'grad_norm': 0.6023358702659607, 'learning_rate': 9.647530241782103e-06, 'epoch': 0.62}

 21%|██        | 2773/13317 [28:34<1:47:44,  1.63it/s]
 21%|██        | 2774/13317 [28:34<1:47:45,  1.63it/s]
                                                      
{'loss': 0.2469, 'grad_norm': 0.5847731232643127, 'learning_rate': 9.647046710844654e-06, 'epoch': 0.62}

 21%|██        | 2774/13317 [28:34<1:47:45,  1.63it/s]
 21%|██        | 2775/13317 [28:35<1:47:48,  1.63it/s]
                                                      
{'loss': 0.2088, 'grad_norm': 0.5210197567939758, 'learning_rate': 9.646562860605894e-06, 'epoch': 0.63}

 21%|██        | 2775/13317 [28:35<1:47:48,  1.63it/s]
 21%|██        | 2776/13317 [28:35<1:47:48,  1.63it/s]
                                                      
{'loss': 0.277, 'grad_norm': 0.5877362489700317, 'learning_rate': 9.646078691099066e-06, 'epoch': 0.63}

 21%|██        | 2776/13317 [28:36<1:47:48,  1.63it/s]
 21%|██        | 2777/13317 [28:36<1:47:48,  1.63it/s]
                                                      
{'loss': 0.2104, 'grad_norm': 0.44975149631500244, 'learning_rate': 9.645594202357438e-06, 'epoch': 0.63}

 21%|██        | 2777/13317 [28:36<1:47:48,  1.63it/s]
 21%|██        | 2778/13317 [28:37<1:47:48,  1.63it/s]
                                                      
{'loss': 0.258, 'grad_norm': 0.5554429292678833, 'learning_rate': 9.6451093944143e-06, 'epoch': 0.63}

 21%|██        | 2778/13317 [28:37<1:47:48,  1.63it/s]
 21%|██        | 2779/13317 [28:37<1:47:47,  1.63it/s]
                                                      
{'loss': 0.3822, 'grad_norm': 0.6590362787246704, 'learning_rate': 9.644624267302964e-06, 'epoch': 0.63}

 21%|██        | 2779/13317 [28:37<1:47:47,  1.63it/s]
 21%|██        | 2780/13317 [28:38<1:47:46,  1.63it/s]
                                                      
{'loss': 0.2353, 'grad_norm': 0.48424363136291504, 'learning_rate': 9.644138821056763e-06, 'epoch': 0.63}

 21%|██        | 2780/13317 [28:38<1:47:46,  1.63it/s]
 21%|██        | 2781/13317 [28:38<1:47:48,  1.63it/s]
                                                      
{'loss': 0.2416, 'grad_norm': 0.5140730738639832, 'learning_rate': 9.643653055709052e-06, 'epoch': 0.63}

 21%|██        | 2781/13317 [28:39<1:47:48,  1.63it/s]
 21%|██        | 2782/13317 [28:39<1:47:47,  1.63it/s]
                                                      
{'loss': 0.234, 'grad_norm': 0.47979480028152466, 'learning_rate': 9.643166971293207e-06, 'epoch': 0.63}

 21%|██        | 2782/13317 [28:39<1:47:47,  1.63it/s]
 21%|██        | 2783/13317 [28:40<1:47:47,  1.63it/s]
                                                      
{'loss': 0.2397, 'grad_norm': 0.47276729345321655, 'learning_rate': 9.642680567842631e-06, 'epoch': 0.63}

 21%|██        | 2783/13317 [28:40<1:47:47,  1.63it/s]
 21%|██        | 2784/13317 [28:40<1:47:43,  1.63it/s]
                                                      
{'loss': 0.2812, 'grad_norm': 0.5713816285133362, 'learning_rate': 9.64219384539074e-06, 'epoch': 0.63}

 21%|██        | 2784/13317 [28:40<1:47:43,  1.63it/s]
 21%|██        | 2785/13317 [28:41<1:47:43,  1.63it/s]
                                                      
{'loss': 0.3049, 'grad_norm': 0.6057060360908508, 'learning_rate': 9.64170680397098e-06, 'epoch': 0.63}

 21%|██        | 2785/13317 [28:41<1:47:43,  1.63it/s]
 21%|██        | 2786/13317 [28:42<1:47:42,  1.63it/s]
                                                      
{'loss': 0.2723, 'grad_norm': 0.5553610324859619, 'learning_rate': 9.641219443616817e-06, 'epoch': 0.63}

 21%|██        | 2786/13317 [28:42<1:47:42,  1.63it/s]
 21%|██        | 2787/13317 [28:42<1:47:40,  1.63it/s]
                                                      
{'loss': 0.1795, 'grad_norm': 0.42166638374328613, 'learning_rate': 9.640731764361734e-06, 'epoch': 0.63}

 21%|██        | 2787/13317 [28:42<1:47:40,  1.63it/s]
 21%|██        | 2788/13317 [28:43<1:47:36,  1.63it/s]
                                                      
{'loss': 0.3678, 'grad_norm': 0.7077832221984863, 'learning_rate': 9.640243766239244e-06, 'epoch': 0.63}

 21%|██        | 2788/13317 [28:43<1:47:36,  1.63it/s]
 21%|██        | 2789/13317 [28:43<1:47:37,  1.63it/s]
                                                      
{'loss': 0.2903, 'grad_norm': 0.6390036344528198, 'learning_rate': 9.639755449282874e-06, 'epoch': 0.63}

 21%|██        | 2789/13317 [28:43<1:47:37,  1.63it/s]
 21%|██        | 2790/13317 [28:44<1:47:37,  1.63it/s]
                                                      
{'loss': 0.3159, 'grad_norm': 0.6282022595405579, 'learning_rate': 9.639266813526179e-06, 'epoch': 0.63}

 21%|██        | 2790/13317 [28:44<1:47:37,  1.63it/s]
 21%|██        | 2791/13317 [28:45<1:47:38,  1.63it/s]
                                                      
{'loss': 0.28, 'grad_norm': 0.5694103837013245, 'learning_rate': 9.638777859002733e-06, 'epoch': 0.63}

 21%|██        | 2791/13317 [28:45<1:47:38,  1.63it/s]
 21%|██        | 2792/13317 [28:45<1:47:38,  1.63it/s]
                                                      
{'loss': 0.2608, 'grad_norm': 0.6256498098373413, 'learning_rate': 9.63828858574613e-06, 'epoch': 0.63}

 21%|██        | 2792/13317 [28:45<1:47:38,  1.63it/s]
 21%|██        | 2793/13317 [28:46<1:47:37,  1.63it/s]
                                                      
{'loss': 0.2762, 'grad_norm': 0.5249968767166138, 'learning_rate': 9.637798993789992e-06, 'epoch': 0.63}

 21%|██        | 2793/13317 [28:46<1:47:37,  1.63it/s]
 21%|██        | 2794/13317 [28:46<1:47:35,  1.63it/s]
                                                      
{'loss': 0.2409, 'grad_norm': 0.6733025312423706, 'learning_rate': 9.637309083167956e-06, 'epoch': 0.63}

 21%|██        | 2794/13317 [28:47<1:47:35,  1.63it/s]
 21%|██        | 2795/13317 [28:47<1:47:33,  1.63it/s]
                                                      
{'loss': 0.2554, 'grad_norm': 0.4816468060016632, 'learning_rate': 9.636818853913685e-06, 'epoch': 0.63}

 21%|██        | 2795/13317 [28:47<1:47:33,  1.63it/s]
 21%|██        | 2796/13317 [28:48<1:47:31,  1.63it/s]
                                                      
{'loss': 0.225, 'grad_norm': 0.5679340362548828, 'learning_rate': 9.636328306060866e-06, 'epoch': 0.63}

 21%|██        | 2796/13317 [28:48<1:47:31,  1.63it/s]
 21%|██        | 2797/13317 [28:48<1:47:34,  1.63it/s]
                                                      
{'loss': 0.2356, 'grad_norm': 0.5877740979194641, 'learning_rate': 9.635837439643198e-06, 'epoch': 0.63}

 21%|██        | 2797/13317 [28:48<1:47:34,  1.63it/s]
 21%|██        | 2798/13317 [28:49<1:47:30,  1.63it/s]
                                                      
{'loss': 0.2712, 'grad_norm': 0.5343776941299438, 'learning_rate': 9.635346254694416e-06, 'epoch': 0.63}

 21%|██        | 2798/13317 [28:49<1:47:30,  1.63it/s]
 21%|██        | 2799/13317 [28:49<1:47:31,  1.63it/s]
                                                      
{'loss': 0.2072, 'grad_norm': 0.44234371185302734, 'learning_rate': 9.634854751248267e-06, 'epoch': 0.63}

 21%|██        | 2799/13317 [28:50<1:47:31,  1.63it/s]
 21%|██        | 2800/13317 [28:50<1:47:32,  1.63it/s]
                                                      
{'loss': 0.2674, 'grad_norm': 0.5446460247039795, 'learning_rate': 9.63436292933852e-06, 'epoch': 0.63}

 21%|██        | 2800/13317 [28:50<1:47:32,  1.63it/s]
 21%|██        | 2801/13317 [28:51<1:47:29,  1.63it/s]
                                                      
{'loss': 0.2521, 'grad_norm': 0.49219974875450134, 'learning_rate': 9.633870788998972e-06, 'epoch': 0.63}

 21%|██        | 2801/13317 [28:51<1:47:29,  1.63it/s]
 21%|██        | 2802/13317 [28:51<1:47:28,  1.63it/s]
                                                      
{'loss': 0.2609, 'grad_norm': 0.5418075919151306, 'learning_rate': 9.633378330263435e-06, 'epoch': 0.63}

 21%|██        | 2802/13317 [28:51<1:47:28,  1.63it/s]
 21%|██        | 2803/13317 [28:52<1:47:24,  1.63it/s]
                                                      
{'loss': 0.3027, 'grad_norm': 0.6369662880897522, 'learning_rate': 9.632885553165747e-06, 'epoch': 0.63}

 21%|██        | 2803/13317 [28:52<1:47:24,  1.63it/s]
 21%|██        | 2804/13317 [28:53<1:47:23,  1.63it/s]
                                                      
{'loss': 0.2835, 'grad_norm': 0.4793683588504791, 'learning_rate': 9.63239245773977e-06, 'epoch': 0.63}

 21%|██        | 2804/13317 [28:53<1:47:23,  1.63it/s]
 21%|██        | 2805/13317 [28:53<1:47:25,  1.63it/s]
                                                      
{'loss': 0.1974, 'grad_norm': 0.45748549699783325, 'learning_rate': 9.631899044019381e-06, 'epoch': 0.63}

 21%|██        | 2805/13317 [28:53<1:47:25,  1.63it/s]
 21%|██        | 2806/13317 [28:54<1:47:24,  1.63it/s]
                                                      
{'loss': 0.3054, 'grad_norm': 0.5940791368484497, 'learning_rate': 9.631405312038485e-06, 'epoch': 0.63}

 21%|██        | 2806/13317 [28:54<1:47:24,  1.63it/s]
 21%|██        | 2807/13317 [28:54<1:47:25,  1.63it/s]
                                                      
{'loss': 0.2585, 'grad_norm': 0.5770608186721802, 'learning_rate': 9.630911261831006e-06, 'epoch': 0.63}

 21%|██        | 2807/13317 [28:55<1:47:25,  1.63it/s]
 21%|██        | 2808/13317 [28:55<1:47:23,  1.63it/s]
                                                      
{'loss': 0.3904, 'grad_norm': 0.7206547856330872, 'learning_rate': 9.630416893430892e-06, 'epoch': 0.63}

 21%|██        | 2808/13317 [28:55<1:47:23,  1.63it/s]
 21%|██        | 2809/13317 [28:56<1:47:22,  1.63it/s]
                                                      
{'loss': 0.2624, 'grad_norm': 0.5326082706451416, 'learning_rate': 9.629922206872106e-06, 'epoch': 0.63}

 21%|██        | 2809/13317 [28:56<1:47:22,  1.63it/s]
 21%|██        | 2810/13317 [28:56<1:47:22,  1.63it/s]
                                                      
{'loss': 0.3533, 'grad_norm': 0.7410778403282166, 'learning_rate': 9.629427202188645e-06, 'epoch': 0.63}

 21%|██        | 2810/13317 [28:56<1:47:22,  1.63it/s]
 21%|██        | 2811/13317 [28:57<1:47:20,  1.63it/s]
                                                      
{'loss': 0.298, 'grad_norm': 0.6070356369018555, 'learning_rate': 9.628931879414519e-06, 'epoch': 0.63}

 21%|██        | 2811/13317 [28:57<1:47:20,  1.63it/s]
 21%|██        | 2812/13317 [28:57<1:47:22,  1.63it/s]
                                                      
{'loss': 0.3025, 'grad_norm': 0.6200470924377441, 'learning_rate': 9.628436238583758e-06, 'epoch': 0.63}

 21%|██        | 2812/13317 [28:58<1:47:22,  1.63it/s]
 21%|██        | 2813/13317 [28:58<1:47:21,  1.63it/s]
                                                      
{'loss': 0.2876, 'grad_norm': 0.47041475772857666, 'learning_rate': 9.627940279730421e-06, 'epoch': 0.63}

 21%|██        | 2813/13317 [28:58<1:47:21,  1.63it/s]
 21%|██        | 2814/13317 [28:59<1:47:22,  1.63it/s]
                                                      
{'loss': 0.2239, 'grad_norm': 0.5541791915893555, 'learning_rate': 9.627444002888586e-06, 'epoch': 0.63}

 21%|██        | 2814/13317 [28:59<1:47:22,  1.63it/s]
 21%|██        | 2815/13317 [28:59<1:47:23,  1.63it/s]
                                                      
{'loss': 0.2199, 'grad_norm': 0.47776079177856445, 'learning_rate': 9.626947408092354e-06, 'epoch': 0.63}

 21%|██        | 2815/13317 [28:59<1:47:23,  1.63it/s]
 21%|██        | 2816/13317 [29:00<1:47:20,  1.63it/s]
                                                      
{'loss': 0.2463, 'grad_norm': 0.6805042624473572, 'learning_rate': 9.62645049537584e-06, 'epoch': 0.63}

 21%|██        | 2816/13317 [29:00<1:47:20,  1.63it/s]
 21%|██        | 2817/13317 [29:01<1:47:17,  1.63it/s]
                                                      
{'loss': 0.2584, 'grad_norm': 0.518505871295929, 'learning_rate': 9.625953264773194e-06, 'epoch': 0.63}

 21%|██        | 2817/13317 [29:01<1:47:17,  1.63it/s]
 21%|██        | 2818/13317 [29:01<1:47:15,  1.63it/s]
                                                      
{'loss': 0.2139, 'grad_norm': 0.49282848834991455, 'learning_rate': 9.625455716318579e-06, 'epoch': 0.63}

 21%|██        | 2818/13317 [29:01<1:47:15,  1.63it/s]
 21%|██        | 2819/13317 [29:02<1:47:16,  1.63it/s]
                                                      
{'loss': 0.3377, 'grad_norm': 0.6073794364929199, 'learning_rate': 9.62495785004618e-06, 'epoch': 0.64}

 21%|██        | 2819/13317 [29:02<1:47:16,  1.63it/s]
 21%|██        | 2820/13317 [29:02<1:47:18,  1.63it/s]
                                                      
{'loss': 0.2203, 'grad_norm': 0.5213946104049683, 'learning_rate': 9.624459665990207e-06, 'epoch': 0.64}

 21%|██        | 2820/13317 [29:03<1:47:18,  1.63it/s]
 21%|██        | 2821/13317 [29:03<1:47:17,  1.63it/s]
                                                      
{'loss': 0.2807, 'grad_norm': 0.5019053220748901, 'learning_rate': 9.62396116418489e-06, 'epoch': 0.64}

 21%|██        | 2821/13317 [29:03<1:47:17,  1.63it/s]
 21%|██        | 2822/13317 [29:04<1:47:16,  1.63it/s]
                                                      
{'loss': 0.192, 'grad_norm': 0.46291568875312805, 'learning_rate': 9.623462344664482e-06, 'epoch': 0.64}

 21%|██        | 2822/13317 [29:04<1:47:16,  1.63it/s]
 21%|██        | 2823/13317 [29:04<1:47:16,  1.63it/s]
                                                      
{'loss': 0.2533, 'grad_norm': 0.5887107849121094, 'learning_rate': 9.622963207463258e-06, 'epoch': 0.64}

 21%|██        | 2823/13317 [29:04<1:47:16,  1.63it/s]
 21%|██        | 2824/13317 [29:05<1:47:16,  1.63it/s]
                                                      
{'loss': 0.2811, 'grad_norm': 0.6244810819625854, 'learning_rate': 9.622463752615511e-06, 'epoch': 0.64}

 21%|██        | 2824/13317 [29:05<1:47:16,  1.63it/s]
 21%|██        | 2825/13317 [29:05<1:47:20,  1.63it/s]
                                                      
{'loss': 0.2513, 'grad_norm': 0.5454853177070618, 'learning_rate': 9.621963980155562e-06, 'epoch': 0.64}

 21%|██        | 2825/13317 [29:06<1:47:20,  1.63it/s]
 21%|██        | 2826/13317 [29:06<1:47:17,  1.63it/s]
                                                      
{'loss': 0.2002, 'grad_norm': 0.4635004699230194, 'learning_rate': 9.621463890117749e-06, 'epoch': 0.64}

 21%|██        | 2826/13317 [29:06<1:47:17,  1.63it/s]
 21%|██        | 2827/13317 [29:07<1:47:17,  1.63it/s]
                                                      
{'loss': 0.2621, 'grad_norm': 0.5332880616188049, 'learning_rate': 9.620963482536434e-06, 'epoch': 0.64}

 21%|██        | 2827/13317 [29:07<1:47:17,  1.63it/s]
 21%|██        | 2828/13317 [29:07<1:47:17,  1.63it/s]
                                                      
{'loss': 0.2425, 'grad_norm': 0.5472901463508606, 'learning_rate': 9.620462757446001e-06, 'epoch': 0.64}

 21%|██        | 2828/13317 [29:07<1:47:17,  1.63it/s]
 21%|██        | 2829/13317 [29:08<1:47:15,  1.63it/s]
                                                      
{'loss': 0.2616, 'grad_norm': 0.5959527492523193, 'learning_rate': 9.619961714880852e-06, 'epoch': 0.64}

 21%|██        | 2829/13317 [29:08<1:47:15,  1.63it/s]
 21%|██▏       | 2830/13317 [29:09<1:47:12,  1.63it/s]
                                                      
{'loss': 0.2816, 'grad_norm': 0.6190267205238342, 'learning_rate': 9.619460354875418e-06, 'epoch': 0.64}

 21%|██▏       | 2830/13317 [29:09<1:47:12,  1.63it/s]
 21%|██▏       | 2831/13317 [29:09<1:47:11,  1.63it/s]
                                                      
{'loss': 0.2747, 'grad_norm': 0.5740389227867126, 'learning_rate': 9.618958677464147e-06, 'epoch': 0.64}

 21%|██▏       | 2831/13317 [29:09<1:47:11,  1.63it/s]
 21%|██▏       | 2832/13317 [29:10<1:47:12,  1.63it/s]
                                                      
{'loss': 0.3546, 'grad_norm': 0.6962354183197021, 'learning_rate': 9.618456682681507e-06, 'epoch': 0.64}

 21%|██▏       | 2832/13317 [29:10<1:47:12,  1.63it/s]
 21%|██▏       | 2833/13317 [29:10<1:47:17,  1.63it/s]
                                                      
{'loss': 0.3132, 'grad_norm': 0.6231335401535034, 'learning_rate': 9.617954370561992e-06, 'epoch': 0.64}

 21%|██▏       | 2833/13317 [29:10<1:47:17,  1.63it/s]
 21%|██▏       | 2834/13317 [29:11<1:47:14,  1.63it/s]
                                                      
{'loss': 0.2699, 'grad_norm': 0.6107228994369507, 'learning_rate': 9.617451741140115e-06, 'epoch': 0.64}

 21%|██▏       | 2834/13317 [29:11<1:47:14,  1.63it/s]
 21%|██▏       | 2835/13317 [29:12<1:47:15,  1.63it/s]
                                                      
{'loss': 0.2407, 'grad_norm': 0.5879822969436646, 'learning_rate': 9.616948794450417e-06, 'epoch': 0.64}

 21%|██▏       | 2835/13317 [29:12<1:47:15,  1.63it/s]
 21%|██▏       | 2836/13317 [29:12<1:47:12,  1.63it/s]
                                                      
{'loss': 0.2208, 'grad_norm': 0.4778575599193573, 'learning_rate': 9.616445530527448e-06, 'epoch': 0.64}

 21%|██▏       | 2836/13317 [29:12<1:47:12,  1.63it/s]
 21%|██▏       | 2837/13317 [29:13<1:47:06,  1.63it/s]
                                                      
{'loss': 0.3004, 'grad_norm': 0.5837284922599792, 'learning_rate': 9.61594194940579e-06, 'epoch': 0.64}

 21%|██▏       | 2837/13317 [29:13<1:47:06,  1.63it/s]
 21%|██▏       | 2838/13317 [29:13<1:47:09,  1.63it/s]
                                                      
{'loss': 0.2322, 'grad_norm': 0.4768630564212799, 'learning_rate': 9.61543805112005e-06, 'epoch': 0.64}

 21%|██▏       | 2838/13317 [29:14<1:47:09,  1.63it/s]
 21%|██▏       | 2839/13317 [29:14<1:47:08,  1.63it/s]
                                                      
{'loss': 0.293, 'grad_norm': 0.6660364270210266, 'learning_rate': 9.614933835704844e-06, 'epoch': 0.64}

 21%|██▏       | 2839/13317 [29:14<1:47:08,  1.63it/s]
 21%|██▏       | 2840/13317 [29:15<1:47:07,  1.63it/s]
                                                      
{'loss': 0.2536, 'grad_norm': 0.5772897005081177, 'learning_rate': 9.614429303194822e-06, 'epoch': 0.64}

 21%|██▏       | 2840/13317 [29:15<1:47:07,  1.63it/s]
 21%|██▏       | 2841/13317 [29:15<1:47:07,  1.63it/s]
                                                      
{'loss': 0.2111, 'grad_norm': 0.4980906844139099, 'learning_rate': 9.613924453624648e-06, 'epoch': 0.64}

 21%|██▏       | 2841/13317 [29:15<1:47:07,  1.63it/s]
 21%|██▏       | 2842/13317 [29:16<1:47:04,  1.63it/s]
                                                      
{'loss': 0.2698, 'grad_norm': 0.6183988451957703, 'learning_rate': 9.613419287029009e-06, 'epoch': 0.64}

 21%|██▏       | 2842/13317 [29:16<1:47:04,  1.63it/s]
 21%|██▏       | 2843/13317 [29:16<1:47:02,  1.63it/s]
                                                      
{'loss': 0.2394, 'grad_norm': 0.550962507724762, 'learning_rate': 9.612913803442619e-06, 'epoch': 0.64}

 21%|██▏       | 2843/13317 [29:17<1:47:02,  1.63it/s]
 21%|██▏       | 2844/13317 [29:17<1:47:02,  1.63it/s]
                                                      
{'loss': 0.2861, 'grad_norm': 0.6307234168052673, 'learning_rate': 9.612408002900208e-06, 'epoch': 0.64}

 21%|██▏       | 2844/13317 [29:17<1:47:02,  1.63it/s]
 21%|██▏       | 2845/13317 [29:18<1:47:02,  1.63it/s]
                                                      
{'loss': 0.2339, 'grad_norm': 0.47673627734184265, 'learning_rate': 9.61190188543653e-06, 'epoch': 0.64}

 21%|██▏       | 2845/13317 [29:18<1:47:02,  1.63it/s]
 21%|██▏       | 2846/13317 [29:18<1:47:01,  1.63it/s]
                                                      
{'loss': 0.2313, 'grad_norm': 0.4947374165058136, 'learning_rate': 9.611395451086361e-06, 'epoch': 0.64}

 21%|██▏       | 2846/13317 [29:18<1:47:01,  1.63it/s]
 21%|██▏       | 2847/13317 [29:19<1:47:01,  1.63it/s]
                                                      
{'loss': 0.3011, 'grad_norm': 0.657004177570343, 'learning_rate': 9.610888699884498e-06, 'epoch': 0.64}

 21%|██▏       | 2847/13317 [29:19<1:47:01,  1.63it/s]
 21%|██▏       | 2848/13317 [29:20<1:47:03,  1.63it/s]
                                                      
{'loss': 0.2991, 'grad_norm': 0.6057623028755188, 'learning_rate': 9.610381631865762e-06, 'epoch': 0.64}

 21%|██▏       | 2848/13317 [29:20<1:47:03,  1.63it/s]
 21%|██▏       | 2849/13317 [29:20<1:47:00,  1.63it/s]
                                                      
{'loss': 0.3359, 'grad_norm': 0.6462408304214478, 'learning_rate': 9.60987424706499e-06, 'epoch': 0.64}

 21%|██▏       | 2849/13317 [29:20<1:47:00,  1.63it/s]
 21%|██▏       | 2850/13317 [29:21<1:47:01,  1.63it/s]
                                                      
{'loss': 0.2936, 'grad_norm': 0.5703580379486084, 'learning_rate': 9.60936654551705e-06, 'epoch': 0.64}

 21%|██▏       | 2850/13317 [29:21<1:47:01,  1.63it/s]
 21%|██▏       | 2851/13317 [29:21<1:46:59,  1.63it/s]
                                                      
{'loss': 0.2266, 'grad_norm': 0.6321879029273987, 'learning_rate': 9.60885852725682e-06, 'epoch': 0.64}

 21%|██▏       | 2851/13317 [29:22<1:46:59,  1.63it/s]
 21%|██▏       | 2852/13317 [29:22<1:47:00,  1.63it/s]
                                                      
{'loss': 0.2094, 'grad_norm': 0.5001485347747803, 'learning_rate': 9.608350192319213e-06, 'epoch': 0.64}

 21%|██▏       | 2852/13317 [29:22<1:47:00,  1.63it/s]
 21%|██▏       | 2853/13317 [29:23<1:46:57,  1.63it/s]
                                                      
{'loss': 0.3002, 'grad_norm': 0.543254017829895, 'learning_rate': 9.607841540739152e-06, 'epoch': 0.64}

 21%|██▏       | 2853/13317 [29:23<1:46:57,  1.63it/s]
 21%|██▏       | 2854/13317 [29:23<1:46:56,  1.63it/s]
                                                      
{'loss': 0.2436, 'grad_norm': 0.5463775396347046, 'learning_rate': 9.607332572551588e-06, 'epoch': 0.64}

 21%|██▏       | 2854/13317 [29:23<1:46:56,  1.63it/s]
 21%|██▏       | 2855/13317 [29:24<1:46:59,  1.63it/s]
                                                      
{'loss': 0.2178, 'grad_norm': 0.5018924474716187, 'learning_rate': 9.606823287791496e-06, 'epoch': 0.64}

 21%|██▏       | 2855/13317 [29:24<1:46:59,  1.63it/s]
 21%|██▏       | 2856/13317 [29:24<1:46:56,  1.63it/s]
                                                      
{'loss': 0.2493, 'grad_norm': 0.5285808444023132, 'learning_rate': 9.606313686493865e-06, 'epoch': 0.64}

 21%|██▏       | 2856/13317 [29:25<1:46:56,  1.63it/s]
 21%|██▏       | 2857/13317 [29:25<1:46:57,  1.63it/s]
                                                      
{'loss': 0.3228, 'grad_norm': 0.5635054111480713, 'learning_rate': 9.60580376869371e-06, 'epoch': 0.64}

 21%|██▏       | 2857/13317 [29:25<1:46:57,  1.63it/s]
 21%|██▏       | 2858/13317 [29:26<1:46:56,  1.63it/s]
                                                      
{'loss': 0.2256, 'grad_norm': 0.44433051347732544, 'learning_rate': 9.605293534426068e-06, 'epoch': 0.64}

 21%|██▏       | 2858/13317 [29:26<1:46:56,  1.63it/s]
 21%|██▏       | 2859/13317 [29:26<1:46:56,  1.63it/s]
                                                      
{'loss': 0.3462, 'grad_norm': 0.6117974519729614, 'learning_rate': 9.604782983726002e-06, 'epoch': 0.64}

 21%|██▏       | 2859/13317 [29:26<1:46:56,  1.63it/s]
 21%|██▏       | 2860/13317 [29:27<1:46:54,  1.63it/s]
                                                      
{'loss': 0.2362, 'grad_norm': 0.5569275617599487, 'learning_rate': 9.604272116628585e-06, 'epoch': 0.64}

 21%|██▏       | 2860/13317 [29:27<1:46:54,  1.63it/s]
 21%|██▏       | 2861/13317 [29:28<1:46:53,  1.63it/s]
                                                      
{'loss': 0.2156, 'grad_norm': 0.4979840815067291, 'learning_rate': 9.603760933168927e-06, 'epoch': 0.64}

 21%|██▏       | 2861/13317 [29:28<1:46:53,  1.63it/s]
 21%|██▏       | 2862/13317 [29:28<1:46:51,  1.63it/s]
                                                      
{'loss': 0.2613, 'grad_norm': 0.5220102667808533, 'learning_rate': 9.603249433382145e-06, 'epoch': 0.64}

 21%|██▏       | 2862/13317 [29:28<1:46:51,  1.63it/s]
 21%|██▏       | 2863/13317 [29:29<1:46:54,  1.63it/s]
                                                      
{'loss': 0.2319, 'grad_norm': 0.548119306564331, 'learning_rate': 9.602737617303387e-06, 'epoch': 0.64}

 21%|██▏       | 2863/13317 [29:29<1:46:54,  1.63it/s]
 22%|██▏       | 2864/13317 [29:29<1:46:52,  1.63it/s]
                                                      
{'loss': 0.3883, 'grad_norm': 0.724042534828186, 'learning_rate': 9.60222548496782e-06, 'epoch': 0.65}

 22%|██▏       | 2864/13317 [29:29<1:46:52,  1.63it/s]
 22%|██▏       | 2865/13317 [29:30<1:46:51,  1.63it/s]
                                                      
{'loss': 0.2531, 'grad_norm': 0.5199875235557556, 'learning_rate': 9.601713036410633e-06, 'epoch': 0.65}

 22%|██▏       | 2865/13317 [29:30<1:46:51,  1.63it/s]
 22%|██▏       | 2866/13317 [29:31<1:46:50,  1.63it/s]
                                                      
{'loss': 0.2958, 'grad_norm': 0.6073821783065796, 'learning_rate': 9.601200271667038e-06, 'epoch': 0.65}

 22%|██▏       | 2866/13317 [29:31<1:46:50,  1.63it/s]
 22%|██▏       | 2867/13317 [29:31<1:46:51,  1.63it/s]
                                                      
{'loss': 0.284, 'grad_norm': 0.5642223954200745, 'learning_rate': 9.600687190772266e-06, 'epoch': 0.65}

 22%|██▏       | 2867/13317 [29:31<1:46:51,  1.63it/s]
 22%|██▏       | 2868/13317 [29:32<1:46:51,  1.63it/s]
                                                      
{'loss': 0.2627, 'grad_norm': 0.5573248863220215, 'learning_rate': 9.600173793761569e-06, 'epoch': 0.65}

 22%|██▏       | 2868/13317 [29:32<1:46:51,  1.63it/s]
 22%|██▏       | 2869/13317 [29:32<1:46:48,  1.63it/s]
                                                      
{'loss': 0.2586, 'grad_norm': 0.6206027269363403, 'learning_rate': 9.599660080670226e-06, 'epoch': 0.65}

 22%|██▏       | 2869/13317 [29:33<1:46:48,  1.63it/s]
 22%|██▏       | 2870/13317 [29:33<1:46:47,  1.63it/s]
                                                      
{'loss': 0.312, 'grad_norm': 0.6056125164031982, 'learning_rate': 9.599146051533534e-06, 'epoch': 0.65}

 22%|██▏       | 2870/13317 [29:33<1:46:47,  1.63it/s]
 22%|██▏       | 2871/13317 [29:34<1:46:45,  1.63it/s]
                                                      
{'loss': 0.239, 'grad_norm': 0.6753525733947754, 'learning_rate': 9.598631706386812e-06, 'epoch': 0.65}

 22%|██▏       | 2871/13317 [29:34<1:46:45,  1.63it/s]
 22%|██▏       | 2872/13317 [29:34<1:46:45,  1.63it/s]
                                                      
{'loss': 0.2249, 'grad_norm': 0.5390564203262329, 'learning_rate': 9.5981170452654e-06, 'epoch': 0.65}

 22%|██▏       | 2872/13317 [29:34<1:46:45,  1.63it/s]
 22%|██▏       | 2873/13317 [29:35<1:46:46,  1.63it/s]
                                                      
{'loss': 0.2789, 'grad_norm': 0.5139696002006531, 'learning_rate': 9.597602068204661e-06, 'epoch': 0.65}

 22%|██▏       | 2873/13317 [29:35<1:46:46,  1.63it/s]
 22%|██▏       | 2874/13317 [29:35<1:46:44,  1.63it/s]
                                                      
{'loss': 0.2245, 'grad_norm': 0.520932137966156, 'learning_rate': 9.597086775239981e-06, 'epoch': 0.65}

 22%|██▏       | 2874/13317 [29:36<1:46:44,  1.63it/s]
 22%|██▏       | 2875/13317 [29:36<1:46:44,  1.63it/s]
                                                      
{'loss': 0.2551, 'grad_norm': 0.6543159484863281, 'learning_rate': 9.596571166406764e-06, 'epoch': 0.65}

 22%|██▏       | 2875/13317 [29:36<1:46:44,  1.63it/s]
 22%|██▏       | 2876/13317 [29:37<1:46:47,  1.63it/s]
                                                      
{'loss': 0.2598, 'grad_norm': 0.5982657670974731, 'learning_rate': 9.596055241740438e-06, 'epoch': 0.65}

 22%|██▏       | 2876/13317 [29:37<1:46:47,  1.63it/s]
 22%|██▏       | 2877/13317 [29:37<1:46:46,  1.63it/s]
                                                      
{'loss': 0.2518, 'grad_norm': 0.5481685996055603, 'learning_rate': 9.595539001276453e-06, 'epoch': 0.65}

 22%|██▏       | 2877/13317 [29:37<1:46:46,  1.63it/s]
 22%|██▏       | 2878/13317 [29:38<1:47:10,  1.62it/s]
                                                      
{'loss': 0.1937, 'grad_norm': 0.47290870547294617, 'learning_rate': 9.595022445050281e-06, 'epoch': 0.65}

 22%|██▏       | 2878/13317 [29:38<1:47:10,  1.62it/s]
 22%|██▏       | 2879/13317 [29:39<1:47:07,  1.62it/s]
                                                      
{'loss': 0.2303, 'grad_norm': 0.6119693517684937, 'learning_rate': 9.594505573097415e-06, 'epoch': 0.65}

 22%|██▏       | 2879/13317 [29:39<1:47:07,  1.62it/s]
 22%|██▏       | 2880/13317 [29:39<1:47:00,  1.63it/s]
                                                      
{'loss': 0.2867, 'grad_norm': 0.6107691526412964, 'learning_rate': 9.593988385453368e-06, 'epoch': 0.65}

 22%|██▏       | 2880/13317 [29:39<1:47:00,  1.63it/s]
 22%|██▏       | 2881/13317 [29:40<1:46:55,  1.63it/s]
                                                      
{'loss': 0.2791, 'grad_norm': 0.630048394203186, 'learning_rate': 9.593470882153678e-06, 'epoch': 0.65}

 22%|██▏       | 2881/13317 [29:40<1:46:55,  1.63it/s]
 22%|██▏       | 2882/13317 [29:40<1:46:55,  1.63it/s]
                                                      
{'loss': 0.3127, 'grad_norm': 0.5758545994758606, 'learning_rate': 9.592953063233899e-06, 'epoch': 0.65}

 22%|██▏       | 2882/13317 [29:41<1:46:55,  1.63it/s]
 22%|██▏       | 2883/13317 [29:41<1:46:50,  1.63it/s]
                                                      
{'loss': 0.2403, 'grad_norm': 0.4991023540496826, 'learning_rate': 9.592434928729617e-06, 'epoch': 0.65}

 22%|██▏       | 2883/13317 [29:41<1:46:50,  1.63it/s]
 22%|██▏       | 2884/13317 [29:42<1:46:53,  1.63it/s]
                                                      
{'loss': 0.2935, 'grad_norm': 0.6028522253036499, 'learning_rate': 9.591916478676426e-06, 'epoch': 0.65}

 22%|██▏       | 2884/13317 [29:42<1:46:53,  1.63it/s]
 22%|██▏       | 2885/13317 [29:42<1:46:50,  1.63it/s]
                                                      
{'loss': 0.2464, 'grad_norm': 0.5905839204788208, 'learning_rate': 9.591397713109955e-06, 'epoch': 0.65}

 22%|██▏       | 2885/13317 [29:42<1:46:50,  1.63it/s]
 22%|██▏       | 2886/13317 [29:43<1:47:28,  1.62it/s]
                                                      
{'loss': 0.3054, 'grad_norm': 0.5814077258110046, 'learning_rate': 9.590878632065848e-06, 'epoch': 0.65}

 22%|██▏       | 2886/13317 [29:43<1:47:28,  1.62it/s]
 22%|██▏       | 2887/13317 [29:44<1:47:21,  1.62it/s]
                                                      
{'loss': 0.2434, 'grad_norm': 0.5535728931427002, 'learning_rate': 9.59035923557977e-06, 'epoch': 0.65}

 22%|██▏       | 2887/13317 [29:44<1:47:21,  1.62it/s]
 22%|██▏       | 2888/13317 [29:44<1:47:06,  1.62it/s]
                                                      
{'loss': 0.3256, 'grad_norm': 0.6414884328842163, 'learning_rate': 9.589839523687406e-06, 'epoch': 0.65}

 22%|██▏       | 2888/13317 [29:44<1:47:06,  1.62it/s]
 22%|██▏       | 2889/13317 [29:45<1:47:03,  1.62it/s]
                                                      
{'loss': 0.1991, 'grad_norm': 0.4432491660118103, 'learning_rate': 9.589319496424472e-06, 'epoch': 0.65}

 22%|██▏       | 2889/13317 [29:45<1:47:03,  1.62it/s]
 22%|██▏       | 2890/13317 [29:45<1:46:53,  1.63it/s]
                                                      
{'loss': 0.2658, 'grad_norm': 0.5193043947219849, 'learning_rate': 9.588799153826693e-06, 'epoch': 0.65}

 22%|██▏       | 2890/13317 [29:45<1:46:53,  1.63it/s]
 22%|██▏       | 2891/13317 [29:46<1:46:50,  1.63it/s]
                                                      
{'loss': 0.3793, 'grad_norm': 0.6170049905776978, 'learning_rate': 9.588278495929828e-06, 'epoch': 0.65}

 22%|██▏       | 2891/13317 [29:46<1:46:50,  1.63it/s]
 22%|██▏       | 2892/13317 [29:47<1:46:55,  1.62it/s]
                                                      
{'loss': 0.3043, 'grad_norm': 0.567533016204834, 'learning_rate': 9.587757522769647e-06, 'epoch': 0.65}

 22%|██▏       | 2892/13317 [29:47<1:46:55,  1.62it/s]
 22%|██▏       | 2893/13317 [29:47<1:46:48,  1.63it/s]
                                                      
{'loss': 0.2652, 'grad_norm': 0.48192843794822693, 'learning_rate': 9.587236234381949e-06, 'epoch': 0.65}

 22%|██▏       | 2893/13317 [29:47<1:46:48,  1.63it/s]
 22%|██▏       | 2894/13317 [29:48<1:46:50,  1.63it/s]
                                                      
{'loss': 0.26, 'grad_norm': 0.607914924621582, 'learning_rate': 9.586714630802552e-06, 'epoch': 0.65}

 22%|██▏       | 2894/13317 [29:48<1:46:50,  1.63it/s]
 22%|██▏       | 2895/13317 [29:48<1:46:50,  1.63it/s]
                                                      
{'loss': 0.183, 'grad_norm': 0.415833055973053, 'learning_rate': 9.586192712067293e-06, 'epoch': 0.65}

 22%|██▏       | 2895/13317 [29:49<1:46:50,  1.63it/s]
 22%|██▏       | 2896/13317 [29:49<1:46:49,  1.63it/s]
                                                      
{'loss': 0.2251, 'grad_norm': 0.4783923625946045, 'learning_rate': 9.585670478212037e-06, 'epoch': 0.65}

 22%|██▏       | 2896/13317 [29:49<1:46:49,  1.63it/s]
 22%|██▏       | 2897/13317 [29:50<1:46:54,  1.62it/s]
                                                      
{'loss': 0.3035, 'grad_norm': 0.6740967035293579, 'learning_rate': 9.585147929272665e-06, 'epoch': 0.65}

 22%|██▏       | 2897/13317 [29:50<1:46:54,  1.62it/s]
 22%|██▏       | 2898/13317 [29:50<1:46:47,  1.63it/s]
                                                      
{'loss': 0.2573, 'grad_norm': 0.5595380663871765, 'learning_rate': 9.584625065285081e-06, 'epoch': 0.65}

 22%|██▏       | 2898/13317 [29:50<1:46:47,  1.63it/s]
 22%|██▏       | 2899/13317 [29:51<1:46:52,  1.62it/s]
                                                      
{'loss': 0.1999, 'grad_norm': 0.5068724751472473, 'learning_rate': 9.584101886285213e-06, 'epoch': 0.65}

 22%|██▏       | 2899/13317 [29:51<1:46:52,  1.62it/s]
 22%|██▏       | 2900/13317 [29:51<1:46:43,  1.63it/s]
                                                      
{'loss': 0.3082, 'grad_norm': 0.5736415982246399, 'learning_rate': 9.583578392309007e-06, 'epoch': 0.65}

 22%|██▏       | 2900/13317 [29:52<1:46:43,  1.63it/s]
 22%|██▏       | 2901/13317 [29:52<1:46:46,  1.63it/s]
                                                      
{'loss': 0.3295, 'grad_norm': 0.7178822755813599, 'learning_rate': 9.583054583392435e-06, 'epoch': 0.65}

 22%|██▏       | 2901/13317 [29:52<1:46:46,  1.63it/s]
 22%|██▏       | 2902/13317 [29:53<1:46:50,  1.62it/s]
                                                      
{'loss': 0.2881, 'grad_norm': 0.677006185054779, 'learning_rate': 9.582530459571487e-06, 'epoch': 0.65}

 22%|██▏       | 2902/13317 [29:53<1:46:50,  1.62it/s]
 22%|██▏       | 2903/13317 [29:53<1:46:41,  1.63it/s]
                                                      
{'loss': 0.2378, 'grad_norm': 0.503710925579071, 'learning_rate': 9.582006020882175e-06, 'epoch': 0.65}

 22%|██▏       | 2903/13317 [29:53<1:46:41,  1.63it/s]
 22%|██▏       | 2904/13317 [29:54<1:46:40,  1.63it/s]
                                                      
{'loss': 0.2699, 'grad_norm': 0.5612732172012329, 'learning_rate': 9.581481267360535e-06, 'epoch': 0.65}

 22%|██▏       | 2904/13317 [29:54<1:46:40,  1.63it/s]
 22%|██▏       | 2905/13317 [29:55<1:46:37,  1.63it/s]
                                                      
{'loss': 0.2404, 'grad_norm': 0.5572289228439331, 'learning_rate': 9.580956199042622e-06, 'epoch': 0.65}

 22%|██▏       | 2905/13317 [29:55<1:46:37,  1.63it/s]
 22%|██▏       | 2906/13317 [29:55<1:46:38,  1.63it/s]
                                                      
{'loss': 0.2628, 'grad_norm': 0.542130172252655, 'learning_rate': 9.580430815964515e-06, 'epoch': 0.65}

 22%|██▏       | 2906/13317 [29:55<1:46:38,  1.63it/s]
 22%|██▏       | 2907/13317 [29:56<1:46:42,  1.63it/s]
                                                      
{'loss': 0.3418, 'grad_norm': 0.5960443019866943, 'learning_rate': 9.57990511816231e-06, 'epoch': 0.65}

 22%|██▏       | 2907/13317 [29:56<1:46:42,  1.63it/s]
 22%|██▏       | 2908/13317 [29:56<1:46:37,  1.63it/s]
                                                      
{'loss': 0.2202, 'grad_norm': 0.5239051580429077, 'learning_rate': 9.579379105672135e-06, 'epoch': 0.66}

 22%|██▏       | 2908/13317 [29:57<1:46:37,  1.63it/s]
 22%|██▏       | 2909/13317 [29:57<1:46:40,  1.63it/s]
                                                      
{'loss': 0.2707, 'grad_norm': 0.5573527812957764, 'learning_rate': 9.578852778530126e-06, 'epoch': 0.66}

 22%|██▏       | 2909/13317 [29:57<1:46:40,  1.63it/s]
 22%|██▏       | 2910/13317 [29:58<1:46:35,  1.63it/s]
                                                      
{'loss': 0.2518, 'grad_norm': 0.48840048909187317, 'learning_rate': 9.578326136772449e-06, 'epoch': 0.66}

 22%|██▏       | 2910/13317 [29:58<1:46:35,  1.63it/s]
 22%|██▏       | 2911/13317 [29:58<1:46:39,  1.63it/s]
                                                      
{'loss': 0.287, 'grad_norm': 0.5683078765869141, 'learning_rate': 9.57779918043529e-06, 'epoch': 0.66}

 22%|██▏       | 2911/13317 [29:58<1:46:39,  1.63it/s]
 22%|██▏       | 2912/13317 [29:59<1:46:41,  1.63it/s]
                                                      
{'loss': 0.2232, 'grad_norm': 0.49132803082466125, 'learning_rate': 9.577271909554858e-06, 'epoch': 0.66}

 22%|██▏       | 2912/13317 [29:59<1:46:41,  1.63it/s]
 22%|██▏       | 2913/13317 [29:59<1:46:34,  1.63it/s]
                                                      
{'loss': 0.2761, 'grad_norm': 0.521704375743866, 'learning_rate': 9.57674432416738e-06, 'epoch': 0.66}

 22%|██▏       | 2913/13317 [30:00<1:46:34,  1.63it/s]
 22%|██▏       | 2914/13317 [30:00<1:46:35,  1.63it/s]
                                                      
{'loss': 0.3064, 'grad_norm': 0.5258693695068359, 'learning_rate': 9.576216424309107e-06, 'epoch': 0.66}

 22%|██▏       | 2914/13317 [30:00<1:46:35,  1.63it/s]
 22%|██▏       | 2915/13317 [30:01<1:46:25,  1.63it/s]
                                                      
{'loss': 0.2312, 'grad_norm': 0.5967009663581848, 'learning_rate': 9.575688210016313e-06, 'epoch': 0.66}

 22%|██▏       | 2915/13317 [30:01<1:46:25,  1.63it/s]
 22%|██▏       | 2916/13317 [30:01<1:46:22,  1.63it/s]
                                                      
{'loss': 0.3184, 'grad_norm': 0.7236337065696716, 'learning_rate': 9.57515968132529e-06, 'epoch': 0.66}

 22%|██▏       | 2916/13317 [30:01<1:46:22,  1.63it/s]
 22%|██▏       | 2917/13317 [30:02<1:46:31,  1.63it/s]
                                                      
{'loss': 0.2981, 'grad_norm': 0.5754890441894531, 'learning_rate': 9.574630838272355e-06, 'epoch': 0.66}

 22%|██▏       | 2917/13317 [30:02<1:46:31,  1.63it/s]
 22%|██▏       | 2918/13317 [30:03<1:46:26,  1.63it/s]
                                                      
{'loss': 0.2504, 'grad_norm': 0.5578694343566895, 'learning_rate': 9.574101680893843e-06, 'epoch': 0.66}

 22%|██▏       | 2918/13317 [30:03<1:46:26,  1.63it/s]
 22%|██▏       | 2919/13317 [30:03<1:46:29,  1.63it/s]
                                                      
{'loss': 0.2293, 'grad_norm': 0.5188772678375244, 'learning_rate': 9.573572209226116e-06, 'epoch': 0.66}

 22%|██▏       | 2919/13317 [30:03<1:46:29,  1.63it/s]
 22%|██▏       | 2920/13317 [30:04<1:46:25,  1.63it/s]
                                                      
{'loss': 0.2261, 'grad_norm': 0.471150279045105, 'learning_rate': 9.573042423305552e-06, 'epoch': 0.66}

 22%|██▏       | 2920/13317 [30:04<1:46:25,  1.63it/s]
 22%|██▏       | 2921/13317 [30:04<1:46:25,  1.63it/s]
                                                      
{'loss': 0.199, 'grad_norm': 0.4371517300605774, 'learning_rate': 9.572512323168553e-06, 'epoch': 0.66}

 22%|██▏       | 2921/13317 [30:05<1:46:25,  1.63it/s]
 22%|██▏       | 2922/13317 [30:05<1:46:27,  1.63it/s]
                                                      
{'loss': 0.3179, 'grad_norm': 0.5991093516349792, 'learning_rate': 9.571981908851543e-06, 'epoch': 0.66}

 22%|██▏       | 2922/13317 [30:05<1:46:27,  1.63it/s]
 22%|██▏       | 2923/13317 [30:06<1:46:24,  1.63it/s]
                                                      
{'loss': 0.3009, 'grad_norm': 0.5595633387565613, 'learning_rate': 9.571451180390968e-06, 'epoch': 0.66}

 22%|██▏       | 2923/13317 [30:06<1:46:24,  1.63it/s]
 22%|██▏       | 2924/13317 [30:06<1:46:21,  1.63it/s]
                                                      
{'loss': 0.2498, 'grad_norm': 0.5269356369972229, 'learning_rate': 9.570920137823292e-06, 'epoch': 0.66}

 22%|██▏       | 2924/13317 [30:06<1:46:21,  1.63it/s]
 22%|██▏       | 2925/13317 [30:07<1:46:17,  1.63it/s]
                                                      
{'loss': 0.3031, 'grad_norm': 0.5625948905944824, 'learning_rate': 9.570388781185005e-06, 'epoch': 0.66}

 22%|██▏       | 2925/13317 [30:07<1:46:17,  1.63it/s]
 22%|██▏       | 2926/13317 [30:07<1:46:20,  1.63it/s]
                                                      
{'loss': 0.3242, 'grad_norm': 0.6022029519081116, 'learning_rate': 9.569857110512618e-06, 'epoch': 0.66}

 22%|██▏       | 2926/13317 [30:08<1:46:20,  1.63it/s]
 22%|██▏       | 2927/13317 [30:08<1:46:25,  1.63it/s]
                                                      
{'loss': 0.2245, 'grad_norm': 0.5597080588340759, 'learning_rate': 9.56932512584266e-06, 'epoch': 0.66}

 22%|██▏       | 2927/13317 [30:08<1:46:25,  1.63it/s]
 22%|██▏       | 2928/13317 [30:09<1:46:18,  1.63it/s]
                                                      
{'loss': 0.2434, 'grad_norm': 0.555671751499176, 'learning_rate': 9.568792827211685e-06, 'epoch': 0.66}

 22%|██▏       | 2928/13317 [30:09<1:46:18,  1.63it/s]
 22%|██▏       | 2929/13317 [30:09<1:46:16,  1.63it/s]
                                                      
{'loss': 0.285, 'grad_norm': 0.5327353477478027, 'learning_rate': 9.568260214656269e-06, 'epoch': 0.66}

 22%|██▏       | 2929/13317 [30:09<1:46:16,  1.63it/s]
 22%|██▏       | 2930/13317 [30:10<1:46:15,  1.63it/s]
                                                      
{'loss': 0.32, 'grad_norm': 0.6584289073944092, 'learning_rate': 9.567727288213005e-06, 'epoch': 0.66}

 22%|██▏       | 2930/13317 [30:10<1:46:15,  1.63it/s]
 22%|██▏       | 2931/13317 [30:11<1:46:19,  1.63it/s]
                                                      
{'loss': 0.2513, 'grad_norm': 0.5345426797866821, 'learning_rate': 9.567194047918513e-06, 'epoch': 0.66}

 22%|██▏       | 2931/13317 [30:11<1:46:19,  1.63it/s]
 22%|██▏       | 2932/13317 [30:11<1:46:24,  1.63it/s]
                                                      
{'loss': 0.2374, 'grad_norm': 0.6110538840293884, 'learning_rate': 9.566660493809432e-06, 'epoch': 0.66}

 22%|██▏       | 2932/13317 [30:11<1:46:24,  1.63it/s]
 22%|██▏       | 2933/13317 [30:12<1:46:20,  1.63it/s]
                                                      
{'loss': 0.2915, 'grad_norm': 0.5865962505340576, 'learning_rate': 9.566126625922422e-06, 'epoch': 0.66}

 22%|██▏       | 2933/13317 [30:12<1:46:20,  1.63it/s]
 22%|██▏       | 2934/13317 [30:12<1:46:14,  1.63it/s]
                                                      
{'loss': 0.1742, 'grad_norm': 0.42101842164993286, 'learning_rate': 9.565592444294167e-06, 'epoch': 0.66}

 22%|██▏       | 2934/13317 [30:13<1:46:14,  1.63it/s]
 22%|██▏       | 2935/13317 [30:13<1:46:20,  1.63it/s]
                                                      
{'loss': 0.3672, 'grad_norm': 0.7456029653549194, 'learning_rate': 9.565057948961368e-06, 'epoch': 0.66}

 22%|██▏       | 2935/13317 [30:13<1:46:20,  1.63it/s]
 22%|██▏       | 2936/13317 [30:14<1:46:22,  1.63it/s]
                                                      
{'loss': 0.3018, 'grad_norm': 0.5591388940811157, 'learning_rate': 9.564523139960754e-06, 'epoch': 0.66}

 22%|██▏       | 2936/13317 [30:14<1:46:22,  1.63it/s]
 22%|██▏       | 2937/13317 [30:14<1:46:19,  1.63it/s]
                                                      
{'loss': 0.2664, 'grad_norm': 0.5688878893852234, 'learning_rate': 9.563988017329068e-06, 'epoch': 0.66}

 22%|██▏       | 2937/13317 [30:14<1:46:19,  1.63it/s]
 22%|██▏       | 2938/13317 [30:15<1:46:18,  1.63it/s]
                                                      
{'loss': 0.2401, 'grad_norm': 0.5391114354133606, 'learning_rate': 9.563452581103084e-06, 'epoch': 0.66}

 22%|██▏       | 2938/13317 [30:15<1:46:18,  1.63it/s]
 22%|██▏       | 2939/13317 [30:15<1:46:12,  1.63it/s]
                                                      
{'loss': 0.2405, 'grad_norm': 0.5430123209953308, 'learning_rate': 9.562916831319588e-06, 'epoch': 0.66}

 22%|██▏       | 2939/13317 [30:16<1:46:12,  1.63it/s]
 22%|██▏       | 2940/13317 [30:16<1:46:10,  1.63it/s]
                                                      
{'loss': 0.3195, 'grad_norm': 0.6240359544754028, 'learning_rate': 9.562380768015392e-06, 'epoch': 0.66}

 22%|██▏       | 2940/13317 [30:16<1:46:10,  1.63it/s]
 22%|██▏       | 2941/13317 [30:17<1:46:21,  1.63it/s]
                                                      
{'loss': 0.2308, 'grad_norm': 0.4764147102832794, 'learning_rate': 9.561844391227331e-06, 'epoch': 0.66}

 22%|██▏       | 2941/13317 [30:17<1:46:21,  1.63it/s]
 22%|██▏       | 2942/13317 [30:17<1:46:21,  1.63it/s]
                                                      
{'loss': 0.2173, 'grad_norm': 0.48910877108573914, 'learning_rate': 9.561307700992257e-06, 'epoch': 0.66}

 22%|██▏       | 2942/13317 [30:17<1:46:21,  1.63it/s]
 22%|██▏       | 2943/13317 [30:18<1:46:14,  1.63it/s]
                                                      
{'loss': 0.2762, 'grad_norm': 0.5519199371337891, 'learning_rate': 9.56077069734705e-06, 'epoch': 0.66}

 22%|██▏       | 2943/13317 [30:18<1:46:14,  1.63it/s]
 22%|██▏       | 2944/13317 [30:19<1:46:11,  1.63it/s]
                                                      
{'loss': 0.2633, 'grad_norm': 0.534943699836731, 'learning_rate': 9.560233380328606e-06, 'epoch': 0.66}

 22%|██▏       | 2944/13317 [30:19<1:46:11,  1.63it/s]
 22%|██▏       | 2945/13317 [30:19<1:46:05,  1.63it/s]
                                                      
{'loss': 0.1868, 'grad_norm': 0.44704708456993103, 'learning_rate': 9.559695749973841e-06, 'epoch': 0.66}

 22%|██▏       | 2945/13317 [30:19<1:46:05,  1.63it/s]
 22%|██▏       | 2946/13317 [30:20<1:46:09,  1.63it/s]
                                                      
{'loss': 0.2515, 'grad_norm': 0.5209429860115051, 'learning_rate': 9.559157806319703e-06, 'epoch': 0.66}

 22%|██▏       | 2946/13317 [30:20<1:46:09,  1.63it/s]
 22%|██▏       | 2947/13317 [30:20<1:46:13,  1.63it/s]
                                                      
{'loss': 0.2275, 'grad_norm': 0.5294981002807617, 'learning_rate': 9.558619549403148e-06, 'epoch': 0.66}

 22%|██▏       | 2947/13317 [30:21<1:46:13,  1.63it/s]
 22%|██▏       | 2948/13317 [30:21<1:52:04,  1.54it/s]
                                                      
{'loss': 0.2661, 'grad_norm': 0.49219977855682373, 'learning_rate': 9.558080979261165e-06, 'epoch': 0.66}

 22%|██▏       | 2948/13317 [30:21<1:52:04,  1.54it/s]
 22%|██▏       | 2949/13317 [30:22<1:50:15,  1.57it/s]
                                                      
{'loss': 0.2571, 'grad_norm': 0.5259037017822266, 'learning_rate': 9.557542095930755e-06, 'epoch': 0.66}

 22%|██▏       | 2949/13317 [30:22<1:50:15,  1.57it/s]
 22%|██▏       | 2950/13317 [30:22<1:48:58,  1.59it/s]
                                                      
{'loss': 0.2812, 'grad_norm': 0.5429623126983643, 'learning_rate': 9.557002899448946e-06, 'epoch': 0.66}

 22%|██▏       | 2950/13317 [30:22<1:48:58,  1.59it/s]
 22%|██▏       | 2951/13317 [30:23<1:48:07,  1.60it/s]
                                                      
{'loss': 0.2162, 'grad_norm': 0.502987802028656, 'learning_rate': 9.556463389852789e-06, 'epoch': 0.66}

 22%|██▏       | 2951/13317 [30:23<1:48:07,  1.60it/s]
 22%|██▏       | 2952/13317 [30:24<1:47:27,  1.61it/s]
                                                      
{'loss': 0.3435, 'grad_norm': 0.5928541421890259, 'learning_rate': 9.555923567179352e-06, 'epoch': 0.67}

 22%|██▏       | 2952/13317 [30:24<1:47:27,  1.61it/s]
 22%|██▏       | 2953/13317 [30:24<1:46:58,  1.61it/s]
                                                      
{'loss': 0.2567, 'grad_norm': 0.5142159461975098, 'learning_rate': 9.555383431465727e-06, 'epoch': 0.67}

 22%|██▏       | 2953/13317 [30:24<1:46:58,  1.61it/s]
 22%|██▏       | 2954/13317 [30:25<1:46:38,  1.62it/s]
                                                      
{'loss': 0.2356, 'grad_norm': 0.518549382686615, 'learning_rate': 9.554842982749029e-06, 'epoch': 0.67}

 22%|██▏       | 2954/13317 [30:25<1:46:38,  1.62it/s]
 22%|██▏       | 2955/13317 [30:25<1:46:26,  1.62it/s]
                                                      
{'loss': 0.2693, 'grad_norm': 0.6247689723968506, 'learning_rate': 9.55430222106639e-06, 'epoch': 0.67}

 22%|██▏       | 2955/13317 [30:26<1:46:26,  1.62it/s]
 22%|██▏       | 2956/13317 [30:26<1:46:15,  1.63it/s]
                                                      
{'loss': 0.2145, 'grad_norm': 0.5074813365936279, 'learning_rate': 9.553761146454964e-06, 'epoch': 0.67}

 22%|██▏       | 2956/13317 [30:26<1:46:15,  1.63it/s]
 22%|██▏       | 2957/13317 [30:27<1:46:07,  1.63it/s]
                                                      
{'loss': 0.1886, 'grad_norm': 0.5898149013519287, 'learning_rate': 9.553219758951935e-06, 'epoch': 0.67}

 22%|██▏       | 2957/13317 [30:27<1:46:07,  1.63it/s]
 22%|██▏       | 2958/13317 [30:27<1:46:04,  1.63it/s]
                                                      
{'loss': 0.225, 'grad_norm': 0.5417472124099731, 'learning_rate': 9.552678058594496e-06, 'epoch': 0.67}

 22%|██▏       | 2958/13317 [30:27<1:46:04,  1.63it/s]
 22%|██▏       | 2959/13317 [30:28<1:46:00,  1.63it/s]
                                                      
{'loss': 0.2704, 'grad_norm': 0.5471581220626831, 'learning_rate': 9.55213604541987e-06, 'epoch': 0.67}

 22%|██▏       | 2959/13317 [30:28<1:46:00,  1.63it/s]
 22%|██▏       | 2960/13317 [30:28<1:45:57,  1.63it/s]
                                                      
{'loss': 0.3072, 'grad_norm': 0.6129259467124939, 'learning_rate': 9.551593719465299e-06, 'epoch': 0.67}

 22%|██▏       | 2960/13317 [30:29<1:45:57,  1.63it/s]
 22%|██▏       | 2961/13317 [30:29<1:45:56,  1.63it/s]
                                                      
{'loss': 0.3404, 'grad_norm': 0.7559263110160828, 'learning_rate': 9.551051080768047e-06, 'epoch': 0.67}

 22%|██▏       | 2961/13317 [30:29<1:45:56,  1.63it/s]
 22%|██▏       | 2962/13317 [30:30<1:45:57,  1.63it/s]
                                                      
{'loss': 0.2115, 'grad_norm': 0.48352867364883423, 'learning_rate': 9.550508129365399e-06, 'epoch': 0.67}

 22%|██▏       | 2962/13317 [30:30<1:45:57,  1.63it/s]
 22%|██▏       | 2963/13317 [30:30<1:45:58,  1.63it/s]
                                                      
{'loss': 0.3718, 'grad_norm': 0.6455190181732178, 'learning_rate': 9.549964865294659e-06, 'epoch': 0.67}

 22%|██▏       | 2963/13317 [30:30<1:45:58,  1.63it/s]
 22%|██▏       | 2964/13317 [30:31<1:45:59,  1.63it/s]
                                                      
{'loss': 0.2677, 'grad_norm': 0.556377112865448, 'learning_rate': 9.549421288593157e-06, 'epoch': 0.67}

 22%|██▏       | 2964/13317 [30:31<1:45:59,  1.63it/s]
 22%|██▏       | 2965/13317 [30:32<1:45:55,  1.63it/s]
                                                      
{'loss': 0.3771, 'grad_norm': 0.6347973346710205, 'learning_rate': 9.548877399298245e-06, 'epoch': 0.67}

 22%|██▏       | 2965/13317 [30:32<1:45:55,  1.63it/s]
 22%|██▏       | 2966/13317 [30:32<1:45:52,  1.63it/s]
                                                      
{'loss': 0.3354, 'grad_norm': 0.6411018967628479, 'learning_rate': 9.548333197447288e-06, 'epoch': 0.67}

 22%|██▏       | 2966/13317 [30:32<1:45:52,  1.63it/s]
 22%|██▏       | 2967/13317 [30:33<1:45:51,  1.63it/s]
                                                      
{'loss': 0.2162, 'grad_norm': 0.4777844250202179, 'learning_rate': 9.547788683077685e-06, 'epoch': 0.67}

 22%|██▏       | 2967/13317 [30:33<1:45:51,  1.63it/s]
 22%|██▏       | 2968/13317 [30:33<1:45:50,  1.63it/s]
                                                      
{'loss': 0.2348, 'grad_norm': 0.5066023468971252, 'learning_rate': 9.547243856226844e-06, 'epoch': 0.67}

 22%|██▏       | 2968/13317 [30:34<1:45:50,  1.63it/s]
 22%|██▏       | 2969/13317 [30:34<1:45:49,  1.63it/s]
                                                      
{'loss': 0.3109, 'grad_norm': 0.6040470600128174, 'learning_rate': 9.546698716932206e-06, 'epoch': 0.67}

 22%|██▏       | 2969/13317 [30:34<1:45:49,  1.63it/s]
 22%|██▏       | 2970/13317 [30:35<1:45:49,  1.63it/s]
                                                      
{'loss': 0.2236, 'grad_norm': 0.4924446642398834, 'learning_rate': 9.546153265231222e-06, 'epoch': 0.67}

 22%|██▏       | 2970/13317 [30:35<1:45:49,  1.63it/s]
 22%|██▏       | 2971/13317 [30:35<1:45:46,  1.63it/s]
                                                      
{'loss': 0.3104, 'grad_norm': 0.5956478118896484, 'learning_rate': 9.545607501161375e-06, 'epoch': 0.67}

 22%|██▏       | 2971/13317 [30:35<1:45:46,  1.63it/s]
 22%|██▏       | 2972/13317 [30:36<1:45:47,  1.63it/s]
                                                      
{'loss': 0.2256, 'grad_norm': 0.5013025403022766, 'learning_rate': 9.545061424760163e-06, 'epoch': 0.67}

 22%|██▏       | 2972/13317 [30:36<1:45:47,  1.63it/s]
 22%|██▏       | 2973/13317 [30:36<1:45:44,  1.63it/s]
                                                      
{'loss': 0.2487, 'grad_norm': 0.543213427066803, 'learning_rate': 9.544515036065107e-06, 'epoch': 0.67}

 22%|██▏       | 2973/13317 [30:37<1:45:44,  1.63it/s]
 22%|██▏       | 2974/13317 [30:37<1:45:42,  1.63it/s]
                                                      
{'loss': 0.226, 'grad_norm': 0.4722316563129425, 'learning_rate': 9.543968335113749e-06, 'epoch': 0.67}

 22%|██▏       | 2974/13317 [30:37<1:45:42,  1.63it/s]
 22%|██▏       | 2975/13317 [30:38<1:45:42,  1.63it/s]
                                                      
{'loss': 0.3208, 'grad_norm': 0.5507001876831055, 'learning_rate': 9.543421321943654e-06, 'epoch': 0.67}

 22%|██▏       | 2975/13317 [30:38<1:45:42,  1.63it/s]
 22%|██▏       | 2976/13317 [30:38<1:45:43,  1.63it/s]
                                                      
{'loss': 0.1964, 'grad_norm': 0.5020887851715088, 'learning_rate': 9.542873996592409e-06, 'epoch': 0.67}

 22%|██▏       | 2976/13317 [30:38<1:45:43,  1.63it/s]
 22%|██▏       | 2977/13317 [30:39<1:45:42,  1.63it/s]
                                                      
{'loss': 0.2452, 'grad_norm': 0.5176296234130859, 'learning_rate': 9.542326359097619e-06, 'epoch': 0.67}

 22%|██▏       | 2977/13317 [30:39<1:45:42,  1.63it/s]
 22%|██▏       | 2978/13317 [30:40<1:45:42,  1.63it/s]
                                                      
{'loss': 0.3156, 'grad_norm': 0.5903765559196472, 'learning_rate': 9.541778409496913e-06, 'epoch': 0.67}

 22%|██▏       | 2978/13317 [30:40<1:45:42,  1.63it/s]
 22%|██▏       | 2979/13317 [30:40<1:45:40,  1.63it/s]
                                                      
{'loss': 0.3832, 'grad_norm': 0.6397297978401184, 'learning_rate': 9.541230147827942e-06, 'epoch': 0.67}

 22%|██▏       | 2979/13317 [30:40<1:45:40,  1.63it/s]
 22%|██▏       | 2980/13317 [30:41<1:45:37,  1.63it/s]
                                                      
{'loss': 0.303, 'grad_norm': 0.5127180218696594, 'learning_rate': 9.540681574128376e-06, 'epoch': 0.67}

 22%|██▏       | 2980/13317 [30:41<1:45:37,  1.63it/s]
 22%|██▏       | 2981/13317 [30:41<1:45:42,  1.63it/s]
                                                      
{'loss': 0.2652, 'grad_norm': 0.6018312573432922, 'learning_rate': 9.540132688435907e-06, 'epoch': 0.67}

 22%|██▏       | 2981/13317 [30:41<1:45:42,  1.63it/s]
 22%|██▏       | 2982/13317 [30:42<1:45:41,  1.63it/s]
                                                      
{'loss': 0.236, 'grad_norm': 0.507774293422699, 'learning_rate': 9.539583490788253e-06, 'epoch': 0.67}

 22%|██▏       | 2982/13317 [30:42<1:45:41,  1.63it/s]
 22%|██▏       | 2983/13317 [30:43<1:45:37,  1.63it/s]
                                                      
{'loss': 0.2761, 'grad_norm': 0.6054124236106873, 'learning_rate': 9.539033981223147e-06, 'epoch': 0.67}

 22%|██▏       | 2983/13317 [30:43<1:45:37,  1.63it/s]
 22%|██▏       | 2984/13317 [30:43<1:45:34,  1.63it/s]
                                                      
{'loss': 0.3196, 'grad_norm': 0.6059771180152893, 'learning_rate': 9.538484159778345e-06, 'epoch': 0.67}

 22%|██▏       | 2984/13317 [30:43<1:45:34,  1.63it/s]
 22%|██▏       | 2985/13317 [30:44<1:45:34,  1.63it/s]
                                                      
{'loss': 0.2934, 'grad_norm': 0.5517271757125854, 'learning_rate': 9.537934026491628e-06, 'epoch': 0.67}

 22%|██▏       | 2985/13317 [30:44<1:45:34,  1.63it/s]
 22%|██▏       | 2986/13317 [30:44<1:45:32,  1.63it/s]
                                                      
{'loss': 0.2044, 'grad_norm': 0.5089526176452637, 'learning_rate': 9.537383581400794e-06, 'epoch': 0.67}

 22%|██▏       | 2986/13317 [30:45<1:45:32,  1.63it/s]
 22%|██▏       | 2987/13317 [30:45<1:45:37,  1.63it/s]
                                                      
{'loss': 0.2017, 'grad_norm': 0.4528932571411133, 'learning_rate': 9.536832824543668e-06, 'epoch': 0.67}

 22%|██▏       | 2987/13317 [30:45<1:45:37,  1.63it/s]
 22%|██▏       | 2988/13317 [30:46<1:45:39,  1.63it/s]
                                                      
{'loss': 0.2769, 'grad_norm': 0.5061888694763184, 'learning_rate': 9.536281755958088e-06, 'epoch': 0.67}

 22%|██▏       | 2988/13317 [30:46<1:45:39,  1.63it/s]
 22%|██▏       | 2989/13317 [30:46<1:45:36,  1.63it/s]
                                                      
{'loss': 0.1985, 'grad_norm': 0.460919052362442, 'learning_rate': 9.535730375681919e-06, 'epoch': 0.67}

 22%|██▏       | 2989/13317 [30:46<1:45:36,  1.63it/s]
 22%|██▏       | 2990/13317 [30:47<1:45:35,  1.63it/s]
                                                      
{'loss': 0.2246, 'grad_norm': 0.4971982538700104, 'learning_rate': 9.53517868375305e-06, 'epoch': 0.67}

 22%|██▏       | 2990/13317 [30:47<1:45:35,  1.63it/s]
 22%|██▏       | 2991/13317 [30:47<1:45:35,  1.63it/s]
                                                      
{'loss': 0.1996, 'grad_norm': 0.4735868573188782, 'learning_rate': 9.534626680209386e-06, 'epoch': 0.67}

 22%|██▏       | 2991/13317 [30:48<1:45:35,  1.63it/s]
 22%|██▏       | 2992/13317 [30:48<1:45:36,  1.63it/s]
                                                      
{'loss': 0.213, 'grad_norm': 0.4858831465244293, 'learning_rate': 9.534074365088857e-06, 'epoch': 0.67}

 22%|██▏       | 2992/13317 [30:48<1:45:36,  1.63it/s]
 22%|██▏       | 2993/13317 [30:49<1:45:35,  1.63it/s]
                                                      
{'loss': 0.2411, 'grad_norm': 0.5449579954147339, 'learning_rate': 9.53352173842941e-06, 'epoch': 0.67}

 22%|██▏       | 2993/13317 [30:49<1:45:35,  1.63it/s]
 22%|██▏       | 2994/13317 [30:49<1:45:31,  1.63it/s]
                                                      
{'loss': 0.3255, 'grad_norm': 0.684065580368042, 'learning_rate': 9.532968800269018e-06, 'epoch': 0.67}

 22%|██▏       | 2994/13317 [30:49<1:45:31,  1.63it/s]
 22%|██▏       | 2995/13317 [30:50<1:45:32,  1.63it/s]
                                                      
{'loss': 0.2208, 'grad_norm': 0.5365663766860962, 'learning_rate': 9.532415550645673e-06, 'epoch': 0.67}

 22%|██▏       | 2995/13317 [30:50<1:45:32,  1.63it/s]
 22%|██▏       | 2996/13317 [30:51<1:45:29,  1.63it/s]
                                                      
{'loss': 0.2496, 'grad_norm': 0.5178642868995667, 'learning_rate': 9.53186198959739e-06, 'epoch': 0.67}

 22%|██▏       | 2996/13317 [30:51<1:45:29,  1.63it/s]
 23%|██▎       | 2997/13317 [30:51<1:45:29,  1.63it/s]
                                                      
{'loss': 0.2133, 'grad_norm': 0.49296656250953674, 'learning_rate': 9.531308117162205e-06, 'epoch': 0.68}

 23%|██▎       | 2997/13317 [30:51<1:45:29,  1.63it/s]
 23%|██▎       | 2998/13317 [30:52<1:45:26,  1.63it/s]
                                                      
{'loss': 0.25, 'grad_norm': 0.48577696084976196, 'learning_rate': 9.530753933378175e-06, 'epoch': 0.68}

 23%|██▎       | 2998/13317 [30:52<1:45:26,  1.63it/s]
 23%|██▎       | 2999/13317 [30:52<1:45:25,  1.63it/s]
                                                      
{'loss': 0.2871, 'grad_norm': 0.6201773285865784, 'learning_rate': 9.530199438283375e-06, 'epoch': 0.68}

 23%|██▎       | 2999/13317 [30:53<1:45:25,  1.63it/s]
 23%|██▎       | 3000/13317 [30:53<1:45:32,  1.63it/s]
                                                      
{'loss': 0.2614, 'grad_norm': 0.5896213054656982, 'learning_rate': 9.52964463191591e-06, 'epoch': 0.68}

 23%|██▎       | 3000/13317 [30:53<1:45:32,  1.63it/s]

  0%|          | 0/13 [00:00<?, ?it/s]

 15%|█▌        | 2/13 [00:00<00:00, 13.38it/s]

 31%|███       | 4/13 [00:00<00:01,  8.41it/s]

 38%|███▊      | 5/13 [00:00<00:01,  7.80it/s]

 46%|████▌     | 6/13 [00:00<00:00,  7.43it/s]

 54%|█████▍    | 7/13 [00:00<00:00,  7.18it/s]

 62%|██████▏   | 8/13 [00:01<00:00,  7.02it/s]

 69%|██████▉   | 9/13 [00:01<00:00,  6.91it/s]

 77%|███████▋  | 10/13 [00:01<00:00,  6.84it/s]

 85%|████████▍ | 11/13 [00:01<00:00,  6.78it/s]

 92%|█████████▏| 12/13 [00:01<00:00,  6.74it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]
                                                      

                                               
{'eval_loss': 0.6902517676353455, 'eval_runtime': 1.952, 'eval_samples_per_second': 102.46, 'eval_steps_per_second': 6.66, 'epoch': 0.68}

 23%|██▎       | 3000/13317 [30:55<1:45:32,  1.63it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]

                                               
 23%|██▎       | 3001/13317 [30:56<3:26:24,  1.20s/it]
                                                      
{'loss': 0.2444, 'grad_norm': 0.5545573830604553, 'learning_rate': 9.529089514313894e-06, 'epoch': 0.68}

 23%|██▎       | 3001/13317 [30:56<3:26:24,  1.20s/it]
 23%|██▎       | 3002/13317 [30:56<2:56:08,  1.02s/it]
                                                      
{'loss': 0.2523, 'grad_norm': 0.5102229118347168, 'learning_rate': 9.528534085515476e-06, 'epoch': 0.68}

 23%|██▎       | 3002/13317 [30:56<2:56:08,  1.02s/it]
 23%|██▎       | 3003/13317 [30:57<2:35:00,  1.11it/s]
                                                      
{'loss': 0.2804, 'grad_norm': 0.6426987051963806, 'learning_rate': 9.527978345558818e-06, 'epoch': 0.68}

 23%|██▎       | 3003/13317 [30:57<2:35:00,  1.11it/s]
 23%|██▎       | 3004/13317 [30:57<2:20:11,  1.23it/s]
                                                      
{'loss': 0.2694, 'grad_norm': 0.5421964526176453, 'learning_rate': 9.527422294482103e-06, 'epoch': 0.68}

 23%|██▎       | 3004/13317 [30:58<2:20:11,  1.23it/s]
 23%|██▎       | 3005/13317 [30:58<2:09:44,  1.32it/s]
                                                      
{'loss': 0.2203, 'grad_norm': 0.5450560450553894, 'learning_rate': 9.526865932323542e-06, 'epoch': 0.68}

 23%|██▎       | 3005/13317 [30:58<2:09:44,  1.32it/s]
 23%|██▎       | 3006/13317 [30:59<2:02:23,  1.40it/s]
                                                      
{'loss': 0.3016, 'grad_norm': 0.6023803949356079, 'learning_rate': 9.526309259121358e-06, 'epoch': 0.68}

 23%|██▎       | 3006/13317 [30:59<2:02:23,  1.40it/s]
 23%|██▎       | 3007/13317 [30:59<1:57:20,  1.46it/s]
                                                      
{'loss': 0.31, 'grad_norm': 0.6092237830162048, 'learning_rate': 9.525752274913803e-06, 'epoch': 0.68}

 23%|██▎       | 3007/13317 [30:59<1:57:20,  1.46it/s]
 23%|██▎       | 3008/13317 [31:00<1:53:46,  1.51it/s]
                                                      
{'loss': 0.2953, 'grad_norm': 0.5959012508392334, 'learning_rate': 9.525194979739146e-06, 'epoch': 0.68}

 23%|██▎       | 3008/13317 [31:00<1:53:46,  1.51it/s]
 23%|██▎       | 3009/13317 [31:00<1:51:13,  1.54it/s]
                                                      
{'loss': 0.2218, 'grad_norm': 0.45957815647125244, 'learning_rate': 9.524637373635682e-06, 'epoch': 0.68}

 23%|██▎       | 3009/13317 [31:01<1:51:13,  1.54it/s]
 23%|██▎       | 3010/13317 [31:01<1:49:22,  1.57it/s]
                                                      
{'loss': 0.2807, 'grad_norm': 0.5752333402633667, 'learning_rate': 9.524079456641723e-06, 'epoch': 0.68}

 23%|██▎       | 3010/13317 [31:01<1:49:22,  1.57it/s]
 23%|██▎       | 3011/13317 [31:02<1:48:13,  1.59it/s]
                                                      
{'loss': 0.3079, 'grad_norm': 0.6698651909828186, 'learning_rate': 9.523521228795602e-06, 'epoch': 0.68}

 23%|██▎       | 3011/13317 [31:02<1:48:13,  1.59it/s]
 23%|██▎       | 3012/13317 [31:02<1:47:21,  1.60it/s]
                                                      
{'loss': 0.3578, 'grad_norm': 0.6849905252456665, 'learning_rate': 9.522962690135679e-06, 'epoch': 0.68}

 23%|██▎       | 3012/13317 [31:02<1:47:21,  1.60it/s]
 23%|██▎       | 3013/13317 [31:03<1:46:43,  1.61it/s]
                                                      
{'loss': 0.2341, 'grad_norm': 0.5167815685272217, 'learning_rate': 9.522403840700328e-06, 'epoch': 0.68}

 23%|██▎       | 3013/13317 [31:03<1:46:43,  1.61it/s]
 23%|██▎       | 3014/13317 [31:04<1:46:19,  1.62it/s]
                                                      
{'loss': 0.2319, 'grad_norm': 0.5100425481796265, 'learning_rate': 9.521844680527948e-06, 'epoch': 0.68}

 23%|██▎       | 3014/13317 [31:04<1:46:19,  1.62it/s]
 23%|██▎       | 3015/13317 [31:04<1:46:04,  1.62it/s]
                                                      
{'loss': 0.266, 'grad_norm': 0.6135210394859314, 'learning_rate': 9.521285209656964e-06, 'epoch': 0.68}

 23%|██▎       | 3015/13317 [31:04<1:46:04,  1.62it/s]
 23%|██▎       | 3016/13317 [31:05<1:45:49,  1.62it/s]
                                                      
{'loss': 0.3097, 'grad_norm': 0.6482551693916321, 'learning_rate': 9.52072542812581e-06, 'epoch': 0.68}

 23%|██▎       | 3016/13317 [31:05<1:45:49,  1.62it/s]
 23%|██▎       | 3017/13317 [31:05<1:45:39,  1.62it/s]
                                                      
{'loss': 0.1891, 'grad_norm': 0.4736221134662628, 'learning_rate': 9.520165335972955e-06, 'epoch': 0.68}

 23%|██▎       | 3017/13317 [31:06<1:45:39,  1.62it/s]
 23%|██▎       | 3018/13317 [31:06<1:45:30,  1.63it/s]
                                                      
{'loss': 0.2685, 'grad_norm': 0.5662933588027954, 'learning_rate': 9.51960493323688e-06, 'epoch': 0.68}

 23%|██▎       | 3018/13317 [31:06<1:45:30,  1.63it/s]
 23%|██▎       | 3019/13317 [31:07<1:45:26,  1.63it/s]
                                                      
{'loss': 0.2815, 'grad_norm': 0.5691407918930054, 'learning_rate': 9.519044219956093e-06, 'epoch': 0.68}

 23%|██▎       | 3019/13317 [31:07<1:45:26,  1.63it/s]
 23%|██▎       | 3020/13317 [31:07<1:45:21,  1.63it/s]
                                                      
{'loss': 0.2072, 'grad_norm': 0.5024381279945374, 'learning_rate': 9.51848319616912e-06, 'epoch': 0.68}

 23%|██▎       | 3020/13317 [31:07<1:45:21,  1.63it/s]
 23%|██▎       | 3021/13317 [31:08<1:45:19,  1.63it/s]
                                                      
{'loss': 0.2541, 'grad_norm': 0.5070144534111023, 'learning_rate': 9.517921861914506e-06, 'epoch': 0.68}

 23%|██▎       | 3021/13317 [31:08<1:45:19,  1.63it/s]
 23%|██▎       | 3022/13317 [31:08<1:45:18,  1.63it/s]
                                                      
{'loss': 0.2116, 'grad_norm': 0.521759033203125, 'learning_rate': 9.517360217230827e-06, 'epoch': 0.68}

 23%|██▎       | 3022/13317 [31:09<1:45:18,  1.63it/s]
 23%|██▎       | 3023/13317 [31:09<1:45:14,  1.63it/s]
                                                      
{'loss': 0.2664, 'grad_norm': 0.5180450677871704, 'learning_rate': 9.516798262156668e-06, 'epoch': 0.68}

 23%|██▎       | 3023/13317 [31:09<1:45:14,  1.63it/s]
 23%|██▎       | 3024/13317 [31:10<1:45:14,  1.63it/s]
                                                      
{'loss': 0.2013, 'grad_norm': 0.45313316583633423, 'learning_rate': 9.516235996730645e-06, 'epoch': 0.68}

 23%|██▎       | 3024/13317 [31:10<1:45:14,  1.63it/s]
 23%|██▎       | 3025/13317 [31:10<1:45:11,  1.63it/s]
                                                      
{'loss': 0.2785, 'grad_norm': 0.5339494347572327, 'learning_rate': 9.515673420991391e-06, 'epoch': 0.68}

 23%|██▎       | 3025/13317 [31:10<1:45:11,  1.63it/s]
 23%|██▎       | 3026/13317 [31:11<1:45:12,  1.63it/s]
                                                      
{'loss': 0.1951, 'grad_norm': 0.4825515151023865, 'learning_rate': 9.515110534977558e-06, 'epoch': 0.68}

 23%|██▎       | 3026/13317 [31:11<1:45:12,  1.63it/s]
 23%|██▎       | 3027/13317 [31:12<1:45:11,  1.63it/s]
                                                      
{'loss': 0.2478, 'grad_norm': 0.5814910531044006, 'learning_rate': 9.514547338727827e-06, 'epoch': 0.68}

 23%|██▎       | 3027/13317 [31:12<1:45:11,  1.63it/s]
 23%|██▎       | 3028/13317 [31:12<1:45:08,  1.63it/s]
                                                      
{'loss': 0.2503, 'grad_norm': 0.6042575240135193, 'learning_rate': 9.513983832280892e-06, 'epoch': 0.68}

 23%|██▎       | 3028/13317 [31:12<1:45:08,  1.63it/s]
 23%|██▎       | 3029/13317 [31:13<1:45:08,  1.63it/s]
                                                      
{'loss': 0.2304, 'grad_norm': 0.5379692912101746, 'learning_rate': 9.513420015675474e-06, 'epoch': 0.68}

 23%|██▎       | 3029/13317 [31:13<1:45:08,  1.63it/s]
 23%|██▎       | 3030/13317 [31:13<1:45:07,  1.63it/s]
                                                      
{'loss': 0.3294, 'grad_norm': 0.6084476709365845, 'learning_rate': 9.51285588895031e-06, 'epoch': 0.68}

 23%|██▎       | 3030/13317 [31:13<1:45:07,  1.63it/s]
 23%|██▎       | 3031/13317 [31:14<1:45:08,  1.63it/s]
                                                      
{'loss': 0.2824, 'grad_norm': 0.5371868014335632, 'learning_rate': 9.512291452144165e-06, 'epoch': 0.68}

 23%|██▎       | 3031/13317 [31:14<1:45:08,  1.63it/s]
 23%|██▎       | 3032/13317 [31:15<1:45:09,  1.63it/s]
                                                      
{'loss': 0.2675, 'grad_norm': 0.560036838054657, 'learning_rate': 9.51172670529582e-06, 'epoch': 0.68}

 23%|██▎       | 3032/13317 [31:15<1:45:09,  1.63it/s]
 23%|██▎       | 3033/13317 [31:15<1:45:08,  1.63it/s]
                                                      
{'loss': 0.2682, 'grad_norm': 0.5276099443435669, 'learning_rate': 9.511161648444079e-06, 'epoch': 0.68}

 23%|██▎       | 3033/13317 [31:15<1:45:08,  1.63it/s]
 23%|██▎       | 3034/13317 [31:16<1:45:09,  1.63it/s]
                                                      
{'loss': 0.1764, 'grad_norm': 0.4120045602321625, 'learning_rate': 9.510596281627768e-06, 'epoch': 0.68}

 23%|██▎       | 3034/13317 [31:16<1:45:09,  1.63it/s]
 23%|██▎       | 3035/13317 [31:16<1:45:07,  1.63it/s]
                                                      
{'loss': 0.286, 'grad_norm': 0.5482808947563171, 'learning_rate': 9.510030604885732e-06, 'epoch': 0.68}

 23%|██▎       | 3035/13317 [31:17<1:45:07,  1.63it/s]
 23%|██▎       | 3036/13317 [31:17<1:45:06,  1.63it/s]
                                                      
{'loss': 0.2593, 'grad_norm': 0.5978070497512817, 'learning_rate': 9.509464618256844e-06, 'epoch': 0.68}

 23%|██▎       | 3036/13317 [31:17<1:45:06,  1.63it/s]
 23%|██▎       | 3037/13317 [31:18<1:45:05,  1.63it/s]
                                                      
{'loss': 0.2434, 'grad_norm': 0.5626807808876038, 'learning_rate': 9.508898321779988e-06, 'epoch': 0.68}

 23%|██▎       | 3037/13317 [31:18<1:45:05,  1.63it/s]
 23%|██▎       | 3038/13317 [31:18<1:45:03,  1.63it/s]
                                                      
{'loss': 0.2857, 'grad_norm': 0.6214340329170227, 'learning_rate': 9.508331715494075e-06, 'epoch': 0.68}

 23%|██▎       | 3038/13317 [31:18<1:45:03,  1.63it/s]
 23%|██▎       | 3039/13317 [31:19<1:45:00,  1.63it/s]
                                                      
{'loss': 0.2533, 'grad_norm': 0.5570645928382874, 'learning_rate': 9.50776479943804e-06, 'epoch': 0.68}

 23%|██▎       | 3039/13317 [31:19<1:45:00,  1.63it/s]
 23%|██▎       | 3040/13317 [31:19<1:45:02,  1.63it/s]
                                                      
{'loss': 0.2284, 'grad_norm': 0.5002202391624451, 'learning_rate': 9.507197573650832e-06, 'epoch': 0.68}

 23%|██▎       | 3040/13317 [31:20<1:45:02,  1.63it/s]
 23%|██▎       | 3041/13317 [31:20<1:45:03,  1.63it/s]
                                                      
{'loss': 0.2348, 'grad_norm': 0.5489780306816101, 'learning_rate': 9.50663003817143e-06, 'epoch': 0.69}

 23%|██▎       | 3041/13317 [31:20<1:45:03,  1.63it/s]
 23%|██▎       | 3042/13317 [31:21<1:45:03,  1.63it/s]
                                                      
{'loss': 0.2461, 'grad_norm': 0.5778727531433105, 'learning_rate': 9.506062193038827e-06, 'epoch': 0.69}

 23%|██▎       | 3042/13317 [31:21<1:45:03,  1.63it/s]
 23%|██▎       | 3043/13317 [31:21<1:45:03,  1.63it/s]
                                                      
{'loss': 0.2163, 'grad_norm': 0.4971732497215271, 'learning_rate': 9.50549403829204e-06, 'epoch': 0.69}

 23%|██▎       | 3043/13317 [31:21<1:45:03,  1.63it/s]
 23%|██▎       | 3044/13317 [31:22<1:45:00,  1.63it/s]
                                                      
{'loss': 0.2988, 'grad_norm': 0.63885498046875, 'learning_rate': 9.504925573970107e-06, 'epoch': 0.69}

 23%|██▎       | 3044/13317 [31:22<1:45:00,  1.63it/s]
 23%|██▎       | 3045/13317 [31:23<1:45:01,  1.63it/s]
                                                      
{'loss': 0.225, 'grad_norm': 0.4692581295967102, 'learning_rate': 9.504356800112089e-06, 'epoch': 0.69}

 23%|██▎       | 3045/13317 [31:23<1:45:01,  1.63it/s]
 23%|██▎       | 3046/13317 [31:23<1:45:06,  1.63it/s]
                                                      
{'loss': 0.278, 'grad_norm': 0.5622608065605164, 'learning_rate': 9.503787716757066e-06, 'epoch': 0.69}

 23%|██▎       | 3046/13317 [31:23<1:45:06,  1.63it/s]
 23%|██▎       | 3047/13317 [31:24<1:45:01,  1.63it/s]
                                                      
{'loss': 0.2556, 'grad_norm': 0.5631139874458313, 'learning_rate': 9.503218323944139e-06, 'epoch': 0.69}

 23%|██▎       | 3047/13317 [31:24<1:45:01,  1.63it/s]
 23%|██▎       | 3048/13317 [31:24<1:45:01,  1.63it/s]
                                                      
{'loss': 0.211, 'grad_norm': 0.5363587141036987, 'learning_rate': 9.502648621712433e-06, 'epoch': 0.69}

 23%|██▎       | 3048/13317 [31:25<1:45:01,  1.63it/s]
 23%|██▎       | 3049/13317 [31:25<1:45:00,  1.63it/s]
                                                      
{'loss': 0.2674, 'grad_norm': 0.5529927015304565, 'learning_rate': 9.502078610101093e-06, 'epoch': 0.69}

 23%|██▎       | 3049/13317 [31:25<1:45:00,  1.63it/s]
 23%|██▎       | 3050/13317 [31:26<1:47:41,  1.59it/s]
                                                      
{'loss': 0.2475, 'grad_norm': 0.5196377038955688, 'learning_rate': 9.501508289149281e-06, 'epoch': 0.69}

 23%|██▎       | 3050/13317 [31:26<1:47:41,  1.59it/s]
 23%|██▎       | 3051/13317 [31:26<1:46:51,  1.60it/s]
                                                      
{'loss': 0.2201, 'grad_norm': 0.49512624740600586, 'learning_rate': 9.50093765889619e-06, 'epoch': 0.69}

 23%|██▎       | 3051/13317 [31:26<1:46:51,  1.60it/s]
 23%|██▎       | 3052/13317 [31:27<1:46:29,  1.61it/s]
                                                      
{'loss': 0.3008, 'grad_norm': 0.5944926738739014, 'learning_rate': 9.500366719381023e-06, 'epoch': 0.69}

 23%|██▎       | 3052/13317 [31:27<1:46:29,  1.61it/s]
 23%|██▎       | 3053/13317 [31:28<1:46:00,  1.61it/s]
                                                      
{'loss': 0.2439, 'grad_norm': 0.5378848314285278, 'learning_rate': 9.499795470643013e-06, 'epoch': 0.69}

 23%|██▎       | 3053/13317 [31:28<1:46:00,  1.61it/s]
 23%|██▎       | 3054/13317 [31:28<1:45:38,  1.62it/s]
                                                      
{'loss': 0.3545, 'grad_norm': 0.5916624069213867, 'learning_rate': 9.49922391272141e-06, 'epoch': 0.69}

 23%|██▎       | 3054/13317 [31:28<1:45:38,  1.62it/s]
 23%|██▎       | 3055/13317 [31:29<1:45:43,  1.62it/s]
                                                      
{'loss': 0.2566, 'grad_norm': 0.5233245491981506, 'learning_rate': 9.498652045655484e-06, 'epoch': 0.69}

 23%|██▎       | 3055/13317 [31:29<1:45:43,  1.62it/s]
 23%|██▎       | 3056/13317 [31:29<1:45:28,  1.62it/s]
                                                      
{'loss': 0.2414, 'grad_norm': 0.4975931942462921, 'learning_rate': 9.498079869484529e-06, 'epoch': 0.69}

 23%|██▎       | 3056/13317 [31:30<1:45:28,  1.62it/s]
 23%|██▎       | 3057/13317 [31:30<1:45:23,  1.62it/s]
                                                      
{'loss': 0.2796, 'grad_norm': 0.5733692646026611, 'learning_rate': 9.497507384247863e-06, 'epoch': 0.69}

 23%|██▎       | 3057/13317 [31:30<1:45:23,  1.62it/s]
 23%|██▎       | 3058/13317 [31:31<1:45:12,  1.63it/s]
                                                      
{'loss': 0.2508, 'grad_norm': 0.5498244762420654, 'learning_rate': 9.49693458998482e-06, 'epoch': 0.69}

 23%|██▎       | 3058/13317 [31:31<1:45:12,  1.63it/s]
 23%|██▎       | 3059/13317 [31:31<1:45:13,  1.62it/s]
                                                      
{'loss': 0.289, 'grad_norm': 0.6125621795654297, 'learning_rate': 9.496361486734753e-06, 'epoch': 0.69}

 23%|██▎       | 3059/13317 [31:31<1:45:13,  1.62it/s]
 23%|██▎       | 3060/13317 [31:32<1:45:15,  1.62it/s]
                                                      
{'loss': 0.3168, 'grad_norm': 0.5898267030715942, 'learning_rate': 9.495788074537045e-06, 'epoch': 0.69}

 23%|██▎       | 3060/13317 [31:32<1:45:15,  1.62it/s]
 23%|██▎       | 3061/13317 [31:32<1:45:09,  1.63it/s]
                                                      
{'loss': 0.2282, 'grad_norm': 0.4295859634876251, 'learning_rate': 9.495214353431094e-06, 'epoch': 0.69}

 23%|██▎       | 3061/13317 [31:33<1:45:09,  1.63it/s]
 23%|██▎       | 3062/13317 [31:33<1:45:03,  1.63it/s]
                                                      
{'loss': 0.3262, 'grad_norm': 0.5688067078590393, 'learning_rate': 9.494640323456322e-06, 'epoch': 0.69}

 23%|██▎       | 3062/13317 [31:33<1:45:03,  1.63it/s]
 23%|██▎       | 3063/13317 [31:34<1:44:57,  1.63it/s]
                                                      
{'loss': 0.293, 'grad_norm': 0.5984587669372559, 'learning_rate': 9.49406598465217e-06, 'epoch': 0.69}

 23%|██▎       | 3063/13317 [31:34<1:44:57,  1.63it/s]
 23%|██▎       | 3064/13317 [31:34<1:44:54,  1.63it/s]
                                                      
{'loss': 0.3083, 'grad_norm': 0.4955705404281616, 'learning_rate': 9.4934913370581e-06, 'epoch': 0.69}

 23%|██▎       | 3064/13317 [31:34<1:44:54,  1.63it/s]
 23%|██▎       | 3065/13317 [31:35<1:44:57,  1.63it/s]
                                                      
{'loss': 0.1793, 'grad_norm': 0.43268024921417236, 'learning_rate': 9.492916380713597e-06, 'epoch': 0.69}

 23%|██▎       | 3065/13317 [31:35<1:44:57,  1.63it/s]
 23%|██▎       | 3066/13317 [31:36<1:44:56,  1.63it/s]
                                                      
{'loss': 0.3896, 'grad_norm': 0.5027807354927063, 'learning_rate': 9.492341115658167e-06, 'epoch': 0.69}

 23%|██▎       | 3066/13317 [31:36<1:44:56,  1.63it/s]
 23%|██▎       | 3067/13317 [31:36<1:44:53,  1.63it/s]
                                                      
{'loss': 0.2767, 'grad_norm': 0.524669349193573, 'learning_rate': 9.491765541931338e-06, 'epoch': 0.69}

 23%|██▎       | 3067/13317 [31:36<1:44:53,  1.63it/s]
 23%|██▎       | 3068/13317 [31:37<1:44:50,  1.63it/s]
                                                      
{'loss': 0.247, 'grad_norm': 0.5610320568084717, 'learning_rate': 9.491189659572655e-06, 'epoch': 0.69}

 23%|██▎       | 3068/13317 [31:37<1:44:50,  1.63it/s]
 23%|██▎       | 3069/13317 [31:37<1:44:44,  1.63it/s]
                                                      
{'loss': 0.3018, 'grad_norm': 0.5272584557533264, 'learning_rate': 9.490613468621689e-06, 'epoch': 0.69}

 23%|██▎       | 3069/13317 [31:37<1:44:44,  1.63it/s]
 23%|██▎       | 3070/13317 [31:38<1:44:43,  1.63it/s]
                                                      
{'loss': 0.3251, 'grad_norm': 0.6479690670967102, 'learning_rate': 9.490036969118031e-06, 'epoch': 0.69}

 23%|██▎       | 3070/13317 [31:38<1:44:43,  1.63it/s]
 23%|██▎       | 3071/13317 [31:39<1:44:44,  1.63it/s]
                                                      
{'loss': 0.1685, 'grad_norm': 0.5285512208938599, 'learning_rate': 9.489460161101291e-06, 'epoch': 0.69}

 23%|██▎       | 3071/13317 [31:39<1:44:44,  1.63it/s]
 23%|██▎       | 3072/13317 [31:39<1:44:44,  1.63it/s]
                                                      
{'loss': 0.2872, 'grad_norm': 0.5756537318229675, 'learning_rate': 9.488883044611103e-06, 'epoch': 0.69}

 23%|██▎       | 3072/13317 [31:39<1:44:44,  1.63it/s]
 23%|██▎       | 3073/13317 [31:40<1:44:45,  1.63it/s]
                                                      
{'loss': 0.2335, 'grad_norm': 0.5752984285354614, 'learning_rate': 9.488305619687121e-06, 'epoch': 0.69}

 23%|██▎       | 3073/13317 [31:40<1:44:45,  1.63it/s]
 23%|██▎       | 3074/13317 [31:40<1:44:46,  1.63it/s]
                                                      
{'loss': 0.1675, 'grad_norm': 0.3932393789291382, 'learning_rate': 9.487727886369021e-06, 'epoch': 0.69}

 23%|██▎       | 3074/13317 [31:41<1:44:46,  1.63it/s]
 23%|██▎       | 3075/13317 [31:41<1:44:45,  1.63it/s]
                                                      
{'loss': 0.2773, 'grad_norm': 0.5963214039802551, 'learning_rate': 9.487149844696498e-06, 'epoch': 0.69}

 23%|██▎       | 3075/13317 [31:41<1:44:45,  1.63it/s]
 23%|██▎       | 3076/13317 [31:42<1:44:44,  1.63it/s]
                                                      
{'loss': 0.3975, 'grad_norm': 0.6439527869224548, 'learning_rate': 9.48657149470927e-06, 'epoch': 0.69}

 23%|██▎       | 3076/13317 [31:42<1:44:44,  1.63it/s]
 23%|██▎       | 3077/13317 [31:42<1:44:42,  1.63it/s]
                                                      
{'loss': 0.1993, 'grad_norm': 0.49417993426322937, 'learning_rate': 9.485992836447075e-06, 'epoch': 0.69}

 23%|██▎       | 3077/13317 [31:42<1:44:42,  1.63it/s]
 23%|██▎       | 3078/13317 [31:43<1:44:41,  1.63it/s]
                                                      
{'loss': 0.2514, 'grad_norm': 0.641754686832428, 'learning_rate': 9.485413869949674e-06, 'epoch': 0.69}

 23%|██▎       | 3078/13317 [31:43<1:44:41,  1.63it/s]
 23%|██▎       | 3079/13317 [31:43<1:44:37,  1.63it/s]
                                                      
{'loss': 0.286, 'grad_norm': 0.545259952545166, 'learning_rate': 9.484834595256848e-06, 'epoch': 0.69}

 23%|██▎       | 3079/13317 [31:44<1:44:37,  1.63it/s]
 23%|██▎       | 3080/13317 [31:44<1:44:38,  1.63it/s]
                                                      
{'loss': 0.2497, 'grad_norm': 0.5640097856521606, 'learning_rate': 9.4842550124084e-06, 'epoch': 0.69}

 23%|██▎       | 3080/13317 [31:44<1:44:38,  1.63it/s]
 23%|██▎       | 3081/13317 [31:45<1:44:36,  1.63it/s]
                                                      
{'loss': 0.1908, 'grad_norm': 0.4476217031478882, 'learning_rate': 9.48367512144415e-06, 'epoch': 0.69}

 23%|██▎       | 3081/13317 [31:45<1:44:36,  1.63it/s]
 23%|██▎       | 3082/13317 [31:45<1:44:37,  1.63it/s]
                                                      
{'loss': 0.2914, 'grad_norm': 0.6114400029182434, 'learning_rate': 9.483094922403947e-06, 'epoch': 0.69}

 23%|██▎       | 3082/13317 [31:45<1:44:37,  1.63it/s]
 23%|██▎       | 3083/13317 [31:46<1:44:37,  1.63it/s]
                                                      
{'loss': 0.3076, 'grad_norm': 0.586969256401062, 'learning_rate': 9.482514415327655e-06, 'epoch': 0.69}

 23%|██▎       | 3083/13317 [31:46<1:44:37,  1.63it/s]
 23%|██▎       | 3084/13317 [31:47<1:44:36,  1.63it/s]
                                                      
{'loss': 0.2358, 'grad_norm': 0.539442241191864, 'learning_rate': 9.481933600255162e-06, 'epoch': 0.69}

 23%|██▎       | 3084/13317 [31:47<1:44:36,  1.63it/s]
 23%|██▎       | 3085/13317 [31:47<1:44:32,  1.63it/s]
                                                      
{'loss': 0.2452, 'grad_norm': 0.5334469676017761, 'learning_rate': 9.481352477226373e-06, 'epoch': 0.69}

 23%|██▎       | 3085/13317 [31:47<1:44:32,  1.63it/s]
 23%|██▎       | 3086/13317 [31:48<1:44:36,  1.63it/s]
                                                      
{'loss': 0.3806, 'grad_norm': 0.6643934845924377, 'learning_rate': 9.480771046281221e-06, 'epoch': 0.7}

 23%|██▎       | 3086/13317 [31:48<1:44:36,  1.63it/s]
 23%|██▎       | 3087/13317 [31:48<1:44:36,  1.63it/s]
                                                      
{'loss': 0.284, 'grad_norm': 0.5407428741455078, 'learning_rate': 9.480189307459656e-06, 'epoch': 0.7}

 23%|██▎       | 3087/13317 [31:49<1:44:36,  1.63it/s]
 23%|██▎       | 3088/13317 [31:49<1:44:39,  1.63it/s]
                                                      
{'loss': 0.2529, 'grad_norm': 0.5489826798439026, 'learning_rate': 9.479607260801645e-06, 'epoch': 0.7}

 23%|██▎       | 3088/13317 [31:49<1:44:39,  1.63it/s]
 23%|██▎       | 3089/13317 [31:50<1:44:40,  1.63it/s]
                                                      
{'loss': 0.2575, 'grad_norm': 0.5311387181282043, 'learning_rate': 9.479024906347188e-06, 'epoch': 0.7}

 23%|██▎       | 3089/13317 [31:50<1:44:40,  1.63it/s]
 23%|██▎       | 3090/13317 [31:50<1:44:38,  1.63it/s]
                                                      
{'loss': 0.3121, 'grad_norm': 0.5440350770950317, 'learning_rate': 9.478442244136292e-06, 'epoch': 0.7}

 23%|██▎       | 3090/13317 [31:50<1:44:38,  1.63it/s]
 23%|██▎       | 3091/13317 [31:51<1:44:45,  1.63it/s]
                                                      
{'loss': 0.2211, 'grad_norm': 0.45717883110046387, 'learning_rate': 9.477859274208998e-06, 'epoch': 0.7}

 23%|██▎       | 3091/13317 [31:51<1:44:45,  1.63it/s]
 23%|██▎       | 3092/13317 [31:51<1:44:41,  1.63it/s]
                                                      
{'loss': 0.2499, 'grad_norm': 0.5364225506782532, 'learning_rate': 9.477275996605359e-06, 'epoch': 0.7}

 23%|██▎       | 3092/13317 [31:52<1:44:41,  1.63it/s]
 23%|██▎       | 3093/13317 [31:52<1:44:38,  1.63it/s]
                                                      
{'loss': 0.1892, 'grad_norm': 0.4699312448501587, 'learning_rate': 9.47669241136545e-06, 'epoch': 0.7}

 23%|██▎       | 3093/13317 [31:52<1:44:38,  1.63it/s]
 23%|██▎       | 3094/13317 [31:53<1:45:05,  1.62it/s]
                                                      
{'loss': 0.2498, 'grad_norm': 0.6016184091567993, 'learning_rate': 9.476108518529377e-06, 'epoch': 0.7}

 23%|██▎       | 3094/13317 [31:53<1:45:05,  1.62it/s]
 23%|██▎       | 3095/13317 [31:53<1:44:55,  1.62it/s]
                                                      
{'loss': 0.3067, 'grad_norm': 0.5616608262062073, 'learning_rate': 9.475524318137252e-06, 'epoch': 0.7}

 23%|██▎       | 3095/13317 [31:53<1:44:55,  1.62it/s]
 23%|██▎       | 3096/13317 [31:54<1:44:44,  1.63it/s]
                                                      
{'loss': 0.2325, 'grad_norm': 0.5313904881477356, 'learning_rate': 9.47493981022922e-06, 'epoch': 0.7}

 23%|██▎       | 3096/13317 [31:54<1:44:44,  1.63it/s]
 23%|██▎       | 3097/13317 [31:55<1:44:40,  1.63it/s]
                                                      
{'loss': 0.2032, 'grad_norm': 0.5117576718330383, 'learning_rate': 9.47435499484544e-06, 'epoch': 0.7}

 23%|██▎       | 3097/13317 [31:55<1:44:40,  1.63it/s]
 23%|██▎       | 3098/13317 [31:55<1:44:39,  1.63it/s]
                                                      
{'loss': 0.2533, 'grad_norm': 0.5295073390007019, 'learning_rate': 9.4737698720261e-06, 'epoch': 0.7}

 23%|██▎       | 3098/13317 [31:55<1:44:39,  1.63it/s]
 23%|██▎       | 3099/13317 [31:56<1:44:34,  1.63it/s]
                                                      
{'loss': 0.2888, 'grad_norm': 0.5460037589073181, 'learning_rate': 9.473184441811397e-06, 'epoch': 0.7}

 23%|██▎       | 3099/13317 [31:56<1:44:34,  1.63it/s]
 23%|██▎       | 3100/13317 [31:56<1:44:31,  1.63it/s]
                                                      
{'loss': 0.2981, 'grad_norm': 0.6712568998336792, 'learning_rate': 9.472598704241562e-06, 'epoch': 0.7}

 23%|██▎       | 3100/13317 [31:57<1:44:31,  1.63it/s]
 23%|██▎       | 3101/13317 [31:57<1:44:27,  1.63it/s]
                                                      
{'loss': 0.3073, 'grad_norm': 0.4781532287597656, 'learning_rate': 9.47201265935684e-06, 'epoch': 0.7}

 23%|██▎       | 3101/13317 [31:57<1:44:27,  1.63it/s]
 23%|██▎       | 3102/13317 [31:58<1:44:32,  1.63it/s]
                                                      
{'loss': 0.2663, 'grad_norm': 0.5577643513679504, 'learning_rate': 9.471426307197496e-06, 'epoch': 0.7}

 23%|██▎       | 3102/13317 [31:58<1:44:32,  1.63it/s]
 23%|██▎       | 3103/13317 [31:58<1:44:29,  1.63it/s]
                                                      
{'loss': 0.2085, 'grad_norm': 0.5741495490074158, 'learning_rate': 9.470839647803821e-06, 'epoch': 0.7}

 23%|██▎       | 3103/13317 [31:58<1:44:29,  1.63it/s]
 23%|██▎       | 3104/13317 [31:59<1:44:27,  1.63it/s]
                                                      
{'loss': 0.2139, 'grad_norm': 0.5212539434432983, 'learning_rate': 9.470252681216124e-06, 'epoch': 0.7}

 23%|██▎       | 3104/13317 [31:59<1:44:27,  1.63it/s]
 23%|██▎       | 3105/13317 [31:59<1:44:25,  1.63it/s]
                                                      
{'loss': 0.203, 'grad_norm': 0.4527615010738373, 'learning_rate': 9.469665407474738e-06, 'epoch': 0.7}

 23%|██▎       | 3105/13317 [32:00<1:44:25,  1.63it/s]
 23%|██▎       | 3106/13317 [32:00<1:44:22,  1.63it/s]
                                                      
{'loss': 0.222, 'grad_norm': 0.5394847393035889, 'learning_rate': 9.46907782662001e-06, 'epoch': 0.7}

 23%|██▎       | 3106/13317 [32:00<1:44:22,  1.63it/s]
 23%|██▎       | 3107/13317 [32:01<1:44:57,  1.62it/s]
                                                      
{'loss': 0.3426, 'grad_norm': 0.6505303382873535, 'learning_rate': 9.468489938692318e-06, 'epoch': 0.7}

 23%|██▎       | 3107/13317 [32:01<1:44:57,  1.62it/s]
 23%|██▎       | 3108/13317 [32:01<1:44:45,  1.62it/s]
                                                      
{'loss': 0.3303, 'grad_norm': 0.6372262835502625, 'learning_rate': 9.467901743732054e-06, 'epoch': 0.7}

 23%|██▎       | 3108/13317 [32:01<1:44:45,  1.62it/s]
 23%|██▎       | 3109/13317 [32:02<1:44:39,  1.63it/s]
                                                      
{'loss': 0.2878, 'grad_norm': 0.5644536018371582, 'learning_rate': 9.467313241779633e-06, 'epoch': 0.7}

 23%|██▎       | 3109/13317 [32:02<1:44:39,  1.63it/s]
 23%|██▎       | 3110/13317 [32:03<1:44:33,  1.63it/s]
                                                      
{'loss': 0.2477, 'grad_norm': 0.5586183071136475, 'learning_rate': 9.46672443287549e-06, 'epoch': 0.7}

 23%|██▎       | 3110/13317 [32:03<1:44:33,  1.63it/s]
 23%|██▎       | 3111/13317 [32:03<1:44:29,  1.63it/s]
                                                      
{'loss': 0.1907, 'grad_norm': 0.4190044105052948, 'learning_rate': 9.466135317060085e-06, 'epoch': 0.7}

 23%|██▎       | 3111/13317 [32:03<1:44:29,  1.63it/s]
 23%|██▎       | 3112/13317 [32:04<1:44:25,  1.63it/s]
                                                      
{'loss': 0.3756, 'grad_norm': 0.6695292592048645, 'learning_rate': 9.465545894373896e-06, 'epoch': 0.7}

 23%|██▎       | 3112/13317 [32:04<1:44:25,  1.63it/s]
 23%|██▎       | 3113/13317 [32:04<1:44:24,  1.63it/s]
                                                      
{'loss': 0.2123, 'grad_norm': 0.4698202908039093, 'learning_rate': 9.464956164857422e-06, 'epoch': 0.7}

 23%|██▎       | 3113/13317 [32:05<1:44:24,  1.63it/s]
 23%|██▎       | 3114/13317 [32:05<1:44:25,  1.63it/s]
                                                      
{'loss': 0.3084, 'grad_norm': 0.5901675820350647, 'learning_rate': 9.464366128551184e-06, 'epoch': 0.7}

 23%|██▎       | 3114/13317 [32:05<1:44:25,  1.63it/s]
 23%|██▎       | 3115/13317 [32:06<1:44:26,  1.63it/s]
                                                      
{'loss': 0.275, 'grad_norm': 0.5018272399902344, 'learning_rate': 9.463775785495723e-06, 'epoch': 0.7}

 23%|██▎       | 3115/13317 [32:06<1:44:26,  1.63it/s]
 23%|██▎       | 3116/13317 [32:06<1:44:26,  1.63it/s]
                                                      
{'loss': 0.2947, 'grad_norm': 0.5842723250389099, 'learning_rate': 9.463185135731602e-06, 'epoch': 0.7}

 23%|██▎       | 3116/13317 [32:06<1:44:26,  1.63it/s]
 23%|██▎       | 3117/13317 [32:07<1:44:23,  1.63it/s]
                                                      
{'loss': 0.2028, 'grad_norm': 0.5450541377067566, 'learning_rate': 9.462594179299408e-06, 'epoch': 0.7}

 23%|██▎       | 3117/13317 [32:07<1:44:23,  1.63it/s]
 23%|██▎       | 3118/13317 [32:07<1:44:23,  1.63it/s]
                                                      
{'loss': 0.2515, 'grad_norm': 0.5344480872154236, 'learning_rate': 9.46200291623974e-06, 'epoch': 0.7}

 23%|██▎       | 3118/13317 [32:08<1:44:23,  1.63it/s]
 23%|██▎       | 3119/13317 [32:08<1:44:26,  1.63it/s]
                                                      
{'loss': 0.166, 'grad_norm': 0.4167653024196625, 'learning_rate': 9.461411346593228e-06, 'epoch': 0.7}

 23%|██▎       | 3119/13317 [32:08<1:44:26,  1.63it/s]
 23%|██▎       | 3120/13317 [32:09<1:44:27,  1.63it/s]
                                                      
{'loss': 0.2868, 'grad_norm': 0.6155035495758057, 'learning_rate': 9.46081947040052e-06, 'epoch': 0.7}

 23%|██▎       | 3120/13317 [32:09<1:44:27,  1.63it/s]
 23%|██▎       | 3121/13317 [32:09<1:44:16,  1.63it/s]
                                                      
{'loss': 0.2524, 'grad_norm': 0.5343681573867798, 'learning_rate': 9.46022728770228e-06, 'epoch': 0.7}

 23%|██▎       | 3121/13317 [32:09<1:44:16,  1.63it/s]
 23%|██▎       | 3122/13317 [32:10<1:44:17,  1.63it/s]
                                                      
{'loss': 0.3231, 'grad_norm': 0.6304108500480652, 'learning_rate': 9.4596347985392e-06, 'epoch': 0.7}

 23%|██▎       | 3122/13317 [32:10<1:44:17,  1.63it/s]
 23%|██▎       | 3123/13317 [32:11<1:44:17,  1.63it/s]
                                                      
{'loss': 0.3039, 'grad_norm': 0.6183767318725586, 'learning_rate': 9.45904200295199e-06, 'epoch': 0.7}

 23%|██▎       | 3123/13317 [32:11<1:44:17,  1.63it/s]
 23%|██▎       | 3124/13317 [32:11<1:44:12,  1.63it/s]
                                                      
{'loss': 0.3573, 'grad_norm': 0.6255594491958618, 'learning_rate': 9.458448900981382e-06, 'epoch': 0.7}

 23%|██▎       | 3124/13317 [32:11<1:44:12,  1.63it/s]
 23%|██▎       | 3125/13317 [32:12<1:44:16,  1.63it/s]
                                                      
{'loss': 0.2166, 'grad_norm': 0.45954859256744385, 'learning_rate': 9.457855492668128e-06, 'epoch': 0.7}

 23%|██▎       | 3125/13317 [32:12<1:44:16,  1.63it/s]
 23%|██▎       | 3126/13317 [32:12<1:44:17,  1.63it/s]
                                                      
{'loss': 0.2849, 'grad_norm': 0.5939781665802002, 'learning_rate': 9.457261778053002e-06, 'epoch': 0.7}

 23%|██▎       | 3126/13317 [32:12<1:44:17,  1.63it/s]
 23%|██▎       | 3127/13317 [32:13<1:44:15,  1.63it/s]
                                                      
{'loss': 0.2509, 'grad_norm': 0.5210291743278503, 'learning_rate': 9.456667757176796e-06, 'epoch': 0.7}

 23%|██▎       | 3127/13317 [32:13<1:44:15,  1.63it/s]
 23%|██▎       | 3128/13317 [32:14<1:44:14,  1.63it/s]
                                                      
{'loss': 0.296, 'grad_norm': 0.5467372536659241, 'learning_rate': 9.456073430080327e-06, 'epoch': 0.7}

 23%|██▎       | 3128/13317 [32:14<1:44:14,  1.63it/s]
 23%|██▎       | 3129/13317 [32:14<1:44:16,  1.63it/s]
                                                      
{'loss': 0.3252, 'grad_norm': 0.6928276419639587, 'learning_rate': 9.455478796804432e-06, 'epoch': 0.7}

 23%|██▎       | 3129/13317 [32:14<1:44:16,  1.63it/s]
 24%|██▎       | 3130/13317 [32:15<1:44:15,  1.63it/s]
                                                      
{'loss': 0.2379, 'grad_norm': 0.5939196944236755, 'learning_rate': 9.45488385738997e-06, 'epoch': 0.71}

 24%|██▎       | 3130/13317 [32:15<1:44:15,  1.63it/s]
 24%|██▎       | 3131/13317 [32:15<1:44:16,  1.63it/s]
                                                      
{'loss': 0.2477, 'grad_norm': 0.5644726753234863, 'learning_rate': 9.454288611877815e-06, 'epoch': 0.71}

 24%|██▎       | 3131/13317 [32:16<1:44:16,  1.63it/s]
 24%|██▎       | 3132/13317 [32:16<1:44:15,  1.63it/s]
                                                      
{'loss': 0.2801, 'grad_norm': 0.5410284399986267, 'learning_rate': 9.453693060308871e-06, 'epoch': 0.71}

 24%|██▎       | 3132/13317 [32:16<1:44:15,  1.63it/s]
 24%|██▎       | 3133/13317 [32:17<1:44:13,  1.63it/s]
                                                      
{'loss': 0.2349, 'grad_norm': 0.5612520575523376, 'learning_rate': 9.453097202724057e-06, 'epoch': 0.71}

 24%|██▎       | 3133/13317 [32:17<1:44:13,  1.63it/s]
 24%|██▎       | 3134/13317 [32:17<1:44:15,  1.63it/s]
                                                      
{'loss': 0.2758, 'grad_norm': 0.6384111046791077, 'learning_rate': 9.452501039164316e-06, 'epoch': 0.71}

 24%|██▎       | 3134/13317 [32:17<1:44:15,  1.63it/s]
 24%|██▎       | 3135/13317 [32:18<1:44:16,  1.63it/s]
                                                      
{'loss': 0.2436, 'grad_norm': 0.5312628746032715, 'learning_rate': 9.451904569670609e-06, 'epoch': 0.71}

 24%|██▎       | 3135/13317 [32:18<1:44:16,  1.63it/s]
 24%|██▎       | 3136/13317 [32:18<1:44:10,  1.63it/s]
                                                      
{'loss': 0.2747, 'grad_norm': 0.5847230553627014, 'learning_rate': 9.451307794283921e-06, 'epoch': 0.71}

 24%|██▎       | 3136/13317 [32:19<1:44:10,  1.63it/s]
 24%|██▎       | 3137/13317 [32:19<1:44:16,  1.63it/s]
                                                      
{'loss': 0.2428, 'grad_norm': 0.5406540632247925, 'learning_rate': 9.450710713045255e-06, 'epoch': 0.71}

 24%|██▎       | 3137/13317 [32:19<1:44:16,  1.63it/s]
 24%|██▎       | 3138/13317 [32:20<1:44:14,  1.63it/s]
                                                      
{'loss': 0.2634, 'grad_norm': 0.566415548324585, 'learning_rate': 9.450113325995641e-06, 'epoch': 0.71}

 24%|██▎       | 3138/13317 [32:20<1:44:14,  1.63it/s]
 24%|██▎       | 3139/13317 [32:20<1:44:20,  1.63it/s]
                                                      
{'loss': 0.3141, 'grad_norm': 0.6003013253211975, 'learning_rate': 9.44951563317612e-06, 'epoch': 0.71}

 24%|██▎       | 3139/13317 [32:20<1:44:20,  1.63it/s]
 24%|██▎       | 3140/13317 [32:21<1:44:27,  1.62it/s]
                                                      
{'loss': 0.2607, 'grad_norm': 0.5431172847747803, 'learning_rate': 9.448917634627764e-06, 'epoch': 0.71}

 24%|██▎       | 3140/13317 [32:21<1:44:27,  1.62it/s]
 24%|██▎       | 3141/13317 [32:22<1:44:19,  1.63it/s]
                                                      
{'loss': 0.2587, 'grad_norm': 0.5426080226898193, 'learning_rate': 9.44831933039166e-06, 'epoch': 0.71}

 24%|██▎       | 3141/13317 [32:22<1:44:19,  1.63it/s]
 24%|██▎       | 3142/13317 [32:22<1:44:17,  1.63it/s]
                                                      
{'loss': 0.1565, 'grad_norm': 0.4276065230369568, 'learning_rate': 9.447720720508918e-06, 'epoch': 0.71}

 24%|██▎       | 3142/13317 [32:22<1:44:17,  1.63it/s]
 24%|██▎       | 3143/13317 [32:23<1:44:11,  1.63it/s]
                                                      
{'loss': 0.256, 'grad_norm': 0.6048504710197449, 'learning_rate': 9.447121805020671e-06, 'epoch': 0.71}

 24%|██▎       | 3143/13317 [32:23<1:44:11,  1.63it/s]
 24%|██▎       | 3144/13317 [32:23<1:44:09,  1.63it/s]
                                                      
{'loss': 0.2658, 'grad_norm': 0.5977084040641785, 'learning_rate': 9.446522583968067e-06, 'epoch': 0.71}

 24%|██▎       | 3144/13317 [32:24<1:44:09,  1.63it/s]
 24%|██▎       | 3145/13317 [32:24<1:44:16,  1.63it/s]
                                                      
{'loss': 0.3005, 'grad_norm': 0.5715791583061218, 'learning_rate': 9.445923057392283e-06, 'epoch': 0.71}

 24%|██▎       | 3145/13317 [32:24<1:44:16,  1.63it/s]
 24%|██▎       | 3146/13317 [32:25<1:44:08,  1.63it/s]
                                                      
{'loss': 0.2797, 'grad_norm': 0.6007423996925354, 'learning_rate': 9.44532322533451e-06, 'epoch': 0.71}

 24%|██▎       | 3146/13317 [32:25<1:44:08,  1.63it/s]
 24%|██▎       | 3147/13317 [32:25<1:44:03,  1.63it/s]
                                                      
{'loss': 0.255, 'grad_norm': 0.4809395968914032, 'learning_rate': 9.444723087835962e-06, 'epoch': 0.71}

 24%|██▎       | 3147/13317 [32:25<1:44:03,  1.63it/s]
 24%|██▎       | 3148/13317 [32:26<1:43:57,  1.63it/s]
                                                      
{'loss': 0.2704, 'grad_norm': 0.565955400466919, 'learning_rate': 9.44412264493788e-06, 'epoch': 0.71}

 24%|██▎       | 3148/13317 [32:26<1:43:57,  1.63it/s]
 24%|██▎       | 3149/13317 [32:26<1:43:58,  1.63it/s]
                                                      
{'loss': 0.2089, 'grad_norm': 0.5513453483581543, 'learning_rate': 9.443521896681515e-06, 'epoch': 0.71}

 24%|██▎       | 3149/13317 [32:27<1:43:58,  1.63it/s]
 24%|██▎       | 3150/13317 [32:27<1:44:07,  1.63it/s]
                                                      
{'loss': 0.2762, 'grad_norm': 0.5983428359031677, 'learning_rate': 9.442920843108148e-06, 'epoch': 0.71}

 24%|██▎       | 3150/13317 [32:27<1:44:07,  1.63it/s]
 24%|██▎       | 3151/13317 [32:28<1:44:02,  1.63it/s]
                                                      
{'loss': 0.207, 'grad_norm': 0.53664630651474, 'learning_rate': 9.442319484259075e-06, 'epoch': 0.71}

 24%|██▎       | 3151/13317 [32:28<1:44:02,  1.63it/s]
 24%|██▎       | 3152/13317 [32:28<1:43:56,  1.63it/s]
                                                      
{'loss': 0.3128, 'grad_norm': 0.5750687122344971, 'learning_rate': 9.441717820175617e-06, 'epoch': 0.71}

 24%|██▎       | 3152/13317 [32:28<1:43:56,  1.63it/s]
 24%|██▎       | 3153/13317 [32:29<1:43:58,  1.63it/s]
                                                      
{'loss': 0.2581, 'grad_norm': 0.4990088641643524, 'learning_rate': 9.441115850899117e-06, 'epoch': 0.71}

 24%|██▎       | 3153/13317 [32:29<1:43:58,  1.63it/s]
 24%|██▎       | 3154/13317 [32:30<1:44:03,  1.63it/s]
                                                      
{'loss': 0.2642, 'grad_norm': 0.5888904929161072, 'learning_rate': 9.440513576470933e-06, 'epoch': 0.71}

 24%|██▎       | 3154/13317 [32:30<1:44:03,  1.63it/s]
 24%|██▎       | 3155/13317 [32:30<1:44:09,  1.63it/s]
                                                      
{'loss': 0.2338, 'grad_norm': 0.45064106583595276, 'learning_rate': 9.43991099693245e-06, 'epoch': 0.71}

 24%|██▎       | 3155/13317 [32:30<1:44:09,  1.63it/s]
 24%|██▎       | 3156/13317 [32:31<1:44:01,  1.63it/s]
                                                      
{'loss': 0.2279, 'grad_norm': 0.5267093181610107, 'learning_rate': 9.439308112325071e-06, 'epoch': 0.71}

 24%|██▎       | 3156/13317 [32:31<1:44:01,  1.63it/s]
 24%|██▎       | 3157/13317 [32:31<1:43:58,  1.63it/s]
                                                      
{'loss': 0.2553, 'grad_norm': 0.4934568703174591, 'learning_rate': 9.438704922690222e-06, 'epoch': 0.71}

 24%|██▎       | 3157/13317 [32:32<1:43:58,  1.63it/s]
 24%|██▎       | 3158/13317 [32:32<1:43:52,  1.63it/s]
                                                      
{'loss': 0.2472, 'grad_norm': 0.5548195838928223, 'learning_rate': 9.438101428069347e-06, 'epoch': 0.71}

 24%|██▎       | 3158/13317 [32:32<1:43:52,  1.63it/s]
 24%|██▎       | 3159/13317 [32:33<1:43:53,  1.63it/s]
                                                      
{'loss': 0.2832, 'grad_norm': 0.5269201993942261, 'learning_rate': 9.437497628503911e-06, 'epoch': 0.71}

 24%|██▎       | 3159/13317 [32:33<1:43:53,  1.63it/s]
 24%|██▎       | 3160/13317 [32:33<1:44:00,  1.63it/s]
                                                      
{'loss': 0.253, 'grad_norm': 0.46618375182151794, 'learning_rate': 9.436893524035403e-06, 'epoch': 0.71}

 24%|██▎       | 3160/13317 [32:33<1:44:00,  1.63it/s]
 24%|██▎       | 3161/13317 [32:34<1:43:54,  1.63it/s]
                                                      
{'loss': 0.2463, 'grad_norm': 0.4996981620788574, 'learning_rate': 9.436289114705332e-06, 'epoch': 0.71}

 24%|██▎       | 3161/13317 [32:34<1:43:54,  1.63it/s]
 24%|██▎       | 3162/13317 [32:34<1:43:51,  1.63it/s]
                                                      
{'loss': 0.2934, 'grad_norm': 0.5584174990653992, 'learning_rate': 9.435684400555226e-06, 'epoch': 0.71}

 24%|██▎       | 3162/13317 [32:35<1:43:51,  1.63it/s]
 24%|██▍       | 3163/13317 [32:35<1:43:49,  1.63it/s]
                                                      
{'loss': 0.2534, 'grad_norm': 0.49436458945274353, 'learning_rate': 9.435079381626637e-06, 'epoch': 0.71}

 24%|██▍       | 3163/13317 [32:35<1:43:49,  1.63it/s]
 24%|██▍       | 3164/13317 [32:36<1:43:48,  1.63it/s]
                                                      
{'loss': 0.3003, 'grad_norm': 0.5853759050369263, 'learning_rate': 9.434474057961134e-06, 'epoch': 0.71}

 24%|██▍       | 3164/13317 [32:36<1:43:48,  1.63it/s]
 24%|██▍       | 3165/13317 [32:36<1:43:55,  1.63it/s]
                                                      
{'loss': 0.2957, 'grad_norm': 0.5971466302871704, 'learning_rate': 9.43386842960031e-06, 'epoch': 0.71}

 24%|██▍       | 3165/13317 [32:36<1:43:55,  1.63it/s]
 24%|██▍       | 3166/13317 [32:37<1:43:53,  1.63it/s]
                                                      
{'loss': 0.246, 'grad_norm': 0.6017525792121887, 'learning_rate': 9.43326249658578e-06, 'epoch': 0.71}

 24%|██▍       | 3166/13317 [32:37<1:43:53,  1.63it/s]
 24%|██▍       | 3167/13317 [32:38<1:43:47,  1.63it/s]
                                                      
{'loss': 0.2808, 'grad_norm': 0.584763765335083, 'learning_rate': 9.432656258959175e-06, 'epoch': 0.71}

 24%|██▍       | 3167/13317 [32:38<1:43:47,  1.63it/s]
 24%|██▍       | 3168/13317 [32:38<1:43:45,  1.63it/s]
                                                      
{'loss': 0.2397, 'grad_norm': 0.5216559171676636, 'learning_rate': 9.432049716762151e-06, 'epoch': 0.71}

 24%|██▍       | 3168/13317 [32:38<1:43:45,  1.63it/s]
 24%|██▍       | 3169/13317 [32:39<1:43:42,  1.63it/s]
                                                      
{'loss': 0.2879, 'grad_norm': 0.6268324851989746, 'learning_rate': 9.431442870036384e-06, 'epoch': 0.71}

 24%|██▍       | 3169/13317 [32:39<1:43:42,  1.63it/s]
 24%|██▍       | 3170/13317 [32:39<1:43:46,  1.63it/s]
                                                      
{'loss': 0.1966, 'grad_norm': 0.48045599460601807, 'learning_rate': 9.430835718823572e-06, 'epoch': 0.71}

 24%|██▍       | 3170/13317 [32:40<1:43:46,  1.63it/s]
 24%|██▍       | 3171/13317 [32:40<1:43:47,  1.63it/s]
                                                      
{'loss': 0.197, 'grad_norm': 0.4868941307067871, 'learning_rate': 9.430228263165434e-06, 'epoch': 0.71}

 24%|██▍       | 3171/13317 [32:40<1:43:47,  1.63it/s]
 24%|██▍       | 3172/13317 [32:41<1:43:44,  1.63it/s]
                                                      
{'loss': 0.2746, 'grad_norm': 0.5925983786582947, 'learning_rate': 9.429620503103703e-06, 'epoch': 0.71}

 24%|██▍       | 3172/13317 [32:41<1:43:44,  1.63it/s]
 24%|██▍       | 3173/13317 [32:41<1:43:45,  1.63it/s]
                                                      
{'loss': 0.1825, 'grad_norm': 0.5113939046859741, 'learning_rate': 9.42901243868014e-06, 'epoch': 0.71}

 24%|██▍       | 3173/13317 [32:41<1:43:45,  1.63it/s]
 24%|██▍       | 3174/13317 [32:42<1:43:44,  1.63it/s]
                                                      
{'loss': 0.2051, 'grad_norm': 0.4566026031970978, 'learning_rate': 9.428404069936533e-06, 'epoch': 0.72}

 24%|██▍       | 3174/13317 [32:42<1:43:44,  1.63it/s]
 24%|██▍       | 3175/13317 [32:42<1:43:49,  1.63it/s]
                                                      
{'loss': 0.2387, 'grad_norm': 0.5870369076728821, 'learning_rate': 9.427795396914673e-06, 'epoch': 0.72}

 24%|██▍       | 3175/13317 [32:43<1:43:49,  1.63it/s]
 24%|██▍       | 3176/13317 [32:43<1:43:47,  1.63it/s]
                                                      
{'loss': 0.1953, 'grad_norm': 0.46254539489746094, 'learning_rate': 9.427186419656389e-06, 'epoch': 0.72}

 24%|██▍       | 3176/13317 [32:43<1:43:47,  1.63it/s]
 24%|██▍       | 3177/13317 [32:44<1:43:44,  1.63it/s]
                                                      
{'loss': 0.2508, 'grad_norm': 0.554401695728302, 'learning_rate': 9.426577138203521e-06, 'epoch': 0.72}

 24%|██▍       | 3177/13317 [32:44<1:43:44,  1.63it/s]
 24%|██▍       | 3178/13317 [32:44<1:43:39,  1.63it/s]
                                                      
{'loss': 0.2403, 'grad_norm': 0.5456129908561707, 'learning_rate': 9.425967552597934e-06, 'epoch': 0.72}

 24%|██▍       | 3178/13317 [32:44<1:43:39,  1.63it/s]
 24%|██▍       | 3179/13317 [32:45<1:43:42,  1.63it/s]
                                                      
{'loss': 0.3214, 'grad_norm': 0.5885248184204102, 'learning_rate': 9.425357662881515e-06, 'epoch': 0.72}

 24%|██▍       | 3179/13317 [32:45<1:43:42,  1.63it/s]
 24%|██▍       | 3180/13317 [32:46<1:43:41,  1.63it/s]
                                                      
{'loss': 0.2404, 'grad_norm': 0.4951915144920349, 'learning_rate': 9.424747469096167e-06, 'epoch': 0.72}

 24%|██▍       | 3180/13317 [32:46<1:43:41,  1.63it/s]
 24%|██▍       | 3181/13317 [32:46<1:43:40,  1.63it/s]
                                                      
{'loss': 0.2116, 'grad_norm': 0.5416821837425232, 'learning_rate': 9.42413697128382e-06, 'epoch': 0.72}

 24%|██▍       | 3181/13317 [32:46<1:43:40,  1.63it/s]
 24%|██▍       | 3182/13317 [32:47<1:43:37,  1.63it/s]
                                                      
{'loss': 0.1782, 'grad_norm': 0.4536038637161255, 'learning_rate': 9.423526169486417e-06, 'epoch': 0.72}

 24%|██▍       | 3182/13317 [32:47<1:43:37,  1.63it/s]
 24%|██▍       | 3183/13317 [32:47<1:43:34,  1.63it/s]
                                                      
{'loss': 0.2453, 'grad_norm': 0.568612277507782, 'learning_rate': 9.42291506374593e-06, 'epoch': 0.72}

 24%|██▍       | 3183/13317 [32:47<1:43:34,  1.63it/s]
 24%|██▍       | 3184/13317 [32:48<1:43:34,  1.63it/s]
                                                      
{'loss': 0.2617, 'grad_norm': 0.5825778245925903, 'learning_rate': 9.422303654104347e-06, 'epoch': 0.72}

 24%|██▍       | 3184/13317 [32:48<1:43:34,  1.63it/s]
 24%|██▍       | 3185/13317 [32:49<1:43:30,  1.63it/s]
                                                      
{'loss': 0.2491, 'grad_norm': 0.5376750826835632, 'learning_rate': 9.42169194060368e-06, 'epoch': 0.72}

 24%|██▍       | 3185/13317 [32:49<1:43:30,  1.63it/s]
 24%|██▍       | 3186/13317 [32:49<1:43:34,  1.63it/s]
                                                      
{'loss': 0.2186, 'grad_norm': 0.4659641981124878, 'learning_rate': 9.421079923285958e-06, 'epoch': 0.72}

 24%|██▍       | 3186/13317 [32:49<1:43:34,  1.63it/s]
 24%|██▍       | 3187/13317 [32:50<1:43:34,  1.63it/s]
                                                      
{'loss': 0.2122, 'grad_norm': 0.49679118394851685, 'learning_rate': 9.420467602193234e-06, 'epoch': 0.72}

 24%|██▍       | 3187/13317 [32:50<1:43:34,  1.63it/s]
 24%|██▍       | 3188/13317 [32:50<1:43:37,  1.63it/s]
                                                      
{'loss': 0.2251, 'grad_norm': 0.5306106805801392, 'learning_rate': 9.419854977367583e-06, 'epoch': 0.72}

 24%|██▍       | 3188/13317 [32:51<1:43:37,  1.63it/s]
 24%|██▍       | 3189/13317 [32:51<1:43:37,  1.63it/s]
                                                      
{'loss': 0.3235, 'grad_norm': 0.689175009727478, 'learning_rate': 9.419242048851093e-06, 'epoch': 0.72}

 24%|██▍       | 3189/13317 [32:51<1:43:37,  1.63it/s]
 24%|██▍       | 3190/13317 [32:52<1:43:36,  1.63it/s]
                                                      
{'loss': 0.2496, 'grad_norm': 0.5662642121315002, 'learning_rate': 9.418628816685886e-06, 'epoch': 0.72}

 24%|██▍       | 3190/13317 [32:52<1:43:36,  1.63it/s]
 24%|██▍       | 3191/13317 [32:52<1:43:32,  1.63it/s]
                                                      
{'loss': 0.2306, 'grad_norm': 0.5643953680992126, 'learning_rate': 9.418015280914094e-06, 'epoch': 0.72}

 24%|██▍       | 3191/13317 [32:52<1:43:32,  1.63it/s]
 24%|██▍       | 3192/13317 [32:53<1:43:46,  1.63it/s]
                                                      
{'loss': 0.3017, 'grad_norm': 0.6024090647697449, 'learning_rate': 9.41740144157787e-06, 'epoch': 0.72}

 24%|██▍       | 3192/13317 [32:53<1:43:46,  1.63it/s]
 24%|██▍       | 3193/13317 [32:53<1:43:42,  1.63it/s]
                                                      
{'loss': 0.1999, 'grad_norm': 0.44020992517471313, 'learning_rate': 9.416787298719398e-06, 'epoch': 0.72}

 24%|██▍       | 3193/13317 [32:54<1:43:42,  1.63it/s]
 24%|██▍       | 3194/13317 [32:54<1:43:38,  1.63it/s]
                                                      
{'loss': 0.3094, 'grad_norm': 0.5790213346481323, 'learning_rate': 9.416172852380874e-06, 'epoch': 0.72}

 24%|██▍       | 3194/13317 [32:54<1:43:38,  1.63it/s]
 24%|██▍       | 3195/13317 [32:55<1:43:32,  1.63it/s]
                                                      
{'loss': 0.223, 'grad_norm': 0.5074520707130432, 'learning_rate': 9.415558102604515e-06, 'epoch': 0.72}

 24%|██▍       | 3195/13317 [32:55<1:43:32,  1.63it/s]
 24%|██▍       | 3196/13317 [32:55<1:43:28,  1.63it/s]
                                                      
{'loss': 0.2648, 'grad_norm': 0.5164620876312256, 'learning_rate': 9.41494304943256e-06, 'epoch': 0.72}

 24%|██▍       | 3196/13317 [32:55<1:43:28,  1.63it/s]
 24%|██▍       | 3197/13317 [32:56<1:43:30,  1.63it/s]
                                                      
{'loss': 0.29, 'grad_norm': 0.7631735801696777, 'learning_rate': 9.414327692907272e-06, 'epoch': 0.72}

 24%|██▍       | 3197/13317 [32:56<1:43:30,  1.63it/s]
 24%|██▍       | 3198/13317 [32:57<1:43:29,  1.63it/s]
                                                      
{'loss': 0.2511, 'grad_norm': 0.5966005921363831, 'learning_rate': 9.413712033070931e-06, 'epoch': 0.72}

 24%|██▍       | 3198/13317 [32:57<1:43:29,  1.63it/s]
 24%|██▍       | 3199/13317 [32:57<1:43:26,  1.63it/s]
                                                      
{'loss': 0.2402, 'grad_norm': 0.5586456060409546, 'learning_rate': 9.413096069965842e-06, 'epoch': 0.72}

 24%|██▍       | 3199/13317 [32:57<1:43:26,  1.63it/s]
 24%|██▍       | 3200/13317 [32:58<1:43:28,  1.63it/s]
                                                      
{'loss': 0.2756, 'grad_norm': 0.5113767981529236, 'learning_rate': 9.412479803634326e-06, 'epoch': 0.72}

 24%|██▍       | 3200/13317 [32:58<1:43:28,  1.63it/s]
 24%|██▍       | 3201/13317 [32:58<1:43:24,  1.63it/s]
                                                      
{'loss': 0.2149, 'grad_norm': 0.4640437066555023, 'learning_rate': 9.411863234118727e-06, 'epoch': 0.72}

 24%|██▍       | 3201/13317 [32:59<1:43:24,  1.63it/s]
 24%|██▍       | 3202/13317 [32:59<1:43:23,  1.63it/s]
                                                      
{'loss': 0.2977, 'grad_norm': 0.5953981280326843, 'learning_rate': 9.41124636146141e-06, 'epoch': 0.72}

 24%|██▍       | 3202/13317 [32:59<1:43:23,  1.63it/s]
 24%|██▍       | 3203/13317 [33:00<1:43:29,  1.63it/s]
                                                      
{'loss': 0.3311, 'grad_norm': 0.6574826240539551, 'learning_rate': 9.41062918570476e-06, 'epoch': 0.72}

 24%|██▍       | 3203/13317 [33:00<1:43:29,  1.63it/s]
 24%|██▍       | 3204/13317 [33:00<1:43:25,  1.63it/s]
                                                      
{'loss': 0.1776, 'grad_norm': 0.3925679624080658, 'learning_rate': 9.410011706891187e-06, 'epoch': 0.72}

 24%|██▍       | 3204/13317 [33:00<1:43:25,  1.63it/s]
 24%|██▍       | 3205/13317 [33:01<1:43:23,  1.63it/s]
                                                      
{'loss': 0.2893, 'grad_norm': 0.6143432259559631, 'learning_rate': 9.409393925063114e-06, 'epoch': 0.72}

 24%|██▍       | 3205/13317 [33:01<1:43:23,  1.63it/s]
 24%|██▍       | 3206/13317 [33:01<1:43:22,  1.63it/s]
                                                      
{'loss': 0.321, 'grad_norm': 0.595749020576477, 'learning_rate': 9.408775840262991e-06, 'epoch': 0.72}

 24%|██▍       | 3206/13317 [33:02<1:43:22,  1.63it/s]
 24%|██▍       | 3207/13317 [33:02<1:43:18,  1.63it/s]
                                                      
{'loss': 0.2216, 'grad_norm': 0.4786226153373718, 'learning_rate': 9.408157452533288e-06, 'epoch': 0.72}

 24%|██▍       | 3207/13317 [33:02<1:43:18,  1.63it/s]
 24%|██▍       | 3208/13317 [33:03<1:43:21,  1.63it/s]
                                                      
{'loss': 0.3024, 'grad_norm': 0.6710831522941589, 'learning_rate': 9.407538761916492e-06, 'epoch': 0.72}

 24%|██▍       | 3208/13317 [33:03<1:43:21,  1.63it/s]
 24%|██▍       | 3209/13317 [33:03<1:43:19,  1.63it/s]
                                                      
{'loss': 0.2237, 'grad_norm': 0.5118451714515686, 'learning_rate': 9.406919768455117e-06, 'epoch': 0.72}

 24%|██▍       | 3209/13317 [33:03<1:43:19,  1.63it/s]
 24%|██▍       | 3210/13317 [33:04<1:43:16,  1.63it/s]
                                                      
{'loss': 0.307, 'grad_norm': 0.5924102663993835, 'learning_rate': 9.406300472191691e-06, 'epoch': 0.72}

 24%|██▍       | 3210/13317 [33:04<1:43:16,  1.63it/s]
 24%|██▍       | 3211/13317 [33:05<1:43:16,  1.63it/s]
                                                      
{'loss': 0.2556, 'grad_norm': 0.640143632888794, 'learning_rate': 9.40568087316877e-06, 'epoch': 0.72}

 24%|██▍       | 3211/13317 [33:05<1:43:16,  1.63it/s]
 24%|██▍       | 3212/13317 [33:05<1:43:17,  1.63it/s]
                                                      
{'loss': 0.2403, 'grad_norm': 0.4418479800224304, 'learning_rate': 9.405060971428924e-06, 'epoch': 0.72}

 24%|██▍       | 3212/13317 [33:05<1:43:17,  1.63it/s]
 24%|██▍       | 3213/13317 [33:06<1:43:16,  1.63it/s]
                                                      
{'loss': 0.3084, 'grad_norm': 0.509601354598999, 'learning_rate': 9.404440767014746e-06, 'epoch': 0.72}

 24%|██▍       | 3213/13317 [33:06<1:43:16,  1.63it/s]
 24%|██▍       | 3214/13317 [33:06<1:43:16,  1.63it/s]
                                                      
{'loss': 0.2682, 'grad_norm': 0.49343422055244446, 'learning_rate': 9.403820259968854e-06, 'epoch': 0.72}

 24%|██▍       | 3214/13317 [33:07<1:43:16,  1.63it/s]
 24%|██▍       | 3215/13317 [33:07<1:43:19,  1.63it/s]
                                                      
{'loss': 0.2838, 'grad_norm': 0.6749040484428406, 'learning_rate': 9.403199450333881e-06, 'epoch': 0.72}

 24%|██▍       | 3215/13317 [33:07<1:43:19,  1.63it/s]
 24%|██▍       | 3216/13317 [33:08<1:43:20,  1.63it/s]
                                                      
{'loss': 0.2371, 'grad_norm': 0.47367173433303833, 'learning_rate': 9.402578338152487e-06, 'epoch': 0.72}

 24%|██▍       | 3216/13317 [33:08<1:43:20,  1.63it/s]
 24%|██▍       | 3217/13317 [33:08<1:43:20,  1.63it/s]
                                                      
{'loss': 0.2841, 'grad_norm': 0.5468733310699463, 'learning_rate': 9.40195692346734e-06, 'epoch': 0.72}

 24%|██▍       | 3217/13317 [33:08<1:43:20,  1.63it/s]
 24%|██▍       | 3218/13317 [33:09<1:43:19,  1.63it/s]
                                                      
{'loss': 0.2496, 'grad_norm': 0.5360663533210754, 'learning_rate': 9.40133520632115e-06, 'epoch': 0.72}

 24%|██▍       | 3218/13317 [33:09<1:43:19,  1.63it/s]
 24%|██▍       | 3219/13317 [33:09<1:43:19,  1.63it/s]
                                                      
{'loss': 0.2318, 'grad_norm': 0.6658249497413635, 'learning_rate': 9.400713186756625e-06, 'epoch': 0.73}

 24%|██▍       | 3219/13317 [33:10<1:43:19,  1.63it/s]
 24%|██▍       | 3220/13317 [33:10<1:43:17,  1.63it/s]
                                                      
{'loss': 0.2901, 'grad_norm': 0.6416469812393188, 'learning_rate': 9.400090864816512e-06, 'epoch': 0.73}

 24%|██▍       | 3220/13317 [33:10<1:43:17,  1.63it/s]
 24%|██▍       | 3221/13317 [33:11<1:43:17,  1.63it/s]
                                                      
{'loss': 0.2902, 'grad_norm': 0.5934906005859375, 'learning_rate': 9.399468240543565e-06, 'epoch': 0.73}

 24%|██▍       | 3221/13317 [33:11<1:43:17,  1.63it/s]
 24%|██▍       | 3222/13317 [33:11<1:43:20,  1.63it/s]
                                                      
{'loss': 0.2836, 'grad_norm': 0.7689126133918762, 'learning_rate': 9.398845313980567e-06, 'epoch': 0.73}

 24%|██▍       | 3222/13317 [33:11<1:43:20,  1.63it/s]
 24%|██▍       | 3223/13317 [33:12<1:43:15,  1.63it/s]
                                                      
{'loss': 0.293, 'grad_norm': 0.5993302464485168, 'learning_rate': 9.398222085170321e-06, 'epoch': 0.73}

 24%|██▍       | 3223/13317 [33:12<1:43:15,  1.63it/s]
 24%|██▍       | 3224/13317 [33:13<1:43:12,  1.63it/s]
                                                      
{'loss': 0.2355, 'grad_norm': 0.5243382453918457, 'learning_rate': 9.397598554155648e-06, 'epoch': 0.73}

 24%|██▍       | 3224/13317 [33:13<1:43:12,  1.63it/s]
 24%|██▍       | 3225/13317 [33:13<1:43:10,  1.63it/s]
                                                      
{'loss': 0.2065, 'grad_norm': 0.4435337483882904, 'learning_rate': 9.396974720979395e-06, 'epoch': 0.73}

 24%|██▍       | 3225/13317 [33:13<1:43:10,  1.63it/s]
 24%|██▍       | 3226/13317 [33:14<1:43:23,  1.63it/s]
                                                      
{'loss': 0.2617, 'grad_norm': 0.5949369072914124, 'learning_rate': 9.39635058568442e-06, 'epoch': 0.73}

 24%|██▍       | 3226/13317 [33:14<1:43:23,  1.63it/s]
 24%|██▍       | 3227/13317 [33:14<1:43:17,  1.63it/s]
                                                      
{'loss': 0.3018, 'grad_norm': 0.5438033938407898, 'learning_rate': 9.39572614831361e-06, 'epoch': 0.73}

 24%|██▍       | 3227/13317 [33:14<1:43:17,  1.63it/s]
 24%|██▍       | 3228/13317 [33:15<1:43:15,  1.63it/s]
                                                      
{'loss': 0.2142, 'grad_norm': 0.4889594614505768, 'learning_rate': 9.395101408909872e-06, 'epoch': 0.73}

 24%|██▍       | 3228/13317 [33:15<1:43:15,  1.63it/s]
 24%|██▍       | 3229/13317 [33:16<1:43:12,  1.63it/s]
                                                      
{'loss': 0.2991, 'grad_norm': 0.4889054298400879, 'learning_rate': 9.394476367516133e-06, 'epoch': 0.73}

 24%|██▍       | 3229/13317 [33:16<1:43:12,  1.63it/s]
 24%|██▍       | 3230/13317 [33:16<1:43:11,  1.63it/s]
                                                      
{'loss': 0.254, 'grad_norm': 0.4914930462837219, 'learning_rate': 9.393851024175336e-06, 'epoch': 0.73}

 24%|██▍       | 3230/13317 [33:16<1:43:11,  1.63it/s]
 24%|██▍       | 3231/13317 [33:17<1:43:12,  1.63it/s]
                                                      
{'loss': 0.2535, 'grad_norm': 0.42127764225006104, 'learning_rate': 9.39322537893045e-06, 'epoch': 0.73}

 24%|██▍       | 3231/13317 [33:17<1:43:12,  1.63it/s]
 24%|██▍       | 3232/13317 [33:17<1:43:24,  1.63it/s]
                                                      
{'loss': 0.2406, 'grad_norm': 0.4972647726535797, 'learning_rate': 9.392599431824465e-06, 'epoch': 0.73}

 24%|██▍       | 3232/13317 [33:18<1:43:24,  1.63it/s]
 24%|██▍       | 3233/13317 [33:18<1:43:15,  1.63it/s]
                                                      
{'loss': 0.3306, 'grad_norm': 0.6644205451011658, 'learning_rate': 9.39197318290039e-06, 'epoch': 0.73}

 24%|██▍       | 3233/13317 [33:18<1:43:15,  1.63it/s]
 24%|██▍       | 3234/13317 [33:19<1:43:12,  1.63it/s]
                                                      
{'loss': 0.2384, 'grad_norm': 0.5974321961402893, 'learning_rate': 9.391346632201255e-06, 'epoch': 0.73}

 24%|██▍       | 3234/13317 [33:19<1:43:12,  1.63it/s]
 24%|██▍       | 3235/13317 [33:19<1:43:06,  1.63it/s]
                                                      
{'loss': 0.2311, 'grad_norm': 0.6589708924293518, 'learning_rate': 9.390719779770108e-06, 'epoch': 0.73}

 24%|██▍       | 3235/13317 [33:19<1:43:06,  1.63it/s]
 24%|██▍       | 3236/13317 [33:20<1:43:04,  1.63it/s]
                                                      
{'loss': 0.2693, 'grad_norm': 0.5879087448120117, 'learning_rate': 9.390092625650024e-06, 'epoch': 0.73}

 24%|██▍       | 3236/13317 [33:20<1:43:04,  1.63it/s]
 24%|██▍       | 3237/13317 [33:21<1:43:11,  1.63it/s]
                                                      
{'loss': 0.2534, 'grad_norm': 0.6260934472084045, 'learning_rate': 9.389465169884094e-06, 'epoch': 0.73}

 24%|██▍       | 3237/13317 [33:21<1:43:11,  1.63it/s]
 24%|██▍       | 3238/13317 [33:21<1:43:08,  1.63it/s]
                                                      
{'loss': 0.3258, 'grad_norm': 0.6069020628929138, 'learning_rate': 9.388837412515428e-06, 'epoch': 0.73}

 24%|██▍       | 3238/13317 [33:21<1:43:08,  1.63it/s]
 24%|██▍       | 3239/13317 [33:22<1:43:04,  1.63it/s]
                                                      
{'loss': 0.3273, 'grad_norm': 0.6184947490692139, 'learning_rate': 9.388209353587164e-06, 'epoch': 0.73}

 24%|██▍       | 3239/13317 [33:22<1:43:04,  1.63it/s]
 24%|██▍       | 3240/13317 [33:22<1:43:01,  1.63it/s]
                                                      
{'loss': 0.2331, 'grad_norm': 0.48545703291893005, 'learning_rate': 9.387580993142453e-06, 'epoch': 0.73}

 24%|██▍       | 3240/13317 [33:22<1:43:01,  1.63it/s]
 24%|██▍       | 3241/13317 [33:23<1:43:00,  1.63it/s]
                                                      
{'loss': 0.3181, 'grad_norm': 0.5757179260253906, 'learning_rate': 9.386952331224472e-06, 'epoch': 0.73}

 24%|██▍       | 3241/13317 [33:23<1:43:00,  1.63it/s]
 24%|██▍       | 3242/13317 [33:24<1:43:03,  1.63it/s]
                                                      
{'loss': 0.2615, 'grad_norm': 0.4949221611022949, 'learning_rate': 9.386323367876417e-06, 'epoch': 0.73}

 24%|██▍       | 3242/13317 [33:24<1:43:03,  1.63it/s]
 24%|██▍       | 3243/13317 [33:24<1:43:01,  1.63it/s]
                                                      
{'loss': 0.3398, 'grad_norm': 0.7010434865951538, 'learning_rate': 9.385694103141502e-06, 'epoch': 0.73}

 24%|██▍       | 3243/13317 [33:24<1:43:01,  1.63it/s]
 24%|██▍       | 3244/13317 [33:25<1:43:00,  1.63it/s]
                                                      
{'loss': 0.2697, 'grad_norm': 0.5749279856681824, 'learning_rate': 9.385064537062967e-06, 'epoch': 0.73}

 24%|██▍       | 3244/13317 [33:25<1:43:00,  1.63it/s]
 24%|██▍       | 3245/13317 [33:25<1:43:01,  1.63it/s]
                                                      
{'loss': 0.2328, 'grad_norm': 0.5537091493606567, 'learning_rate': 9.384434669684068e-06, 'epoch': 0.73}

 24%|██▍       | 3245/13317 [33:26<1:43:01,  1.63it/s]
 24%|██▍       | 3246/13317 [33:26<1:42:59,  1.63it/s]
                                                      
{'loss': 0.3618, 'grad_norm': 0.533527135848999, 'learning_rate': 9.383804501048083e-06, 'epoch': 0.73}

 24%|██▍       | 3246/13317 [33:26<1:42:59,  1.63it/s]
 24%|██▍       | 3247/13317 [33:27<1:43:02,  1.63it/s]
                                                      
{'loss': 0.3016, 'grad_norm': 0.5777076482772827, 'learning_rate': 9.383174031198315e-06, 'epoch': 0.73}

 24%|██▍       | 3247/13317 [33:27<1:43:02,  1.63it/s]
 24%|██▍       | 3248/13317 [33:27<1:42:56,  1.63it/s]
                                                      
{'loss': 0.2048, 'grad_norm': 0.4101469814777374, 'learning_rate': 9.382543260178079e-06, 'epoch': 0.73}

 24%|██▍       | 3248/13317 [33:27<1:42:56,  1.63it/s]
 24%|██▍       | 3249/13317 [33:28<1:42:55,  1.63it/s]
                                                      
{'loss': 0.2844, 'grad_norm': 0.6141491532325745, 'learning_rate': 9.381912188030718e-06, 'epoch': 0.73}

 24%|██▍       | 3249/13317 [33:28<1:42:55,  1.63it/s]
 24%|██▍       | 3250/13317 [33:28<1:42:58,  1.63it/s]
                                                      
{'loss': 0.2317, 'grad_norm': 0.5034469366073608, 'learning_rate': 9.381280814799594e-06, 'epoch': 0.73}

 24%|██▍       | 3250/13317 [33:29<1:42:58,  1.63it/s]
 24%|██▍       | 3251/13317 [33:29<1:42:53,  1.63it/s]
                                                      
{'loss': 0.2868, 'grad_norm': 0.5435609221458435, 'learning_rate': 9.380649140528087e-06, 'epoch': 0.73}

 24%|██▍       | 3251/13317 [33:29<1:42:53,  1.63it/s]
 24%|██▍       | 3252/13317 [33:30<1:42:57,  1.63it/s]
                                                      
{'loss': 0.3157, 'grad_norm': 0.6111819744110107, 'learning_rate': 9.380017165259602e-06, 'epoch': 0.73}

 24%|██▍       | 3252/13317 [33:30<1:42:57,  1.63it/s]
 24%|██▍       | 3253/13317 [33:30<1:42:57,  1.63it/s]
                                                      
{'loss': 0.2641, 'grad_norm': 0.5748182535171509, 'learning_rate': 9.379384889037562e-06, 'epoch': 0.73}

 24%|██▍       | 3253/13317 [33:30<1:42:57,  1.63it/s]
 24%|██▍       | 3254/13317 [33:31<1:42:53,  1.63it/s]
                                                      
{'loss': 0.2468, 'grad_norm': 0.6102429032325745, 'learning_rate': 9.37875231190541e-06, 'epoch': 0.73}

 24%|██▍       | 3254/13317 [33:31<1:42:53,  1.63it/s]
 24%|██▍       | 3255/13317 [33:32<1:42:53,  1.63it/s]
                                                      
{'loss': 0.2134, 'grad_norm': 0.49370530247688293, 'learning_rate': 9.37811943390661e-06, 'epoch': 0.73}

 24%|██▍       | 3255/13317 [33:32<1:42:53,  1.63it/s]
 24%|██▍       | 3256/13317 [33:32<1:42:52,  1.63it/s]
                                                      
{'loss': 0.2542, 'grad_norm': 0.5705927014350891, 'learning_rate': 9.37748625508465e-06, 'epoch': 0.73}

 24%|██▍       | 3256/13317 [33:32<1:42:52,  1.63it/s]
 24%|██▍       | 3257/13317 [33:33<1:42:53,  1.63it/s]
                                                      
{'loss': 0.3133, 'grad_norm': 0.6113415360450745, 'learning_rate': 9.376852775483034e-06, 'epoch': 0.73}

 24%|██▍       | 3257/13317 [33:33<1:42:53,  1.63it/s]
 24%|██▍       | 3258/13317 [33:33<1:42:51,  1.63it/s]
                                                      
{'loss': 0.3025, 'grad_norm': 0.636140763759613, 'learning_rate': 9.37621899514529e-06, 'epoch': 0.73}

 24%|██▍       | 3258/13317 [33:34<1:42:51,  1.63it/s]
 24%|██▍       | 3259/13317 [33:34<1:42:51,  1.63it/s]
                                                      
{'loss': 0.2406, 'grad_norm': 0.5181713104248047, 'learning_rate': 9.375584914114963e-06, 'epoch': 0.73}

 24%|██▍       | 3259/13317 [33:34<1:42:51,  1.63it/s]
 24%|██▍       | 3260/13317 [33:35<1:42:51,  1.63it/s]
                                                      
{'loss': 0.2119, 'grad_norm': 0.5194216966629028, 'learning_rate': 9.374950532435626e-06, 'epoch': 0.73}

 24%|██▍       | 3260/13317 [33:35<1:42:51,  1.63it/s]
 24%|██▍       | 3261/13317 [33:35<1:42:51,  1.63it/s]
                                                      
{'loss': 0.1789, 'grad_norm': 0.4453338384628296, 'learning_rate': 9.374315850150864e-06, 'epoch': 0.73}

 24%|██▍       | 3261/13317 [33:35<1:42:51,  1.63it/s]
 24%|██▍       | 3262/13317 [33:36<1:42:56,  1.63it/s]
                                                      
{'loss': 0.1942, 'grad_norm': 0.4902389347553253, 'learning_rate': 9.373680867304285e-06, 'epoch': 0.73}

 24%|██▍       | 3262/13317 [33:36<1:42:56,  1.63it/s]
 25%|██▍       | 3263/13317 [33:36<1:42:53,  1.63it/s]
                                                      
{'loss': 0.2483, 'grad_norm': 0.5625793933868408, 'learning_rate': 9.373045583939523e-06, 'epoch': 0.74}

 25%|██▍       | 3263/13317 [33:37<1:42:53,  1.63it/s]
 25%|██▍       | 3264/13317 [33:37<1:42:55,  1.63it/s]
                                                      
{'loss': 0.3024, 'grad_norm': 0.6923137307167053, 'learning_rate': 9.372410000100225e-06, 'epoch': 0.74}

 25%|██▍       | 3264/13317 [33:37<1:42:55,  1.63it/s]
 25%|██▍       | 3265/13317 [33:38<1:42:49,  1.63it/s]
                                                      
{'loss': 0.2419, 'grad_norm': 0.5076466798782349, 'learning_rate': 9.371774115830066e-06, 'epoch': 0.74}

 25%|██▍       | 3265/13317 [33:38<1:42:49,  1.63it/s]
 25%|██▍       | 3266/13317 [33:38<1:42:46,  1.63it/s]
                                                      
{'loss': 0.2293, 'grad_norm': 0.5361841320991516, 'learning_rate': 9.371137931172735e-06, 'epoch': 0.74}

 25%|██▍       | 3266/13317 [33:38<1:42:46,  1.63it/s]
 25%|██▍       | 3267/13317 [33:39<1:42:49,  1.63it/s]
                                                      
{'loss': 0.2257, 'grad_norm': 0.5355136394500732, 'learning_rate': 9.370501446171948e-06, 'epoch': 0.74}

 25%|██▍       | 3267/13317 [33:39<1:42:49,  1.63it/s]
 25%|██▍       | 3268/13317 [33:40<1:42:44,  1.63it/s]
                                                      
{'loss': 0.2665, 'grad_norm': 0.5101928114891052, 'learning_rate': 9.369864660871434e-06, 'epoch': 0.74}

 25%|██▍       | 3268/13317 [33:40<1:42:44,  1.63it/s]
 25%|██▍       | 3269/13317 [33:40<1:42:43,  1.63it/s]
                                                      
{'loss': 0.2311, 'grad_norm': 0.4978584051132202, 'learning_rate': 9.36922757531495e-06, 'epoch': 0.74}

 25%|██▍       | 3269/13317 [33:40<1:42:43,  1.63it/s]
 25%|██▍       | 3270/13317 [33:41<1:42:43,  1.63it/s]
                                                      
{'loss': 0.2689, 'grad_norm': 0.5298690795898438, 'learning_rate': 9.368590189546268e-06, 'epoch': 0.74}

 25%|██▍       | 3270/13317 [33:41<1:42:43,  1.63it/s]
 25%|██▍       | 3271/13317 [33:41<1:42:48,  1.63it/s]
                                                      
{'loss': 0.2173, 'grad_norm': 0.5130891799926758, 'learning_rate': 9.367952503609186e-06, 'epoch': 0.74}

 25%|██▍       | 3271/13317 [33:41<1:42:48,  1.63it/s]
 25%|██▍       | 3272/13317 [33:42<1:42:48,  1.63it/s]
                                                      
{'loss': 0.2874, 'grad_norm': 0.5753807425498962, 'learning_rate': 9.367314517547518e-06, 'epoch': 0.74}

 25%|██▍       | 3272/13317 [33:42<1:42:48,  1.63it/s]
 25%|██▍       | 3273/13317 [33:43<1:42:44,  1.63it/s]
                                                      
{'loss': 0.2209, 'grad_norm': 0.4756157100200653, 'learning_rate': 9.3666762314051e-06, 'epoch': 0.74}

 25%|██▍       | 3273/13317 [33:43<1:42:44,  1.63it/s]
 25%|██▍       | 3274/13317 [33:43<1:42:42,  1.63it/s]
                                                      
{'loss': 0.2719, 'grad_norm': 0.529017984867096, 'learning_rate': 9.36603764522579e-06, 'epoch': 0.74}

 25%|██▍       | 3274/13317 [33:43<1:42:42,  1.63it/s]
 25%|██▍       | 3275/13317 [33:44<1:42:42,  1.63it/s]
                                                      
{'loss': 0.1919, 'grad_norm': 0.4789835810661316, 'learning_rate': 9.365398759053466e-06, 'epoch': 0.74}

 25%|██▍       | 3275/13317 [33:44<1:42:42,  1.63it/s]
 25%|██▍       | 3276/13317 [33:44<1:42:41,  1.63it/s]
                                                      
{'loss': 0.2207, 'grad_norm': 0.46050798892974854, 'learning_rate': 9.364759572932027e-06, 'epoch': 0.74}

 25%|██▍       | 3276/13317 [33:45<1:42:41,  1.63it/s]
 25%|██▍       | 3277/13317 [33:45<1:42:44,  1.63it/s]
                                                      
{'loss': 0.2444, 'grad_norm': 0.549436628818512, 'learning_rate': 9.364120086905388e-06, 'epoch': 0.74}

 25%|██▍       | 3277/13317 [33:45<1:42:44,  1.63it/s]
 25%|██▍       | 3278/13317 [33:46<1:42:40,  1.63it/s]
                                                      
{'loss': 0.2963, 'grad_norm': 0.5907732248306274, 'learning_rate': 9.363480301017493e-06, 'epoch': 0.74}

 25%|██▍       | 3278/13317 [33:46<1:42:40,  1.63it/s]
 25%|██▍       | 3279/13317 [33:46<1:42:40,  1.63it/s]
                                                      
{'loss': 0.2117, 'grad_norm': 0.4991857409477234, 'learning_rate': 9.3628402153123e-06, 'epoch': 0.74}

 25%|██▍       | 3279/13317 [33:46<1:42:40,  1.63it/s]
 25%|██▍       | 3280/13317 [33:47<1:42:39,  1.63it/s]
                                                      
{'loss': 0.1908, 'grad_norm': 0.4613651633262634, 'learning_rate': 9.362199829833786e-06, 'epoch': 0.74}

 25%|██▍       | 3280/13317 [33:47<1:42:39,  1.63it/s]
 25%|██▍       | 3281/13317 [33:48<1:42:40,  1.63it/s]
                                                      
{'loss': 0.2386, 'grad_norm': 0.5277611613273621, 'learning_rate': 9.36155914462596e-06, 'epoch': 0.74}

 25%|██▍       | 3281/13317 [33:48<1:42:40,  1.63it/s]
 25%|██▍       | 3282/13317 [33:48<1:42:42,  1.63it/s]
                                                      
{'loss': 0.2511, 'grad_norm': 0.5255680680274963, 'learning_rate': 9.36091815973284e-06, 'epoch': 0.74}

 25%|██▍       | 3282/13317 [33:48<1:42:42,  1.63it/s]
 25%|██▍       | 3283/13317 [33:49<1:42:40,  1.63it/s]
                                                      
{'loss': 0.2804, 'grad_norm': 0.5905022621154785, 'learning_rate': 9.360276875198467e-06, 'epoch': 0.74}

 25%|██▍       | 3283/13317 [33:49<1:42:40,  1.63it/s]
 25%|██▍       | 3284/13317 [33:49<1:42:37,  1.63it/s]
                                                      
{'loss': 0.286, 'grad_norm': 0.5772977471351624, 'learning_rate': 9.359635291066904e-06, 'epoch': 0.74}

 25%|██▍       | 3284/13317 [33:49<1:42:37,  1.63it/s]
 25%|██▍       | 3285/13317 [33:50<1:42:37,  1.63it/s]
                                                      
{'loss': 0.2313, 'grad_norm': 0.5099091529846191, 'learning_rate': 9.358993407382238e-06, 'epoch': 0.74}

 25%|██▍       | 3285/13317 [33:50<1:42:37,  1.63it/s]
 25%|██▍       | 3286/13317 [33:51<1:42:37,  1.63it/s]
                                                      
{'loss': 0.2258, 'grad_norm': 0.4830155372619629, 'learning_rate': 9.358351224188568e-06, 'epoch': 0.74}

 25%|██▍       | 3286/13317 [33:51<1:42:37,  1.63it/s]
 25%|██▍       | 3287/13317 [33:51<1:42:38,  1.63it/s]
                                                      
{'loss': 0.3403, 'grad_norm': 0.6154513359069824, 'learning_rate': 9.357708741530025e-06, 'epoch': 0.74}

 25%|██▍       | 3287/13317 [33:51<1:42:38,  1.63it/s]
 25%|██▍       | 3288/13317 [33:52<1:42:36,  1.63it/s]
                                                      
{'loss': 0.2738, 'grad_norm': 0.5186351537704468, 'learning_rate': 9.35706595945075e-06, 'epoch': 0.74}

 25%|██▍       | 3288/13317 [33:52<1:42:36,  1.63it/s]
 25%|██▍       | 3289/13317 [33:52<1:42:33,  1.63it/s]
                                                      
{'loss': 0.2259, 'grad_norm': 0.4731694757938385, 'learning_rate': 9.356422877994911e-06, 'epoch': 0.74}

 25%|██▍       | 3289/13317 [33:53<1:42:33,  1.63it/s]
 25%|██▍       | 3290/13317 [33:53<1:42:33,  1.63it/s]
                                                      
{'loss': 0.2265, 'grad_norm': 0.5122162103652954, 'learning_rate': 9.355779497206694e-06, 'epoch': 0.74}

 25%|██▍       | 3290/13317 [33:53<1:42:33,  1.63it/s]
 25%|██▍       | 3291/13317 [33:54<1:42:30,  1.63it/s]
                                                      
{'loss': 0.2085, 'grad_norm': 0.5068756937980652, 'learning_rate': 9.355135817130306e-06, 'epoch': 0.74}

 25%|██▍       | 3291/13317 [33:54<1:42:30,  1.63it/s]
 25%|██▍       | 3292/13317 [33:54<1:42:40,  1.63it/s]
                                                      
{'loss': 0.2615, 'grad_norm': 0.5146831274032593, 'learning_rate': 9.354491837809973e-06, 'epoch': 0.74}

 25%|██▍       | 3292/13317 [33:54<1:42:40,  1.63it/s]
 25%|██▍       | 3293/13317 [33:55<1:42:35,  1.63it/s]
                                                      
{'loss': 0.2968, 'grad_norm': 0.6372454762458801, 'learning_rate': 9.353847559289947e-06, 'epoch': 0.74}

 25%|██▍       | 3293/13317 [33:55<1:42:35,  1.63it/s]
 25%|██▍       | 3294/13317 [33:55<1:42:33,  1.63it/s]
                                                      
{'loss': 0.2743, 'grad_norm': 0.5453668236732483, 'learning_rate': 9.353202981614491e-06, 'epoch': 0.74}

 25%|██▍       | 3294/13317 [33:56<1:42:33,  1.63it/s]
 25%|██▍       | 3295/13317 [33:56<1:42:28,  1.63it/s]
                                                      
{'loss': 0.2577, 'grad_norm': 0.5378718376159668, 'learning_rate': 9.3525581048279e-06, 'epoch': 0.74}

 25%|██▍       | 3295/13317 [33:56<1:42:28,  1.63it/s]
 25%|██▍       | 3296/13317 [33:57<1:42:25,  1.63it/s]
                                                      
{'loss': 0.2988, 'grad_norm': 0.6403523087501526, 'learning_rate': 9.351912928974481e-06, 'epoch': 0.74}

 25%|██▍       | 3296/13317 [33:57<1:42:25,  1.63it/s]
 25%|██▍       | 3297/13317 [33:57<1:42:28,  1.63it/s]
                                                      
{'loss': 0.3176, 'grad_norm': 0.6303818225860596, 'learning_rate': 9.351267454098565e-06, 'epoch': 0.74}

 25%|██▍       | 3297/13317 [33:57<1:42:28,  1.63it/s]
 25%|██▍       | 3298/13317 [33:58<1:42:27,  1.63it/s]
                                                      
{'loss': 0.2199, 'grad_norm': 0.5141533613204956, 'learning_rate': 9.350621680244503e-06, 'epoch': 0.74}

 25%|██▍       | 3298/13317 [33:58<1:42:27,  1.63it/s]
 25%|██▍       | 3299/13317 [33:59<1:42:24,  1.63it/s]
                                                      
{'loss': 0.2447, 'grad_norm': 0.5360631942749023, 'learning_rate': 9.349975607456668e-06, 'epoch': 0.74}

 25%|██▍       | 3299/13317 [33:59<1:42:24,  1.63it/s]
 25%|██▍       | 3300/13317 [33:59<1:42:29,  1.63it/s]
                                                      
{'loss': 0.1936, 'grad_norm': 0.4186443090438843, 'learning_rate': 9.349329235779447e-06, 'epoch': 0.74}

 25%|██▍       | 3300/13317 [33:59<1:42:29,  1.63it/s]
 25%|██▍       | 3301/13317 [34:00<1:42:28,  1.63it/s]
                                                      
{'loss': 0.2084, 'grad_norm': 0.5200955867767334, 'learning_rate': 9.34868256525726e-06, 'epoch': 0.74}

 25%|██▍       | 3301/13317 [34:00<1:42:28,  1.63it/s]
 25%|██▍       | 3302/13317 [34:00<1:42:38,  1.63it/s]
                                                      
{'loss': 0.2229, 'grad_norm': 0.49868443608283997, 'learning_rate': 9.348035595934533e-06, 'epoch': 0.74}

 25%|██▍       | 3302/13317 [34:01<1:42:38,  1.63it/s]
 25%|██▍       | 3303/13317 [34:01<1:42:31,  1.63it/s]
                                                      
{'loss': 0.3191, 'grad_norm': 0.5609248280525208, 'learning_rate': 9.347388327855724e-06, 'epoch': 0.74}

 25%|██▍       | 3303/13317 [34:01<1:42:31,  1.63it/s]
 25%|██▍       | 3304/13317 [34:02<1:42:30,  1.63it/s]
                                                      
{'loss': 0.2809, 'grad_norm': 0.5198149085044861, 'learning_rate': 9.346740761065306e-06, 'epoch': 0.74}

 25%|██▍       | 3304/13317 [34:02<1:42:30,  1.63it/s]
 25%|██▍       | 3305/13317 [34:02<1:42:31,  1.63it/s]
                                                      
{'loss': 0.2334, 'grad_norm': 0.490193247795105, 'learning_rate': 9.346092895607774e-06, 'epoch': 0.74}

 25%|██▍       | 3305/13317 [34:02<1:42:31,  1.63it/s]
 25%|██▍       | 3306/13317 [34:03<1:42:26,  1.63it/s]
                                                      
{'loss': 0.3227, 'grad_norm': 0.6197907328605652, 'learning_rate': 9.345444731527642e-06, 'epoch': 0.74}

 25%|██▍       | 3306/13317 [34:03<1:42:26,  1.63it/s]
 25%|██▍       | 3307/13317 [34:03<1:42:38,  1.63it/s]
                                                      
{'loss': 0.2205, 'grad_norm': 0.5482670664787292, 'learning_rate': 9.344796268869447e-06, 'epoch': 0.74}

 25%|██▍       | 3307/13317 [34:04<1:42:38,  1.63it/s]
 25%|██▍       | 3308/13317 [34:04<1:42:36,  1.63it/s]
                                                      
{'loss': 0.2806, 'grad_norm': 0.5755503177642822, 'learning_rate': 9.344147507677746e-06, 'epoch': 0.75}

 25%|██▍       | 3308/13317 [34:04<1:42:36,  1.63it/s]
 25%|██▍       | 3309/13317 [34:05<1:42:27,  1.63it/s]
                                                      
{'loss': 0.2446, 'grad_norm': 0.5110052227973938, 'learning_rate': 9.343498447997114e-06, 'epoch': 0.75}

 25%|██▍       | 3309/13317 [34:05<1:42:27,  1.63it/s]
 25%|██▍       | 3310/13317 [34:05<1:42:32,  1.63it/s]
                                                      
{'loss': 0.2638, 'grad_norm': 0.6136289238929749, 'learning_rate': 9.342849089872148e-06, 'epoch': 0.75}

 25%|██▍       | 3310/13317 [34:05<1:42:32,  1.63it/s]
 25%|██▍       | 3311/13317 [34:06<1:42:28,  1.63it/s]
                                                      
{'loss': 0.3008, 'grad_norm': 0.6531462073326111, 'learning_rate': 9.342199433347467e-06, 'epoch': 0.75}

 25%|██▍       | 3311/13317 [34:06<1:42:28,  1.63it/s]
 25%|██▍       | 3312/13317 [34:07<1:42:23,  1.63it/s]
                                                      
{'loss': 0.2572, 'grad_norm': 0.6020426154136658, 'learning_rate': 9.34154947846771e-06, 'epoch': 0.75}

 25%|██▍       | 3312/13317 [34:07<1:42:23,  1.63it/s]
 25%|██▍       | 3313/13317 [34:07<1:42:24,  1.63it/s]
                                                      
{'loss': 0.3079, 'grad_norm': 0.6200039386749268, 'learning_rate': 9.340899225277534e-06, 'epoch': 0.75}

 25%|██▍       | 3313/13317 [34:07<1:42:24,  1.63it/s]
 25%|██▍       | 3314/13317 [34:08<1:42:22,  1.63it/s]
                                                      
{'loss': 0.2746, 'grad_norm': 0.5376090407371521, 'learning_rate': 9.340248673821618e-06, 'epoch': 0.75}

 25%|██▍       | 3314/13317 [34:08<1:42:22,  1.63it/s]
 25%|██▍       | 3315/13317 [34:08<1:42:28,  1.63it/s]
                                                      
{'loss': 0.3206, 'grad_norm': 0.575253427028656, 'learning_rate': 9.339597824144667e-06, 'epoch': 0.75}

 25%|██▍       | 3315/13317 [34:09<1:42:28,  1.63it/s]
 25%|██▍       | 3316/13317 [34:09<1:42:28,  1.63it/s]
                                                      
{'loss': 0.2202, 'grad_norm': 0.497102290391922, 'learning_rate': 9.338946676291392e-06, 'epoch': 0.75}

 25%|██▍       | 3316/13317 [34:09<1:42:28,  1.63it/s]
 25%|██▍       | 3317/13317 [34:10<1:42:23,  1.63it/s]
                                                      
{'loss': 0.2845, 'grad_norm': 0.5809880495071411, 'learning_rate': 9.338295230306542e-06, 'epoch': 0.75}

 25%|██▍       | 3317/13317 [34:10<1:42:23,  1.63it/s]
 25%|██▍       | 3318/13317 [34:10<1:42:21,  1.63it/s]
                                                      
{'loss': 0.2629, 'grad_norm': 0.5378087759017944, 'learning_rate': 9.337643486234874e-06, 'epoch': 0.75}

 25%|██▍       | 3318/13317 [34:10<1:42:21,  1.63it/s]
 25%|██▍       | 3319/13317 [34:11<1:42:19,  1.63it/s]
                                                      
{'loss': 0.2448, 'grad_norm': 0.5480355024337769, 'learning_rate': 9.336991444121171e-06, 'epoch': 0.75}

 25%|██▍       | 3319/13317 [34:11<1:42:19,  1.63it/s]
 25%|██▍       | 3320/13317 [34:11<1:42:25,  1.63it/s]
                                                      
{'loss': 0.2311, 'grad_norm': 0.54151850938797, 'learning_rate': 9.336339104010234e-06, 'epoch': 0.75}

 25%|██▍       | 3320/13317 [34:12<1:42:25,  1.63it/s]
 25%|██▍       | 3321/13317 [34:12<1:42:16,  1.63it/s]
                                                      
{'loss': 0.2373, 'grad_norm': 0.55311518907547, 'learning_rate': 9.335686465946888e-06, 'epoch': 0.75}

 25%|██▍       | 3321/13317 [34:12<1:42:16,  1.63it/s]
 25%|██▍       | 3322/13317 [34:13<1:42:16,  1.63it/s]
                                                      
{'loss': 0.2531, 'grad_norm': 0.5773836374282837, 'learning_rate': 9.335033529975974e-06, 'epoch': 0.75}

 25%|██▍       | 3322/13317 [34:13<1:42:16,  1.63it/s]
 25%|██▍       | 3323/13317 [34:13<1:42:17,  1.63it/s]
                                                      
{'loss': 0.251, 'grad_norm': 0.47005417943000793, 'learning_rate': 9.334380296142356e-06, 'epoch': 0.75}

 25%|██▍       | 3323/13317 [34:13<1:42:17,  1.63it/s]
 25%|██▍       | 3324/13317 [34:14<1:42:13,  1.63it/s]
                                                      
{'loss': 0.233, 'grad_norm': 0.4805987477302551, 'learning_rate': 9.333726764490921e-06, 'epoch': 0.75}

 25%|██▍       | 3324/13317 [34:14<1:42:13,  1.63it/s]
 25%|██▍       | 3325/13317 [34:15<1:42:39,  1.62it/s]
                                                      
{'loss': 0.2271, 'grad_norm': 0.5108808279037476, 'learning_rate': 9.333072935066568e-06, 'epoch': 0.75}

 25%|██▍       | 3325/13317 [34:15<1:42:39,  1.62it/s]
 25%|██▍       | 3326/13317 [34:15<1:42:28,  1.62it/s]
                                                      
{'loss': 0.2759, 'grad_norm': 0.5958229899406433, 'learning_rate': 9.332418807914226e-06, 'epoch': 0.75}

 25%|██▍       | 3326/13317 [34:15<1:42:28,  1.62it/s]
 25%|██▍       | 3327/13317 [34:16<1:42:24,  1.63it/s]
                                                      
{'loss': 0.2227, 'grad_norm': 0.4667026996612549, 'learning_rate': 9.33176438307884e-06, 'epoch': 0.75}

 25%|██▍       | 3327/13317 [34:16<1:42:24,  1.63it/s]
 25%|██▍       | 3328/13317 [34:16<1:42:19,  1.63it/s]
                                                      
{'loss': 0.2922, 'grad_norm': 0.5591367483139038, 'learning_rate': 9.331109660605376e-06, 'epoch': 0.75}

 25%|██▍       | 3328/13317 [34:17<1:42:19,  1.63it/s]
 25%|██▍       | 3329/13317 [34:17<1:42:20,  1.63it/s]
                                                      
{'loss': 0.3109, 'grad_norm': 0.5957015752792358, 'learning_rate': 9.330454640538818e-06, 'epoch': 0.75}

 25%|██▍       | 3329/13317 [34:17<1:42:20,  1.63it/s]
 25%|██▌       | 3330/13317 [34:18<1:42:48,  1.62it/s]
                                                      
{'loss': 0.3093, 'grad_norm': 0.547873318195343, 'learning_rate': 9.329799322924176e-06, 'epoch': 0.75}

 25%|██▌       | 3330/13317 [34:18<1:42:48,  1.62it/s]
 25%|██▌       | 3331/13317 [34:18<1:42:37,  1.62it/s]
                                                      
{'loss': 0.2013, 'grad_norm': 0.4417094886302948, 'learning_rate': 9.329143707806477e-06, 'epoch': 0.75}

 25%|██▌       | 3331/13317 [34:18<1:42:37,  1.62it/s]
 25%|██▌       | 3332/13317 [34:19<1:42:37,  1.62it/s]
                                                      
{'loss': 0.1998, 'grad_norm': 0.5171881318092346, 'learning_rate': 9.328487795230765e-06, 'epoch': 0.75}

 25%|██▌       | 3332/13317 [34:19<1:42:37,  1.62it/s]
 25%|██▌       | 3333/13317 [34:19<1:42:25,  1.62it/s]
                                                      
{'loss': 0.2466, 'grad_norm': 0.5360876321792603, 'learning_rate': 9.327831585242111e-06, 'epoch': 0.75}

 25%|██▌       | 3333/13317 [34:20<1:42:25,  1.62it/s]
 25%|██▌       | 3334/13317 [34:20<1:42:24,  1.62it/s]
                                                      
{'loss': 0.2452, 'grad_norm': 0.610788106918335, 'learning_rate': 9.327175077885604e-06, 'epoch': 0.75}

 25%|██▌       | 3334/13317 [34:20<1:42:24,  1.62it/s]
 25%|██▌       | 3335/13317 [34:21<1:42:28,  1.62it/s]
                                                      
{'loss': 0.279, 'grad_norm': 0.5677272081375122, 'learning_rate': 9.326518273206353e-06, 'epoch': 0.75}

 25%|██▌       | 3335/13317 [34:21<1:42:28,  1.62it/s]
 25%|██▌       | 3336/13317 [34:21<1:42:18,  1.63it/s]
                                                      
{'loss': 0.1833, 'grad_norm': 0.4535437226295471, 'learning_rate': 9.325861171249486e-06, 'epoch': 0.75}

 25%|██▌       | 3336/13317 [34:21<1:42:18,  1.63it/s]
 25%|██▌       | 3337/13317 [34:22<1:42:25,  1.62it/s]
                                                      
{'loss': 0.254, 'grad_norm': 0.4840242862701416, 'learning_rate': 9.325203772060154e-06, 'epoch': 0.75}

 25%|██▌       | 3337/13317 [34:22<1:42:25,  1.62it/s]
 25%|██▌       | 3338/13317 [34:23<1:42:17,  1.63it/s]
                                                      
{'loss': 0.2764, 'grad_norm': 0.5609190464019775, 'learning_rate': 9.324546075683525e-06, 'epoch': 0.75}

 25%|██▌       | 3338/13317 [34:23<1:42:17,  1.63it/s]
 25%|██▌       | 3339/13317 [34:23<1:42:16,  1.63it/s]
                                                      
{'loss': 0.2386, 'grad_norm': 0.5557900667190552, 'learning_rate': 9.323888082164794e-06, 'epoch': 0.75}

 25%|██▌       | 3339/13317 [34:23<1:42:16,  1.63it/s]
 25%|██▌       | 3340/13317 [34:24<1:42:20,  1.62it/s]
                                                      
{'loss': 0.2168, 'grad_norm': 0.4668806493282318, 'learning_rate': 9.323229791549169e-06, 'epoch': 0.75}

 25%|██▌       | 3340/13317 [34:24<1:42:20,  1.62it/s]
 25%|██▌       | 3341/13317 [34:24<1:42:15,  1.63it/s]
                                                      
{'loss': 0.2419, 'grad_norm': 0.5513315200805664, 'learning_rate': 9.322571203881881e-06, 'epoch': 0.75}

 25%|██▌       | 3341/13317 [34:25<1:42:15,  1.63it/s]
 25%|██▌       | 3342/13317 [34:25<1:42:16,  1.63it/s]
                                                      
{'loss': 0.3197, 'grad_norm': 0.7003463506698608, 'learning_rate': 9.321912319208184e-06, 'epoch': 0.75}

 25%|██▌       | 3342/13317 [34:25<1:42:16,  1.63it/s]
 25%|██▌       | 3343/13317 [34:26<1:42:11,  1.63it/s]
                                                      
{'loss': 0.1746, 'grad_norm': 0.4503879249095917, 'learning_rate': 9.321253137573351e-06, 'epoch': 0.75}

 25%|██▌       | 3343/13317 [34:26<1:42:11,  1.63it/s]
 25%|██▌       | 3344/13317 [34:26<1:42:15,  1.63it/s]
                                                      
{'loss': 0.2638, 'grad_norm': 0.529470682144165, 'learning_rate': 9.320593659022673e-06, 'epoch': 0.75}

 25%|██▌       | 3344/13317 [34:26<1:42:15,  1.63it/s]
 25%|██▌       | 3345/13317 [34:27<1:42:17,  1.62it/s]
                                                      
{'loss': 0.2531, 'grad_norm': 0.5681689977645874, 'learning_rate': 9.31993388360146e-06, 'epoch': 0.75}

 25%|██▌       | 3345/13317 [34:27<1:42:17,  1.62it/s]
 25%|██▌       | 3346/13317 [34:27<1:42:15,  1.63it/s]
                                                      
{'loss': 0.2319, 'grad_norm': 0.49720311164855957, 'learning_rate': 9.319273811355052e-06, 'epoch': 0.75}

 25%|██▌       | 3346/13317 [34:28<1:42:15,  1.63it/s]
 25%|██▌       | 3347/13317 [34:28<1:42:22,  1.62it/s]
                                                      
{'loss': 0.2582, 'grad_norm': 0.5618807673454285, 'learning_rate': 9.318613442328799e-06, 'epoch': 0.75}

 25%|██▌       | 3347/13317 [34:28<1:42:22,  1.62it/s]
 25%|██▌       | 3348/13317 [34:29<1:42:13,  1.63it/s]
                                                      
{'loss': 0.2362, 'grad_norm': 0.5201533436775208, 'learning_rate': 9.317952776568078e-06, 'epoch': 0.75}

 25%|██▌       | 3348/13317 [34:29<1:42:13,  1.63it/s]
 25%|██▌       | 3349/13317 [34:29<1:42:16,  1.62it/s]
                                                      
{'loss': 0.3104, 'grad_norm': 0.5647103786468506, 'learning_rate': 9.31729181411828e-06, 'epoch': 0.75}

 25%|██▌       | 3349/13317 [34:29<1:42:16,  1.62it/s]
 25%|██▌       | 3350/13317 [34:30<1:42:19,  1.62it/s]
                                                      
{'loss': 0.2816, 'grad_norm': 0.5374722480773926, 'learning_rate': 9.316630555024822e-06, 'epoch': 0.75}

 25%|██▌       | 3350/13317 [34:30<1:42:19,  1.62it/s]
 25%|██▌       | 3351/13317 [34:31<1:42:10,  1.63it/s]
                                                      
{'loss': 0.2226, 'grad_norm': 0.5662817358970642, 'learning_rate': 9.315968999333142e-06, 'epoch': 0.75}

 25%|██▌       | 3351/13317 [34:31<1:42:10,  1.63it/s]
 25%|██▌       | 3352/13317 [34:31<1:42:10,  1.63it/s]
                                                      
{'loss': 0.2923, 'grad_norm': 0.5697516202926636, 'learning_rate': 9.315307147088692e-06, 'epoch': 0.76}

 25%|██▌       | 3352/13317 [34:31<1:42:10,  1.63it/s]
 25%|██▌       | 3353/13317 [34:32<1:42:05,  1.63it/s]
                                                      
{'loss': 0.2478, 'grad_norm': 0.4702986180782318, 'learning_rate': 9.31464499833695e-06, 'epoch': 0.76}

 25%|██▌       | 3353/13317 [34:32<1:42:05,  1.63it/s]
 25%|██▌       | 3354/13317 [34:32<1:42:08,  1.63it/s]
                                                      
{'loss': 0.2614, 'grad_norm': 0.49667733907699585, 'learning_rate': 9.31398255312341e-06, 'epoch': 0.76}

 25%|██▌       | 3354/13317 [34:33<1:42:08,  1.63it/s]
 25%|██▌       | 3355/13317 [34:33<1:42:10,  1.62it/s]
                                                      
{'loss': 0.2324, 'grad_norm': 0.5429592728614807, 'learning_rate': 9.313319811493596e-06, 'epoch': 0.76}

 25%|██▌       | 3355/13317 [34:33<1:42:10,  1.62it/s]
 25%|██▌       | 3356/13317 [34:34<1:42:05,  1.63it/s]
                                                      
{'loss': 0.2131, 'grad_norm': 0.5193588733673096, 'learning_rate': 9.312656773493038e-06, 'epoch': 0.76}

 25%|██▌       | 3356/13317 [34:34<1:42:05,  1.63it/s]
 25%|██▌       | 3357/13317 [34:34<1:42:06,  1.63it/s]
                                                      
{'loss': 0.2513, 'grad_norm': 0.6076443791389465, 'learning_rate': 9.311993439167297e-06, 'epoch': 0.76}

 25%|██▌       | 3357/13317 [34:34<1:42:06,  1.63it/s]
 25%|██▌       | 3358/13317 [34:35<1:42:00,  1.63it/s]
                                                      
{'loss': 0.2742, 'grad_norm': 0.5701371431350708, 'learning_rate': 9.31132980856195e-06, 'epoch': 0.76}

 25%|██▌       | 3358/13317 [34:35<1:42:00,  1.63it/s]
 25%|██▌       | 3359/13317 [34:35<1:41:55,  1.63it/s]
                                                      
{'loss': 0.262, 'grad_norm': 0.5379364490509033, 'learning_rate': 9.310665881722597e-06, 'epoch': 0.76}

 25%|██▌       | 3359/13317 [34:36<1:41:55,  1.63it/s]
 25%|██▌       | 3360/13317 [34:36<1:41:58,  1.63it/s]
                                                      
{'loss': 0.2468, 'grad_norm': 0.5438183546066284, 'learning_rate': 9.310001658694856e-06, 'epoch': 0.76}

 25%|██▌       | 3360/13317 [34:36<1:41:58,  1.63it/s]
 25%|██▌       | 3361/13317 [34:37<1:41:49,  1.63it/s]
                                                      
{'loss': 0.2158, 'grad_norm': 0.4597002863883972, 'learning_rate': 9.309337139524365e-06, 'epoch': 0.76}

 25%|██▌       | 3361/13317 [34:37<1:41:49,  1.63it/s]
 25%|██▌       | 3362/13317 [34:37<1:41:50,  1.63it/s]
                                                      
{'loss': 0.3662, 'grad_norm': 0.7021230459213257, 'learning_rate': 9.308672324256785e-06, 'epoch': 0.76}

 25%|██▌       | 3362/13317 [34:37<1:41:50,  1.63it/s]
 25%|██▌       | 3363/13317 [34:38<1:41:51,  1.63it/s]
                                                      
{'loss': 0.2091, 'grad_norm': 0.4570665657520294, 'learning_rate': 9.308007212937795e-06, 'epoch': 0.76}

 25%|██▌       | 3363/13317 [34:38<1:41:51,  1.63it/s]
 25%|██▌       | 3364/13317 [34:39<1:41:46,  1.63it/s]
                                                      
{'loss': 0.1892, 'grad_norm': 0.43364012241363525, 'learning_rate': 9.307341805613096e-06, 'epoch': 0.76}

 25%|██▌       | 3364/13317 [34:39<1:41:46,  1.63it/s]
 25%|██▌       | 3365/13317 [34:39<1:41:55,  1.63it/s]
                                                      
{'loss': 0.3044, 'grad_norm': 0.6414267420768738, 'learning_rate': 9.306676102328408e-06, 'epoch': 0.76}

 25%|██▌       | 3365/13317 [34:39<1:41:55,  1.63it/s]
 25%|██▌       | 3366/13317 [34:40<1:41:53,  1.63it/s]
                                                      
{'loss': 0.2818, 'grad_norm': 0.576647937297821, 'learning_rate': 9.306010103129473e-06, 'epoch': 0.76}

 25%|██▌       | 3366/13317 [34:40<1:41:53,  1.63it/s]
 25%|██▌       | 3367/13317 [34:40<1:41:48,  1.63it/s]
                                                      
{'loss': 0.2249, 'grad_norm': 0.5426637530326843, 'learning_rate': 9.30534380806205e-06, 'epoch': 0.76}

 25%|██▌       | 3367/13317 [34:40<1:41:48,  1.63it/s]
 25%|██▌       | 3368/13317 [34:41<1:41:47,  1.63it/s]
                                                      
{'loss': 0.1924, 'grad_norm': 0.46103301644325256, 'learning_rate': 9.304677217171922e-06, 'epoch': 0.76}

 25%|██▌       | 3368/13317 [34:41<1:41:47,  1.63it/s]
 25%|██▌       | 3369/13317 [34:42<1:41:44,  1.63it/s]
                                                      
{'loss': 0.3576, 'grad_norm': 0.6485114693641663, 'learning_rate': 9.30401033050489e-06, 'epoch': 0.76}

 25%|██▌       | 3369/13317 [34:42<1:41:44,  1.63it/s]
 25%|██▌       | 3370/13317 [34:42<1:41:47,  1.63it/s]
                                                      
{'loss': 0.3442, 'grad_norm': 0.6198927164077759, 'learning_rate': 9.303343148106778e-06, 'epoch': 0.76}

 25%|██▌       | 3370/13317 [34:42<1:41:47,  1.63it/s]
 25%|██▌       | 3371/13317 [34:43<1:41:46,  1.63it/s]
                                                      
{'loss': 0.246, 'grad_norm': 0.4863899350166321, 'learning_rate': 9.302675670023425e-06, 'epoch': 0.76}

 25%|██▌       | 3371/13317 [34:43<1:41:46,  1.63it/s]
 25%|██▌       | 3372/13317 [34:43<1:41:56,  1.63it/s]
                                                      
{'loss': 0.2682, 'grad_norm': 0.5177212953567505, 'learning_rate': 9.302007896300697e-06, 'epoch': 0.76}

 25%|██▌       | 3372/13317 [34:44<1:41:56,  1.63it/s]
 25%|██▌       | 3373/13317 [34:44<1:41:51,  1.63it/s]
                                                      
{'loss': 0.3095, 'grad_norm': 0.6270712614059448, 'learning_rate': 9.301339826984477e-06, 'epoch': 0.76}

 25%|██▌       | 3373/13317 [34:44<1:41:51,  1.63it/s]
 25%|██▌       | 3374/13317 [34:45<1:41:47,  1.63it/s]
                                                      
{'loss': 0.2379, 'grad_norm': 0.5084636211395264, 'learning_rate': 9.300671462120667e-06, 'epoch': 0.76}

 25%|██▌       | 3374/13317 [34:45<1:41:47,  1.63it/s]
 25%|██▌       | 3375/13317 [34:45<1:41:55,  1.63it/s]
                                                      
{'loss': 0.2557, 'grad_norm': 0.4901258051395416, 'learning_rate': 9.300002801755192e-06, 'epoch': 0.76}

 25%|██▌       | 3375/13317 [34:45<1:41:55,  1.63it/s]
 25%|██▌       | 3376/13317 [34:46<1:41:51,  1.63it/s]
                                                      
{'loss': 0.2062, 'grad_norm': 0.46588194370269775, 'learning_rate': 9.299333845933994e-06, 'epoch': 0.76}

 25%|██▌       | 3376/13317 [34:46<1:41:51,  1.63it/s]
 25%|██▌       | 3377/13317 [34:47<1:41:48,  1.63it/s]
                                                      
{'loss': 0.2875, 'grad_norm': 0.5555844902992249, 'learning_rate': 9.29866459470304e-06, 'epoch': 0.76}

 25%|██▌       | 3377/13317 [34:47<1:41:48,  1.63it/s]
 25%|██▌       | 3378/13317 [34:47<1:41:44,  1.63it/s]
                                                      
{'loss': 0.2952, 'grad_norm': 0.6022647023200989, 'learning_rate': 9.297995048108313e-06, 'epoch': 0.76}

 25%|██▌       | 3378/13317 [34:47<1:41:44,  1.63it/s]
 25%|██▌       | 3379/13317 [34:48<1:41:46,  1.63it/s]
                                                      
{'loss': 0.256, 'grad_norm': 0.5391560196876526, 'learning_rate': 9.297325206195817e-06, 'epoch': 0.76}

 25%|██▌       | 3379/13317 [34:48<1:41:46,  1.63it/s]
 25%|██▌       | 3380/13317 [34:48<1:41:52,  1.63it/s]
                                                      
{'loss': 0.2698, 'grad_norm': 0.554832935333252, 'learning_rate': 9.29665506901158e-06, 'epoch': 0.76}

 25%|██▌       | 3380/13317 [34:48<1:41:52,  1.63it/s]
 25%|██▌       | 3381/13317 [34:49<1:41:47,  1.63it/s]
                                                      
{'loss': 0.2676, 'grad_norm': 0.5493577122688293, 'learning_rate': 9.295984636601645e-06, 'epoch': 0.76}

 25%|██▌       | 3381/13317 [34:49<1:41:47,  1.63it/s]
 25%|██▌       | 3382/13317 [34:50<1:41:43,  1.63it/s]
                                                      
{'loss': 0.2683, 'grad_norm': 0.5571169257164001, 'learning_rate': 9.29531390901208e-06, 'epoch': 0.76}

 25%|██▌       | 3382/13317 [34:50<1:41:43,  1.63it/s]
 25%|██▌       | 3383/13317 [34:50<1:41:40,  1.63it/s]
                                                      
{'loss': 0.303, 'grad_norm': 0.549866259098053, 'learning_rate': 9.29464288628897e-06, 'epoch': 0.76}

 25%|██▌       | 3383/13317 [34:50<1:41:40,  1.63it/s]
 25%|██▌       | 3384/13317 [34:51<1:41:41,  1.63it/s]
                                                      
{'loss': 0.3267, 'grad_norm': 0.6561495661735535, 'learning_rate': 9.293971568478419e-06, 'epoch': 0.76}

 25%|██▌       | 3384/13317 [34:51<1:41:41,  1.63it/s]
 25%|██▌       | 3385/13317 [34:51<1:41:45,  1.63it/s]
                                                      
{'loss': 0.2313, 'grad_norm': 0.4814583957195282, 'learning_rate': 9.293299955626557e-06, 'epoch': 0.76}

 25%|██▌       | 3385/13317 [34:52<1:41:45,  1.63it/s]
 25%|██▌       | 3386/13317 [34:52<1:41:44,  1.63it/s]
                                                      
{'loss': 0.2752, 'grad_norm': 0.5246928334236145, 'learning_rate': 9.292628047779532e-06, 'epoch': 0.76}

 25%|██▌       | 3386/13317 [34:52<1:41:44,  1.63it/s]
 25%|██▌       | 3387/13317 [34:53<1:41:46,  1.63it/s]
                                                      
{'loss': 0.3437, 'grad_norm': 0.6478704810142517, 'learning_rate': 9.291955844983507e-06, 'epoch': 0.76}

 25%|██▌       | 3387/13317 [34:53<1:41:46,  1.63it/s]
 25%|██▌       | 3388/13317 [34:53<1:41:41,  1.63it/s]
                                                      
{'loss': 0.2459, 'grad_norm': 0.49957555532455444, 'learning_rate': 9.291283347284673e-06, 'epoch': 0.76}

 25%|██▌       | 3388/13317 [34:53<1:41:41,  1.63it/s]
 25%|██▌       | 3389/13317 [34:54<1:41:51,  1.62it/s]
                                                      
{'loss': 0.3095, 'grad_norm': 0.587152361869812, 'learning_rate': 9.290610554729234e-06, 'epoch': 0.76}

 25%|██▌       | 3389/13317 [34:54<1:41:51,  1.62it/s]
 25%|██▌       | 3390/13317 [34:54<1:41:47,  1.63it/s]
                                                      
{'loss': 0.1885, 'grad_norm': 0.43126752972602844, 'learning_rate': 9.289937467363422e-06, 'epoch': 0.76}

 25%|██▌       | 3390/13317 [34:55<1:41:47,  1.63it/s]
 25%|██▌       | 3391/13317 [34:55<1:41:44,  1.63it/s]
                                                      
{'loss': 0.2056, 'grad_norm': 0.4759160876274109, 'learning_rate': 9.289264085233482e-06, 'epoch': 0.76}

 25%|██▌       | 3391/13317 [34:55<1:41:44,  1.63it/s]
 25%|██▌       | 3392/13317 [34:56<1:41:45,  1.63it/s]
                                                      
{'loss': 0.4009, 'grad_norm': 0.6495324969291687, 'learning_rate': 9.288590408385686e-06, 'epoch': 0.76}

 25%|██▌       | 3392/13317 [34:56<1:41:45,  1.63it/s]
 25%|██▌       | 3393/13317 [34:56<1:41:39,  1.63it/s]
                                                      
{'loss': 0.2816, 'grad_norm': 0.5114995241165161, 'learning_rate': 9.28791643686632e-06, 'epoch': 0.76}

 25%|██▌       | 3393/13317 [34:56<1:41:39,  1.63it/s]
 25%|██▌       | 3394/13317 [34:57<1:41:45,  1.63it/s]
                                                      
{'loss': 0.2187, 'grad_norm': 0.4772107005119324, 'learning_rate': 9.287242170721692e-06, 'epoch': 0.76}

 25%|██▌       | 3394/13317 [34:57<1:41:45,  1.63it/s]
 25%|██▌       | 3395/13317 [34:58<1:41:47,  1.62it/s]
                                                      
{'loss': 0.289, 'grad_norm': 0.5278670787811279, 'learning_rate': 9.286567609998134e-06, 'epoch': 0.76}

 25%|██▌       | 3395/13317 [34:58<1:41:47,  1.62it/s]
 26%|██▌       | 3396/13317 [34:58<1:41:38,  1.63it/s]
                                                      
{'loss': 0.2966, 'grad_norm': 0.5524573922157288, 'learning_rate': 9.285892754741994e-06, 'epoch': 0.77}

 26%|██▌       | 3396/13317 [34:58<1:41:38,  1.63it/s]
 26%|██▌       | 3397/13317 [34:59<1:41:34,  1.63it/s]
                                                      
{'loss': 0.25, 'grad_norm': 0.6104183197021484, 'learning_rate': 9.28521760499964e-06, 'epoch': 0.77}

 26%|██▌       | 3397/13317 [34:59<1:41:34,  1.63it/s]
 26%|██▌       | 3398/13317 [34:59<1:41:34,  1.63it/s]
                                                      
{'loss': 0.1469, 'grad_norm': 0.4179726839065552, 'learning_rate': 9.284542160817468e-06, 'epoch': 0.77}

 26%|██▌       | 3398/13317 [35:00<1:41:34,  1.63it/s]
 26%|██▌       | 3399/13317 [35:00<1:41:33,  1.63it/s]
                                                      
{'loss': 0.2876, 'grad_norm': 0.6031035780906677, 'learning_rate': 9.283866422241884e-06, 'epoch': 0.77}

 26%|██▌       | 3399/13317 [35:00<1:41:33,  1.63it/s]
 26%|██▌       | 3400/13317 [35:01<1:41:35,  1.63it/s]
                                                      
{'loss': 0.2834, 'grad_norm': 0.6833616495132446, 'learning_rate': 9.283190389319315e-06, 'epoch': 0.77}

 26%|██▌       | 3400/13317 [35:01<1:41:35,  1.63it/s]
 26%|██▌       | 3401/13317 [35:01<1:41:32,  1.63it/s]
                                                      
{'loss': 0.2396, 'grad_norm': 0.5227854251861572, 'learning_rate': 9.282514062096215e-06, 'epoch': 0.77}

 26%|██▌       | 3401/13317 [35:01<1:41:32,  1.63it/s]
 26%|██▌       | 3402/13317 [35:02<1:41:35,  1.63it/s]
                                                      
{'loss': 0.2947, 'grad_norm': 0.6020106077194214, 'learning_rate': 9.281837440619058e-06, 'epoch': 0.77}

 26%|██▌       | 3402/13317 [35:02<1:41:35,  1.63it/s]
 26%|██▌       | 3403/13317 [35:02<1:41:27,  1.63it/s]
                                                      
{'loss': 0.2488, 'grad_norm': 0.6557766199111938, 'learning_rate': 9.281160524934331e-06, 'epoch': 0.77}

 26%|██▌       | 3403/13317 [35:03<1:41:27,  1.63it/s]
 26%|██▌       | 3404/13317 [35:03<1:41:26,  1.63it/s]
                                                      
{'loss': 0.2698, 'grad_norm': 0.5501858592033386, 'learning_rate': 9.280483315088545e-06, 'epoch': 0.77}

 26%|██▌       | 3404/13317 [35:03<1:41:26,  1.63it/s]
 26%|██▌       | 3405/13317 [35:04<1:41:31,  1.63it/s]
                                                      
{'loss': 0.1969, 'grad_norm': 0.49970105290412903, 'learning_rate': 9.279805811128235e-06, 'epoch': 0.77}

 26%|██▌       | 3405/13317 [35:04<1:41:31,  1.63it/s]
 26%|██▌       | 3406/13317 [35:04<1:41:29,  1.63it/s]
                                                      
{'loss': 0.2268, 'grad_norm': 0.5084642767906189, 'learning_rate': 9.279128013099948e-06, 'epoch': 0.77}

 26%|██▌       | 3406/13317 [35:04<1:41:29,  1.63it/s]
 26%|██▌       | 3407/13317 [35:05<1:41:39,  1.62it/s]
                                                      
{'loss': 0.2727, 'grad_norm': 0.5627431273460388, 'learning_rate': 9.278449921050259e-06, 'epoch': 0.77}

 26%|██▌       | 3407/13317 [35:05<1:41:39,  1.62it/s]
 26%|██▌       | 3408/13317 [35:06<1:41:32,  1.63it/s]
                                                      
{'loss': 0.2387, 'grad_norm': 0.5230651497840881, 'learning_rate': 9.277771535025761e-06, 'epoch': 0.77}

 26%|██▌       | 3408/13317 [35:06<1:41:32,  1.63it/s]
 26%|██▌       | 3409/13317 [35:06<1:41:24,  1.63it/s]
                                                      
{'loss': 0.2516, 'grad_norm': 0.5486571788787842, 'learning_rate': 9.277092855073063e-06, 'epoch': 0.77}

 26%|██▌       | 3409/13317 [35:06<1:41:24,  1.63it/s]
 26%|██▌       | 3410/13317 [35:07<1:41:29,  1.63it/s]
                                                      
{'loss': 0.2734, 'grad_norm': 0.5286996364593506, 'learning_rate': 9.276413881238801e-06, 'epoch': 0.77}

 26%|██▌       | 3410/13317 [35:07<1:41:29,  1.63it/s]
 26%|██▌       | 3411/13317 [35:07<1:41:28,  1.63it/s]
                                                      
{'loss': 0.2358, 'grad_norm': 0.4824065566062927, 'learning_rate': 9.275734613569625e-06, 'epoch': 0.77}

 26%|██▌       | 3411/13317 [35:08<1:41:28,  1.63it/s]
 26%|██▌       | 3412/13317 [35:08<1:41:27,  1.63it/s]
                                                      
{'loss': 0.202, 'grad_norm': 0.4652090072631836, 'learning_rate': 9.275055052112209e-06, 'epoch': 0.77}

 26%|██▌       | 3412/13317 [35:08<1:41:27,  1.63it/s]
 26%|██▌       | 3413/13317 [35:09<1:41:24,  1.63it/s]
                                                      
{'loss': 0.2126, 'grad_norm': 0.4831688702106476, 'learning_rate': 9.274375196913246e-06, 'epoch': 0.77}

 26%|██▌       | 3413/13317 [35:09<1:41:24,  1.63it/s]
 26%|██▌       | 3414/13317 [35:09<1:41:26,  1.63it/s]
                                                      
{'loss': 0.3535, 'grad_norm': 0.5820297002792358, 'learning_rate': 9.273695048019448e-06, 'epoch': 0.77}

 26%|██▌       | 3414/13317 [35:09<1:41:26,  1.63it/s]
 26%|██▌       | 3415/13317 [35:10<1:41:52,  1.62it/s]
                                                      
{'loss': 0.2058, 'grad_norm': 0.5187715888023376, 'learning_rate': 9.27301460547755e-06, 'epoch': 0.77}

 26%|██▌       | 3415/13317 [35:10<1:41:52,  1.62it/s]
 26%|██▌       | 3416/13317 [35:10<1:41:39,  1.62it/s]
                                                      
{'loss': 0.3217, 'grad_norm': 0.6692126989364624, 'learning_rate': 9.272333869334306e-06, 'epoch': 0.77}

 26%|██▌       | 3416/13317 [35:11<1:41:39,  1.62it/s]
 26%|██▌       | 3417/13317 [35:11<1:41:44,  1.62it/s]
                                                      
{'loss': 0.2941, 'grad_norm': 0.6633061170578003, 'learning_rate': 9.27165283963649e-06, 'epoch': 0.77}

 26%|██▌       | 3417/13317 [35:11<1:41:44,  1.62it/s]
 26%|██▌       | 3418/13317 [35:12<1:41:33,  1.62it/s]
                                                      
{'loss': 0.3297, 'grad_norm': 0.6172178983688354, 'learning_rate': 9.270971516430893e-06, 'epoch': 0.77}

 26%|██▌       | 3418/13317 [35:12<1:41:33,  1.62it/s]
 26%|██▌       | 3419/13317 [35:12<1:41:24,  1.63it/s]
                                                      
{'loss': 0.2717, 'grad_norm': 0.5158414840698242, 'learning_rate': 9.270289899764331e-06, 'epoch': 0.77}

 26%|██▌       | 3419/13317 [35:12<1:41:24,  1.63it/s]
 26%|██▌       | 3420/13317 [35:13<1:41:26,  1.63it/s]
                                                      
{'loss': 0.2853, 'grad_norm': 0.5437746047973633, 'learning_rate': 9.269607989683639e-06, 'epoch': 0.77}

 26%|██▌       | 3420/13317 [35:13<1:41:26,  1.63it/s]
 26%|██▌       | 3421/13317 [35:14<1:41:19,  1.63it/s]
                                                      
{'loss': 0.2649, 'grad_norm': 0.5204260349273682, 'learning_rate': 9.26892578623567e-06, 'epoch': 0.77}

 26%|██▌       | 3421/13317 [35:14<1:41:19,  1.63it/s]
 26%|██▌       | 3422/13317 [35:14<1:41:26,  1.63it/s]
                                                      
{'loss': 0.2192, 'grad_norm': 0.5301178693771362, 'learning_rate': 9.2682432894673e-06, 'epoch': 0.77}

 26%|██▌       | 3422/13317 [35:14<1:41:26,  1.63it/s]
 26%|██▌       | 3423/13317 [35:15<1:41:20,  1.63it/s]
                                                      
{'loss': 0.3943, 'grad_norm': 0.5918915271759033, 'learning_rate': 9.267560499425425e-06, 'epoch': 0.77}

 26%|██▌       | 3423/13317 [35:15<1:41:20,  1.63it/s]
 26%|██▌       | 3424/13317 [35:15<1:41:23,  1.63it/s]
                                                      
{'loss': 0.1858, 'grad_norm': 0.5074665546417236, 'learning_rate': 9.266877416156955e-06, 'epoch': 0.77}

 26%|██▌       | 3424/13317 [35:16<1:41:23,  1.63it/s]
 26%|██▌       | 3425/13317 [35:16<1:41:22,  1.63it/s]
                                                      
{'loss': 0.1961, 'grad_norm': 0.45684072375297546, 'learning_rate': 9.26619403970883e-06, 'epoch': 0.77}

 26%|██▌       | 3425/13317 [35:16<1:41:22,  1.63it/s]
 26%|██▌       | 3426/13317 [35:17<1:41:16,  1.63it/s]
                                                      
{'loss': 0.2581, 'grad_norm': 0.5114890336990356, 'learning_rate': 9.265510370128003e-06, 'epoch': 0.77}

 26%|██▌       | 3426/13317 [35:17<1:41:16,  1.63it/s]
 26%|██▌       | 3427/13317 [35:17<1:41:46,  1.62it/s]
                                                      
{'loss': 0.2684, 'grad_norm': 0.5782044529914856, 'learning_rate': 9.26482640746145e-06, 'epoch': 0.77}

 26%|██▌       | 3427/13317 [35:17<1:41:46,  1.62it/s]
 26%|██▌       | 3428/13317 [35:18<1:41:33,  1.62it/s]
                                                      
{'loss': 0.2939, 'grad_norm': 0.6110339164733887, 'learning_rate': 9.264142151756164e-06, 'epoch': 0.77}

 26%|██▌       | 3428/13317 [35:18<1:41:33,  1.62it/s]
 26%|██▌       | 3429/13317 [35:18<1:41:23,  1.63it/s]
                                                      
{'loss': 0.2498, 'grad_norm': 0.554132878780365, 'learning_rate': 9.263457603059165e-06, 'epoch': 0.77}

 26%|██▌       | 3429/13317 [35:19<1:41:23,  1.63it/s]
 26%|██▌       | 3430/13317 [35:19<1:41:25,  1.62it/s]
                                                      
{'loss': 0.148, 'grad_norm': 0.4164541959762573, 'learning_rate': 9.262772761417485e-06, 'epoch': 0.77}

 26%|██▌       | 3430/13317 [35:19<1:41:25,  1.62it/s]
 26%|██▌       | 3431/13317 [35:20<1:41:20,  1.63it/s]
                                                      
{'loss': 0.1758, 'grad_norm': 0.40084198117256165, 'learning_rate': 9.262087626878183e-06, 'epoch': 0.77}

 26%|██▌       | 3431/13317 [35:20<1:41:20,  1.63it/s]
 26%|██▌       | 3432/13317 [35:20<1:41:22,  1.63it/s]
                                                      
{'loss': 0.2502, 'grad_norm': 0.5586901307106018, 'learning_rate': 9.26140219948833e-06, 'epoch': 0.77}

 26%|██▌       | 3432/13317 [35:20<1:41:22,  1.63it/s]
 26%|██▌       | 3433/13317 [35:21<1:41:19,  1.63it/s]
                                                      
{'loss': 0.24, 'grad_norm': 0.4366651773452759, 'learning_rate': 9.26071647929503e-06, 'epoch': 0.77}

 26%|██▌       | 3433/13317 [35:21<1:41:19,  1.63it/s]
 26%|██▌       | 3434/13317 [35:22<1:41:24,  1.62it/s]
                                                      
{'loss': 0.2472, 'grad_norm': 0.5550927519798279, 'learning_rate': 9.260030466345392e-06, 'epoch': 0.77}

 26%|██▌       | 3434/13317 [35:22<1:41:24,  1.62it/s]
 26%|██▌       | 3435/13317 [35:22<1:41:20,  1.63it/s]
                                                      
{'loss': 0.223, 'grad_norm': 0.45456239581108093, 'learning_rate': 9.259344160686555e-06, 'epoch': 0.77}

 26%|██▌       | 3435/13317 [35:22<1:41:20,  1.63it/s]
 26%|██▌       | 3436/13317 [35:23<1:41:16,  1.63it/s]
                                                      
{'loss': 0.2481, 'grad_norm': 0.5585009455680847, 'learning_rate': 9.258657562365674e-06, 'epoch': 0.77}

 26%|██▌       | 3436/13317 [35:23<1:41:16,  1.63it/s]
 26%|██▌       | 3437/13317 [35:23<1:41:17,  1.63it/s]
                                                      
{'loss': 0.254, 'grad_norm': 0.4967055916786194, 'learning_rate': 9.25797067142993e-06, 'epoch': 0.77}

 26%|██▌       | 3437/13317 [35:24<1:41:17,  1.63it/s]
 26%|██▌       | 3438/13317 [35:24<1:41:11,  1.63it/s]
                                                      
{'loss': 0.1918, 'grad_norm': 0.4766235649585724, 'learning_rate': 9.257283487926516e-06, 'epoch': 0.77}

 26%|██▌       | 3438/13317 [35:24<1:41:11,  1.63it/s]
 26%|██▌       | 3439/13317 [35:25<1:41:16,  1.63it/s]
                                                      
{'loss': 0.2962, 'grad_norm': 0.5733106136322021, 'learning_rate': 9.256596011902649e-06, 'epoch': 0.77}

 26%|██▌       | 3439/13317 [35:25<1:41:16,  1.63it/s]
 26%|██▌       | 3440/13317 [35:25<1:41:24,  1.62it/s]
                                                      
{'loss': 0.2735, 'grad_norm': 0.5782309770584106, 'learning_rate': 9.255908243405567e-06, 'epoch': 0.77}

 26%|██▌       | 3440/13317 [35:25<1:41:24,  1.62it/s]
 26%|██▌       | 3441/13317 [35:26<1:41:12,  1.63it/s]
                                                      
{'loss': 0.2106, 'grad_norm': 0.5544835925102234, 'learning_rate': 9.255220182482526e-06, 'epoch': 0.78}

 26%|██▌       | 3441/13317 [35:26<1:41:12,  1.63it/s]
 26%|██▌       | 3442/13317 [35:26<1:41:13,  1.63it/s]
                                                      
{'loss': 0.2208, 'grad_norm': 0.5546177625656128, 'learning_rate': 9.254531829180804e-06, 'epoch': 0.78}

 26%|██▌       | 3442/13317 [35:27<1:41:13,  1.63it/s]
 26%|██▌       | 3443/13317 [35:27<1:41:07,  1.63it/s]
                                                      
{'loss': 0.2114, 'grad_norm': 0.5070033669471741, 'learning_rate': 9.253843183547696e-06, 'epoch': 0.78}

 26%|██▌       | 3443/13317 [35:27<1:41:07,  1.63it/s]
 26%|██▌       | 3444/13317 [35:28<1:41:03,  1.63it/s]
                                                      
{'loss': 0.2374, 'grad_norm': 0.5393062829971313, 'learning_rate': 9.253154245630523e-06, 'epoch': 0.78}

 26%|██▌       | 3444/13317 [35:28<1:41:03,  1.63it/s]
 26%|██▌       | 3445/13317 [35:28<1:41:07,  1.63it/s]
                                                      
{'loss': 0.2493, 'grad_norm': 0.5523160696029663, 'learning_rate': 9.25246501547662e-06, 'epoch': 0.78}

 26%|██▌       | 3445/13317 [35:28<1:41:07,  1.63it/s]
 26%|██▌       | 3446/13317 [35:29<1:41:01,  1.63it/s]
                                                      
{'loss': 0.2467, 'grad_norm': 0.5815749168395996, 'learning_rate': 9.251775493133342e-06, 'epoch': 0.78}

 26%|██▌       | 3446/13317 [35:29<1:41:01,  1.63it/s]
 26%|██▌       | 3447/13317 [35:30<1:40:58,  1.63it/s]
                                                      
{'loss': 0.2294, 'grad_norm': 0.49460652470588684, 'learning_rate': 9.251085678648072e-06, 'epoch': 0.78}

 26%|██▌       | 3447/13317 [35:30<1:40:58,  1.63it/s]
 26%|██▌       | 3448/13317 [35:30<1:41:00,  1.63it/s]
                                                      
{'loss': 0.2392, 'grad_norm': 0.5501682162284851, 'learning_rate': 9.250395572068202e-06, 'epoch': 0.78}

 26%|██▌       | 3448/13317 [35:30<1:41:00,  1.63it/s]
 26%|██▌       | 3449/13317 [35:31<1:40:54,  1.63it/s]
                                                      
{'loss': 0.2255, 'grad_norm': 0.5403647422790527, 'learning_rate': 9.249705173441154e-06, 'epoch': 0.78}

 26%|██▌       | 3449/13317 [35:31<1:40:54,  1.63it/s]
 26%|██▌       | 3450/13317 [35:31<1:41:23,  1.62it/s]
                                                      
{'loss': 0.2692, 'grad_norm': 0.6424387693405151, 'learning_rate': 9.249014482814364e-06, 'epoch': 0.78}

 26%|██▌       | 3450/13317 [35:32<1:41:23,  1.62it/s]
 26%|██▌       | 3451/13317 [35:32<1:41:14,  1.62it/s]
                                                      
{'loss': 0.1918, 'grad_norm': 0.44241082668304443, 'learning_rate': 9.24832350023529e-06, 'epoch': 0.78}

 26%|██▌       | 3451/13317 [35:32<1:41:14,  1.62it/s]
 26%|██▌       | 3452/13317 [35:33<1:41:11,  1.62it/s]
                                                      
{'loss': 0.2898, 'grad_norm': 0.5901171565055847, 'learning_rate': 9.247632225751411e-06, 'epoch': 0.78}

 26%|██▌       | 3452/13317 [35:33<1:41:11,  1.62it/s]
 26%|██▌       | 3453/13317 [35:33<1:41:07,  1.63it/s]
                                                      
{'loss': 0.3308, 'grad_norm': 0.6344764828681946, 'learning_rate': 9.24694065941022e-06, 'epoch': 0.78}

 26%|██▌       | 3453/13317 [35:33<1:41:07,  1.63it/s]
 26%|██▌       | 3454/13317 [35:34<1:41:05,  1.63it/s]
                                                      
{'loss': 0.3018, 'grad_norm': 0.5884047746658325, 'learning_rate': 9.24624880125924e-06, 'epoch': 0.78}

 26%|██▌       | 3454/13317 [35:34<1:41:05,  1.63it/s]
 26%|██▌       | 3455/13317 [35:34<1:41:05,  1.63it/s]
                                                      
{'loss': 0.2027, 'grad_norm': 0.45342329144477844, 'learning_rate': 9.245556651346008e-06, 'epoch': 0.78}

 26%|██▌       | 3455/13317 [35:35<1:41:05,  1.63it/s]
 26%|██▌       | 3456/13317 [35:35<1:41:00,  1.63it/s]
                                                      
{'loss': 0.2486, 'grad_norm': 0.5458234548568726, 'learning_rate': 9.24486420971808e-06, 'epoch': 0.78}

 26%|██▌       | 3456/13317 [35:35<1:41:00,  1.63it/s]
 26%|██▌       | 3457/13317 [35:36<1:41:00,  1.63it/s]
                                                      
{'loss': 0.2524, 'grad_norm': 0.532820999622345, 'learning_rate': 9.244171476423037e-06, 'epoch': 0.78}

 26%|██▌       | 3457/13317 [35:36<1:41:00,  1.63it/s]
 26%|██▌       | 3458/13317 [35:36<1:40:55,  1.63it/s]
                                                      
{'loss': 0.2862, 'grad_norm': 0.5741249322891235, 'learning_rate': 9.243478451508473e-06, 'epoch': 0.78}

 26%|██▌       | 3458/13317 [35:36<1:40:55,  1.63it/s]
 26%|██▌       | 3459/13317 [35:37<1:41:00,  1.63it/s]
                                                      
{'loss': 0.2703, 'grad_norm': 0.6631406545639038, 'learning_rate': 9.242785135022012e-06, 'epoch': 0.78}

 26%|██▌       | 3459/13317 [35:37<1:41:00,  1.63it/s]
 26%|██▌       | 3460/13317 [35:38<1:41:00,  1.63it/s]
                                                      
{'loss': 0.2088, 'grad_norm': 0.44081437587738037, 'learning_rate': 9.242091527011287e-06, 'epoch': 0.78}

 26%|██▌       | 3460/13317 [35:38<1:41:00,  1.63it/s]
 26%|██▌       | 3461/13317 [35:38<1:40:54,  1.63it/s]
                                                      
{'loss': 0.2869, 'grad_norm': 0.5899417996406555, 'learning_rate': 9.241397627523959e-06, 'epoch': 0.78}

 26%|██▌       | 3461/13317 [35:38<1:40:54,  1.63it/s]
 26%|██▌       | 3462/13317 [35:39<1:40:51,  1.63it/s]
                                                      
{'loss': 0.2313, 'grad_norm': 0.534121572971344, 'learning_rate': 9.240703436607706e-06, 'epoch': 0.78}

 26%|██▌       | 3462/13317 [35:39<1:40:51,  1.63it/s]
 26%|██▌       | 3463/13317 [35:39<1:40:47,  1.63it/s]
                                                      
{'loss': 0.2653, 'grad_norm': 0.5827072262763977, 'learning_rate': 9.240008954310224e-06, 'epoch': 0.78}

 26%|██▌       | 3463/13317 [35:40<1:40:47,  1.63it/s]
 26%|██▌       | 3464/13317 [35:40<1:40:52,  1.63it/s]
                                                      
{'loss': 0.2385, 'grad_norm': 0.5842999219894409, 'learning_rate': 9.239314180679232e-06, 'epoch': 0.78}

 26%|██▌       | 3464/13317 [35:40<1:40:52,  1.63it/s]
 26%|██▌       | 3465/13317 [35:41<1:40:51,  1.63it/s]
                                                      
{'loss': 0.2807, 'grad_norm': 0.5974473357200623, 'learning_rate': 9.23861911576247e-06, 'epoch': 0.78}

 26%|██▌       | 3465/13317 [35:41<1:40:51,  1.63it/s]
 26%|██▌       | 3466/13317 [35:41<1:40:50,  1.63it/s]
                                                      
{'loss': 0.2877, 'grad_norm': 0.6243665814399719, 'learning_rate': 9.237923759607698e-06, 'epoch': 0.78}

 26%|██▌       | 3466/13317 [35:41<1:40:50,  1.63it/s]
 26%|██▌       | 3467/13317 [35:42<1:40:49,  1.63it/s]
                                                      
{'loss': 0.257, 'grad_norm': 0.491012841463089, 'learning_rate': 9.23722811226269e-06, 'epoch': 0.78}

 26%|██▌       | 3467/13317 [35:42<1:40:49,  1.63it/s]
 26%|██▌       | 3468/13317 [35:42<1:40:52,  1.63it/s]
                                                      
{'loss': 0.1837, 'grad_norm': 0.5736685395240784, 'learning_rate': 9.236532173775246e-06, 'epoch': 0.78}

 26%|██▌       | 3468/13317 [35:43<1:40:52,  1.63it/s]
 26%|██▌       | 3469/13317 [35:43<1:40:50,  1.63it/s]
                                                      
{'loss': 0.3323, 'grad_norm': 0.5557296872138977, 'learning_rate': 9.235835944193182e-06, 'epoch': 0.78}

 26%|██▌       | 3469/13317 [35:43<1:40:50,  1.63it/s]
 26%|██▌       | 3470/13317 [35:44<1:40:50,  1.63it/s]
                                                      
{'loss': 0.227, 'grad_norm': 0.5157596468925476, 'learning_rate': 9.235139423564344e-06, 'epoch': 0.78}

 26%|██▌       | 3470/13317 [35:44<1:40:50,  1.63it/s]
 26%|██▌       | 3471/13317 [35:44<1:40:47,  1.63it/s]
                                                      
{'loss': 0.2203, 'grad_norm': 0.47764265537261963, 'learning_rate': 9.234442611936582e-06, 'epoch': 0.78}

 26%|██▌       | 3471/13317 [35:44<1:40:47,  1.63it/s]
 26%|██▌       | 3472/13317 [35:45<1:40:54,  1.63it/s]
                                                      
{'loss': 0.2393, 'grad_norm': 0.5076362490653992, 'learning_rate': 9.233745509357778e-06, 'epoch': 0.78}

 26%|██▌       | 3472/13317 [35:45<1:40:54,  1.63it/s]
 26%|██▌       | 3473/13317 [35:46<1:40:48,  1.63it/s]
                                                      
{'loss': 0.2613, 'grad_norm': 0.5420170426368713, 'learning_rate': 9.233048115875831e-06, 'epoch': 0.78}

 26%|██▌       | 3473/13317 [35:46<1:40:48,  1.63it/s]
 26%|██▌       | 3474/13317 [35:46<1:40:49,  1.63it/s]
                                                      
{'loss': 0.3002, 'grad_norm': 0.5452030301094055, 'learning_rate': 9.232350431538656e-06, 'epoch': 0.78}

 26%|██▌       | 3474/13317 [35:46<1:40:49,  1.63it/s]
 26%|██▌       | 3475/13317 [35:47<1:40:56,  1.63it/s]
                                                      
{'loss': 0.2635, 'grad_norm': 0.5598823428153992, 'learning_rate': 9.231652456394197e-06, 'epoch': 0.78}

 26%|██▌       | 3475/13317 [35:47<1:40:56,  1.63it/s]
 26%|██▌       | 3476/13317 [35:47<1:40:48,  1.63it/s]
                                                      
{'loss': 0.2754, 'grad_norm': 0.5745025277137756, 'learning_rate': 9.230954190490406e-06, 'epoch': 0.78}

 26%|██▌       | 3476/13317 [35:48<1:40:48,  1.63it/s]
 26%|██▌       | 3477/13317 [35:48<1:40:46,  1.63it/s]
                                                      
{'loss': 0.3103, 'grad_norm': 0.5588862299919128, 'learning_rate': 9.230255633875265e-06, 'epoch': 0.78}

 26%|██▌       | 3477/13317 [35:48<1:40:46,  1.63it/s]
 26%|██▌       | 3478/13317 [35:49<1:40:41,  1.63it/s]
                                                      
{'loss': 0.2877, 'grad_norm': 0.5923765301704407, 'learning_rate': 9.229556786596772e-06, 'epoch': 0.78}

 26%|██▌       | 3478/13317 [35:49<1:40:41,  1.63it/s]
 26%|██▌       | 3479/13317 [35:49<1:40:44,  1.63it/s]
                                                      
{'loss': 0.2478, 'grad_norm': 0.4977611303329468, 'learning_rate': 9.228857648702942e-06, 'epoch': 0.78}

 26%|██▌       | 3479/13317 [35:49<1:40:44,  1.63it/s]
 26%|██▌       | 3480/13317 [35:50<1:40:50,  1.63it/s]
                                                      
{'loss': 0.2146, 'grad_norm': 0.46298375725746155, 'learning_rate': 9.228158220241818e-06, 'epoch': 0.78}

 26%|██▌       | 3480/13317 [35:50<1:40:50,  1.63it/s]
 26%|██▌       | 3481/13317 [35:50<1:40:48,  1.63it/s]
                                                      
{'loss': 0.2574, 'grad_norm': 0.5008623003959656, 'learning_rate': 9.227458501261455e-06, 'epoch': 0.78}

 26%|██▌       | 3481/13317 [35:51<1:40:48,  1.63it/s]
 26%|██▌       | 3482/13317 [35:51<1:40:42,  1.63it/s]
                                                      
{'loss': 0.2494, 'grad_norm': 0.5870835781097412, 'learning_rate': 9.226758491809933e-06, 'epoch': 0.78}

 26%|██▌       | 3482/13317 [35:51<1:40:42,  1.63it/s]
 26%|██▌       | 3483/13317 [35:52<1:40:34,  1.63it/s]
                                                      
{'loss': 0.3042, 'grad_norm': 0.636352002620697, 'learning_rate': 9.226058191935348e-06, 'epoch': 0.78}

 26%|██▌       | 3483/13317 [35:52<1:40:34,  1.63it/s]
 26%|██▌       | 3484/13317 [35:52<1:40:43,  1.63it/s]
                                                      
{'loss': 0.2423, 'grad_norm': 0.6191544532775879, 'learning_rate': 9.225357601685818e-06, 'epoch': 0.78}

 26%|██▌       | 3484/13317 [35:52<1:40:43,  1.63it/s]
 26%|██▌       | 3485/13317 [35:53<1:40:43,  1.63it/s]
                                                      
{'loss': 0.2267, 'grad_norm': 0.5904302000999451, 'learning_rate': 9.224656721109483e-06, 'epoch': 0.79}

 26%|██▌       | 3485/13317 [35:53<1:40:43,  1.63it/s]
 26%|██▌       | 3486/13317 [35:54<1:40:37,  1.63it/s]
                                                      
{'loss': 0.289, 'grad_norm': 0.5473377704620361, 'learning_rate': 9.2239555502545e-06, 'epoch': 0.79}

 26%|██▌       | 3486/13317 [35:54<1:40:37,  1.63it/s]
 26%|██▌       | 3487/13317 [35:54<1:40:37,  1.63it/s]
                                                      
{'loss': 0.2294, 'grad_norm': 0.47058892250061035, 'learning_rate': 9.223254089169047e-06, 'epoch': 0.79}

 26%|██▌       | 3487/13317 [35:54<1:40:37,  1.63it/s]
 26%|██▌       | 3488/13317 [35:55<1:40:31,  1.63it/s]
                                                      
{'loss': 0.2791, 'grad_norm': 0.509850025177002, 'learning_rate': 9.22255233790132e-06, 'epoch': 0.79}

 26%|██▌       | 3488/13317 [35:55<1:40:31,  1.63it/s]
 26%|██▌       | 3489/13317 [35:55<1:40:32,  1.63it/s]
                                                      
{'loss': 0.2424, 'grad_norm': 0.5557462573051453, 'learning_rate': 9.22185029649954e-06, 'epoch': 0.79}

 26%|██▌       | 3489/13317 [35:55<1:40:32,  1.63it/s]
 26%|██▌       | 3490/13317 [35:56<1:40:35,  1.63it/s]
                                                      
{'loss': 0.338, 'grad_norm': 0.6271080374717712, 'learning_rate': 9.221147965011942e-06, 'epoch': 0.79}

 26%|██▌       | 3490/13317 [35:56<1:40:35,  1.63it/s]
 26%|██▌       | 3491/13317 [35:57<1:40:33,  1.63it/s]
                                                      
{'loss': 0.2394, 'grad_norm': 0.486126571893692, 'learning_rate': 9.220445343486785e-06, 'epoch': 0.79}

 26%|██▌       | 3491/13317 [35:57<1:40:33,  1.63it/s]
 26%|██▌       | 3492/13317 [35:57<1:40:32,  1.63it/s]
                                                      
{'loss': 0.2445, 'grad_norm': 0.5284334421157837, 'learning_rate': 9.219742431972347e-06, 'epoch': 0.79}

 26%|██▌       | 3492/13317 [35:57<1:40:32,  1.63it/s]
 26%|██▌       | 3493/13317 [35:58<1:40:30,  1.63it/s]
                                                      
{'loss': 0.2294, 'grad_norm': 0.6081074476242065, 'learning_rate': 9.219039230516923e-06, 'epoch': 0.79}

 26%|██▌       | 3493/13317 [35:58<1:40:30,  1.63it/s]
 26%|██▌       | 3494/13317 [35:58<1:40:30,  1.63it/s]
                                                      
{'loss': 0.2039, 'grad_norm': 0.5241208076477051, 'learning_rate': 9.218335739168833e-06, 'epoch': 0.79}

 26%|██▌       | 3494/13317 [35:59<1:40:30,  1.63it/s]
 26%|██▌       | 3495/13317 [35:59<1:40:30,  1.63it/s]
                                                      
{'loss': 0.247, 'grad_norm': 0.5511093139648438, 'learning_rate': 9.217631957976412e-06, 'epoch': 0.79}

 26%|██▌       | 3495/13317 [35:59<1:40:30,  1.63it/s]
 26%|██▋       | 3496/13317 [36:00<1:40:29,  1.63it/s]
                                                      
{'loss': 0.2542, 'grad_norm': 0.5330905318260193, 'learning_rate': 9.21692788698802e-06, 'epoch': 0.79}

 26%|██▋       | 3496/13317 [36:00<1:40:29,  1.63it/s]
 26%|██▋       | 3497/13317 [36:00<1:40:35,  1.63it/s]
                                                      
{'loss': 0.19, 'grad_norm': 0.4763564169406891, 'learning_rate': 9.216223526252032e-06, 'epoch': 0.79}

 26%|██▋       | 3497/13317 [36:00<1:40:35,  1.63it/s]
 26%|██▋       | 3498/13317 [36:01<1:40:28,  1.63it/s]
                                                      
{'loss': 0.2892, 'grad_norm': 0.6088959574699402, 'learning_rate': 9.215518875816844e-06, 'epoch': 0.79}

 26%|██▋       | 3498/13317 [36:01<1:40:28,  1.63it/s]
 26%|██▋       | 3499/13317 [36:01<1:40:24,  1.63it/s]
                                                      
{'loss': 0.2713, 'grad_norm': 0.6143534779548645, 'learning_rate': 9.214813935730877e-06, 'epoch': 0.79}

 26%|██▋       | 3499/13317 [36:02<1:40:24,  1.63it/s]
 26%|██▋       | 3500/13317 [36:02<1:40:33,  1.63it/s]
                                                      
{'loss': 0.2193, 'grad_norm': 0.5480522513389587, 'learning_rate': 9.214108706042566e-06, 'epoch': 0.79}

 26%|██▋       | 3500/13317 [36:02<1:40:33,  1.63it/s]

  0%|          | 0/13 [00:00<?, ?it/s]

 15%|█▌        | 2/13 [00:00<00:00, 13.37it/s]

 31%|███       | 4/13 [00:00<00:01,  8.40it/s]

 38%|███▊      | 5/13 [00:00<00:01,  7.80it/s]

 46%|████▌     | 6/13 [00:00<00:00,  7.43it/s]

 54%|█████▍    | 7/13 [00:00<00:00,  7.19it/s]

 62%|██████▏   | 8/13 [00:01<00:00,  7.03it/s]

 69%|██████▉   | 9/13 [00:01<00:00,  6.91it/s]

 77%|███████▋  | 10/13 [00:01<00:00,  6.84it/s]

 85%|████████▍ | 11/13 [00:01<00:00,  6.79it/s]

 92%|█████████▏| 12/13 [00:01<00:00,  6.75it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]
                                                      

                                               
{'eval_loss': 0.6658998131752014, 'eval_runtime': 1.9515, 'eval_samples_per_second': 102.487, 'eval_steps_per_second': 6.662, 'epoch': 0.79}

 26%|██▋       | 3500/13317 [36:04<1:40:33,  1.63it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]

                                               
 26%|██▋       | 3501/13317 [36:05<3:16:32,  1.20s/it]
                                                      
{'loss': 0.2137, 'grad_norm': 0.49138590693473816, 'learning_rate': 9.213403186800364e-06, 'epoch': 0.79}

 26%|██▋       | 3501/13317 [36:05<3:16:32,  1.20s/it]
 26%|██▋       | 3502/13317 [36:05<2:47:40,  1.02s/it]
                                                      
{'loss': 0.291, 'grad_norm': 0.5515320897102356, 'learning_rate': 9.212697378052754e-06, 'epoch': 0.79}

 26%|██▋       | 3502/13317 [36:05<2:47:40,  1.02s/it]
 26%|██▋       | 3503/13317 [36:06<2:27:26,  1.11it/s]
                                                      
{'loss': 0.2885, 'grad_norm': 0.7303850054740906, 'learning_rate': 9.211991279848227e-06, 'epoch': 0.79}

 26%|██▋       | 3503/13317 [36:06<2:27:26,  1.11it/s]
 26%|██▋       | 3504/13317 [36:07<2:13:17,  1.23it/s]
                                                      
{'loss': 0.2575, 'grad_norm': 0.5417729020118713, 'learning_rate': 9.211284892235303e-06, 'epoch': 0.79}

 26%|██▋       | 3504/13317 [36:07<2:13:17,  1.23it/s]
 26%|██▋       | 3505/13317 [36:07<2:03:23,  1.33it/s]
                                                      
{'loss': 0.3029, 'grad_norm': 0.5802406072616577, 'learning_rate': 9.210578215262518e-06, 'epoch': 0.79}

 26%|██▋       | 3505/13317 [36:07<2:03:23,  1.33it/s]
 26%|██▋       | 3506/13317 [36:08<1:56:26,  1.40it/s]
                                                      
{'loss': 0.2483, 'grad_norm': 0.5794926285743713, 'learning_rate': 9.209871248978425e-06, 'epoch': 0.79}

 26%|██▋       | 3506/13317 [36:08<1:56:26,  1.40it/s]
 26%|██▋       | 3507/13317 [36:08<1:51:41,  1.46it/s]
                                                      
{'loss': 0.2355, 'grad_norm': 0.45871198177337646, 'learning_rate': 9.209163993431605e-06, 'epoch': 0.79}

 26%|██▋       | 3507/13317 [36:08<1:51:41,  1.46it/s]
 26%|██▋       | 3508/13317 [36:09<1:48:15,  1.51it/s]
                                                      
{'loss': 0.2557, 'grad_norm': 0.6210447549819946, 'learning_rate': 9.208456448670648e-06, 'epoch': 0.79}

 26%|██▋       | 3508/13317 [36:09<1:48:15,  1.51it/s]
 26%|██▋       | 3509/13317 [36:10<1:45:49,  1.54it/s]
                                                      
{'loss': 0.2907, 'grad_norm': 0.6146055459976196, 'learning_rate': 9.207748614744176e-06, 'epoch': 0.79}

 26%|██▋       | 3509/13317 [36:10<1:45:49,  1.54it/s]
 26%|██▋       | 3510/13317 [36:10<1:44:08,  1.57it/s]
                                                      
{'loss': 0.2882, 'grad_norm': 0.5188525915145874, 'learning_rate': 9.207040491700821e-06, 'epoch': 0.79}

 26%|██▋       | 3510/13317 [36:10<1:44:08,  1.57it/s]
 26%|██▋       | 3511/13317 [36:11<1:42:58,  1.59it/s]
                                                      
{'loss': 0.2935, 'grad_norm': 0.6695197224617004, 'learning_rate': 9.20633207958924e-06, 'epoch': 0.79}

 26%|██▋       | 3511/13317 [36:11<1:42:58,  1.59it/s]
 26%|██▋       | 3512/13317 [36:11<1:42:13,  1.60it/s]
                                                      
{'loss': 0.348, 'grad_norm': 0.6801777482032776, 'learning_rate': 9.205623378458105e-06, 'epoch': 0.79}

 26%|██▋       | 3512/13317 [36:12<1:42:13,  1.60it/s]
 26%|██▋       | 3513/13317 [36:12<1:41:37,  1.61it/s]
                                                      
{'loss': 0.276, 'grad_norm': 0.5969319939613342, 'learning_rate': 9.204914388356116e-06, 'epoch': 0.79}

 26%|██▋       | 3513/13317 [36:12<1:41:37,  1.61it/s]
 26%|██▋       | 3514/13317 [36:13<1:41:13,  1.61it/s]
                                                      
{'loss': 0.3109, 'grad_norm': 0.5713356733322144, 'learning_rate': 9.204205109331987e-06, 'epoch': 0.79}

 26%|██▋       | 3514/13317 [36:13<1:41:13,  1.61it/s]
 26%|██▋       | 3515/13317 [36:13<1:40:57,  1.62it/s]
                                                      
{'loss': 0.2302, 'grad_norm': 0.48213014006614685, 'learning_rate': 9.203495541434452e-06, 'epoch': 0.79}

 26%|██▋       | 3515/13317 [36:13<1:40:57,  1.62it/s]
 26%|██▋       | 3516/13317 [36:14<1:40:46,  1.62it/s]
                                                      
{'loss': 0.2369, 'grad_norm': 0.5117536187171936, 'learning_rate': 9.202785684712266e-06, 'epoch': 0.79}

 26%|██▋       | 3516/13317 [36:14<1:40:46,  1.62it/s]
 26%|██▋       | 3517/13317 [36:15<1:40:42,  1.62it/s]
                                                      
{'loss': 0.2217, 'grad_norm': 0.41476139426231384, 'learning_rate': 9.202075539214201e-06, 'epoch': 0.79}

 26%|██▋       | 3517/13317 [36:15<1:40:42,  1.62it/s]
 26%|██▋       | 3518/13317 [36:15<1:40:31,  1.62it/s]
                                                      
{'loss': 0.2428, 'grad_norm': 0.4983362555503845, 'learning_rate': 9.201365104989057e-06, 'epoch': 0.79}

 26%|██▋       | 3518/13317 [36:15<1:40:31,  1.62it/s]
 26%|██▋       | 3519/13317 [36:16<1:40:24,  1.63it/s]
                                                      
{'loss': 0.204, 'grad_norm': 0.5022507905960083, 'learning_rate': 9.200654382085645e-06, 'epoch': 0.79}

 26%|██▋       | 3519/13317 [36:16<1:40:24,  1.63it/s]
 26%|██▋       | 3520/13317 [36:16<1:40:18,  1.63it/s]
                                                      
{'loss': 0.29, 'grad_norm': 0.5973839163780212, 'learning_rate': 9.1999433705528e-06, 'epoch': 0.79}

 26%|██▋       | 3520/13317 [36:16<1:40:18,  1.63it/s]
 26%|██▋       | 3521/13317 [36:17<1:40:15,  1.63it/s]
                                                      
{'loss': 0.2576, 'grad_norm': 0.5080167651176453, 'learning_rate': 9.199232070439374e-06, 'epoch': 0.79}

 26%|██▋       | 3521/13317 [36:17<1:40:15,  1.63it/s]
 26%|██▋       | 3522/13317 [36:18<1:40:19,  1.63it/s]
                                                      
{'loss': 0.2408, 'grad_norm': 0.455711305141449, 'learning_rate': 9.198520481794244e-06, 'epoch': 0.79}

 26%|██▋       | 3522/13317 [36:18<1:40:19,  1.63it/s]
 26%|██▋       | 3523/13317 [36:18<1:40:13,  1.63it/s]
                                                      
{'loss': 0.2449, 'grad_norm': 0.5181347131729126, 'learning_rate': 9.197808604666301e-06, 'epoch': 0.79}

 26%|██▋       | 3523/13317 [36:18<1:40:13,  1.63it/s]
 26%|██▋       | 3524/13317 [36:19<1:40:10,  1.63it/s]
                                                      
{'loss': 0.2716, 'grad_norm': 0.5418415665626526, 'learning_rate': 9.197096439104461e-06, 'epoch': 0.79}

 26%|██▋       | 3524/13317 [36:19<1:40:10,  1.63it/s]
 26%|██▋       | 3525/13317 [36:19<1:40:10,  1.63it/s]
                                                      
{'loss': 0.2627, 'grad_norm': 0.5582274794578552, 'learning_rate': 9.196383985157657e-06, 'epoch': 0.79}

 26%|██▋       | 3525/13317 [36:20<1:40:10,  1.63it/s]
 26%|██▋       | 3526/13317 [36:20<1:40:05,  1.63it/s]
                                                      
{'loss': 0.2733, 'grad_norm': 0.47838497161865234, 'learning_rate': 9.195671242874839e-06, 'epoch': 0.79}

 26%|██▋       | 3526/13317 [36:20<1:40:05,  1.63it/s]
 26%|██▋       | 3527/13317 [36:21<1:40:16,  1.63it/s]
                                                      
{'loss': 0.2278, 'grad_norm': 0.5190288424491882, 'learning_rate': 9.194958212304982e-06, 'epoch': 0.79}

 26%|██▋       | 3527/13317 [36:21<1:40:16,  1.63it/s]
 26%|██▋       | 3528/13317 [36:21<1:40:14,  1.63it/s]
                                                      
{'loss': 0.259, 'grad_norm': 0.5642428994178772, 'learning_rate': 9.194244893497081e-06, 'epoch': 0.79}

 26%|██▋       | 3528/13317 [36:21<1:40:14,  1.63it/s]
 26%|██▋       | 3529/13317 [36:22<1:40:08,  1.63it/s]
                                                      
{'loss': 0.3155, 'grad_norm': 0.6101175546646118, 'learning_rate': 9.193531286500145e-06, 'epoch': 0.79}

 26%|██▋       | 3529/13317 [36:22<1:40:08,  1.63it/s]
 27%|██▋       | 3530/13317 [36:22<1:40:08,  1.63it/s]
                                                      
{'loss': 0.2087, 'grad_norm': 0.46871957182884216, 'learning_rate': 9.192817391363209e-06, 'epoch': 0.8}

 27%|██▋       | 3530/13317 [36:23<1:40:08,  1.63it/s]
 27%|██▋       | 3531/13317 [36:23<1:40:04,  1.63it/s]
                                                      
{'loss': 0.287, 'grad_norm': 0.5224075317382812, 'learning_rate': 9.192103208135324e-06, 'epoch': 0.8}

 27%|██▋       | 3531/13317 [36:23<1:40:04,  1.63it/s]
 27%|██▋       | 3532/13317 [36:24<1:40:09,  1.63it/s]
                                                      
{'loss': 0.2338, 'grad_norm': 0.4730091691017151, 'learning_rate': 9.19138873686556e-06, 'epoch': 0.8}

 27%|██▋       | 3532/13317 [36:24<1:40:09,  1.63it/s]
 27%|██▋       | 3533/13317 [36:24<1:40:08,  1.63it/s]
                                                      
{'loss': 0.2139, 'grad_norm': 0.6035314798355103, 'learning_rate': 9.190673977603011e-06, 'epoch': 0.8}

 27%|██▋       | 3533/13317 [36:24<1:40:08,  1.63it/s]
 27%|██▋       | 3534/13317 [36:25<1:40:06,  1.63it/s]
                                                      
{'loss': 0.2391, 'grad_norm': 0.5000602602958679, 'learning_rate': 9.18995893039679e-06, 'epoch': 0.8}

 27%|██▋       | 3534/13317 [36:25<1:40:06,  1.63it/s]
 27%|██▋       | 3535/13317 [36:26<1:40:02,  1.63it/s]
                                                      
{'loss': 0.2432, 'grad_norm': 0.5861485004425049, 'learning_rate': 9.189243595296025e-06, 'epoch': 0.8}

 27%|██▋       | 3535/13317 [36:26<1:40:02,  1.63it/s]
 27%|██▋       | 3536/13317 [36:26<1:39:58,  1.63it/s]
                                                      
{'loss': 0.2035, 'grad_norm': 0.5545871257781982, 'learning_rate': 9.18852797234987e-06, 'epoch': 0.8}

 27%|██▋       | 3536/13317 [36:26<1:39:58,  1.63it/s]
 27%|██▋       | 3537/13317 [36:27<1:40:02,  1.63it/s]
                                                      
{'loss': 0.2143, 'grad_norm': 0.5195302963256836, 'learning_rate': 9.187812061607493e-06, 'epoch': 0.8}

 27%|██▋       | 3537/13317 [36:27<1:40:02,  1.63it/s]
 27%|██▋       | 3538/13317 [36:27<1:40:00,  1.63it/s]
                                                      
{'loss': 0.2272, 'grad_norm': 0.4677848219871521, 'learning_rate': 9.187095863118084e-06, 'epoch': 0.8}

 27%|██▋       | 3538/13317 [36:28<1:40:00,  1.63it/s]
 27%|██▋       | 3539/13317 [36:28<1:39:56,  1.63it/s]
                                                      
{'loss': 0.2555, 'grad_norm': 0.5721904635429382, 'learning_rate': 9.18637937693086e-06, 'epoch': 0.8}

 27%|██▋       | 3539/13317 [36:28<1:39:56,  1.63it/s]
 27%|██▋       | 3540/13317 [36:29<1:39:56,  1.63it/s]
                                                      
{'loss': 0.268, 'grad_norm': 0.6252016425132751, 'learning_rate': 9.185662603095045e-06, 'epoch': 0.8}

 27%|██▋       | 3540/13317 [36:29<1:39:56,  1.63it/s]
 27%|██▋       | 3541/13317 [36:29<1:39:54,  1.63it/s]
                                                      
{'loss': 0.2639, 'grad_norm': 0.48441997170448303, 'learning_rate': 9.18494554165989e-06, 'epoch': 0.8}

 27%|██▋       | 3541/13317 [36:29<1:39:54,  1.63it/s]
 27%|██▋       | 3542/13317 [36:30<1:39:52,  1.63it/s]
                                                      
{'loss': 0.2968, 'grad_norm': 0.5512303113937378, 'learning_rate': 9.184228192674667e-06, 'epoch': 0.8}

 27%|██▋       | 3542/13317 [36:30<1:39:52,  1.63it/s]
 27%|██▋       | 3543/13317 [36:30<1:39:54,  1.63it/s]
                                                      
{'loss': 0.1932, 'grad_norm': 0.541110634803772, 'learning_rate': 9.183510556188662e-06, 'epoch': 0.8}

 27%|██▋       | 3543/13317 [36:31<1:39:54,  1.63it/s]
 27%|██▋       | 3544/13317 [36:31<1:39:54,  1.63it/s]
                                                      
{'loss': 0.2783, 'grad_norm': 0.48559191823005676, 'learning_rate': 9.182792632251187e-06, 'epoch': 0.8}

 27%|██▋       | 3544/13317 [36:31<1:39:54,  1.63it/s]
 27%|██▋       | 3545/13317 [36:32<1:39:55,  1.63it/s]
                                                      
{'loss': 0.2508, 'grad_norm': 0.554400622844696, 'learning_rate': 9.182074420911568e-06, 'epoch': 0.8}

 27%|██▋       | 3545/13317 [36:32<1:39:55,  1.63it/s]
 27%|██▋       | 3546/13317 [36:32<1:39:56,  1.63it/s]
                                                      
{'loss': 0.2716, 'grad_norm': 0.5269930362701416, 'learning_rate': 9.181355922219158e-06, 'epoch': 0.8}

 27%|██▋       | 3546/13317 [36:32<1:39:56,  1.63it/s]
 27%|██▋       | 3547/13317 [36:33<1:39:52,  1.63it/s]
                                                      
{'loss': 0.1891, 'grad_norm': 0.5247080326080322, 'learning_rate': 9.180637136223323e-06, 'epoch': 0.8}

 27%|██▋       | 3547/13317 [36:33<1:39:52,  1.63it/s]
 27%|██▋       | 3548/13317 [36:34<1:39:54,  1.63it/s]
                                                      
{'loss': 0.2825, 'grad_norm': 0.5237602591514587, 'learning_rate': 9.179918062973451e-06, 'epoch': 0.8}

 27%|██▋       | 3548/13317 [36:34<1:39:54,  1.63it/s]
 27%|██▋       | 3549/13317 [36:34<1:39:56,  1.63it/s]
                                                      
{'loss': 0.2547, 'grad_norm': 0.5882977843284607, 'learning_rate': 9.179198702518952e-06, 'epoch': 0.8}

 27%|██▋       | 3549/13317 [36:34<1:39:56,  1.63it/s]
 27%|██▋       | 3550/13317 [36:35<1:39:52,  1.63it/s]
                                                      
{'loss': 0.2953, 'grad_norm': 0.6192547082901001, 'learning_rate': 9.178479054909249e-06, 'epoch': 0.8}

 27%|██▋       | 3550/13317 [36:35<1:39:52,  1.63it/s]
 27%|██▋       | 3551/13317 [36:35<1:39:52,  1.63it/s]
                                                      
{'loss': 0.3344, 'grad_norm': 0.6345459222793579, 'learning_rate': 9.177759120193794e-06, 'epoch': 0.8}

 27%|██▋       | 3551/13317 [36:36<1:39:52,  1.63it/s]
 27%|██▋       | 3552/13317 [36:36<1:39:51,  1.63it/s]
                                                      
{'loss': 0.1796, 'grad_norm': 0.47931358218193054, 'learning_rate': 9.177038898422053e-06, 'epoch': 0.8}

 27%|██▋       | 3552/13317 [36:36<1:39:51,  1.63it/s]
 27%|██▋       | 3553/13317 [36:37<1:39:50,  1.63it/s]
                                                      
{'loss': 0.2237, 'grad_norm': 0.5919163823127747, 'learning_rate': 9.176318389643512e-06, 'epoch': 0.8}

 27%|██▋       | 3553/13317 [36:37<1:39:50,  1.63it/s]
 27%|██▋       | 3554/13317 [36:37<1:39:45,  1.63it/s]
                                                      
{'loss': 0.2117, 'grad_norm': 0.4960648715496063, 'learning_rate': 9.175597593907679e-06, 'epoch': 0.8}

 27%|██▋       | 3554/13317 [36:37<1:39:45,  1.63it/s]
 27%|██▋       | 3555/13317 [36:38<1:39:43,  1.63it/s]
                                                      
{'loss': 0.2641, 'grad_norm': 0.5935308337211609, 'learning_rate': 9.17487651126408e-06, 'epoch': 0.8}

 27%|██▋       | 3555/13317 [36:38<1:39:43,  1.63it/s]
 27%|██▋       | 3556/13317 [36:38<1:39:42,  1.63it/s]
                                                      
{'loss': 0.3439, 'grad_norm': 0.5976787805557251, 'learning_rate': 9.174155141762258e-06, 'epoch': 0.8}

 27%|██▋       | 3556/13317 [36:39<1:39:42,  1.63it/s]
 27%|██▋       | 3557/13317 [36:39<1:39:49,  1.63it/s]
                                                      
{'loss': 0.181, 'grad_norm': 0.4946650266647339, 'learning_rate': 9.173433485451781e-06, 'epoch': 0.8}

 27%|██▋       | 3557/13317 [36:39<1:39:49,  1.63it/s]
 27%|██▋       | 3558/13317 [36:40<1:39:50,  1.63it/s]
                                                      
{'loss': 0.226, 'grad_norm': 0.550121545791626, 'learning_rate': 9.172711542382236e-06, 'epoch': 0.8}

 27%|██▋       | 3558/13317 [36:40<1:39:50,  1.63it/s]
 27%|██▋       | 3559/13317 [36:40<1:39:49,  1.63it/s]
                                                      
{'loss': 0.2662, 'grad_norm': 0.5758847594261169, 'learning_rate': 9.171989312603226e-06, 'epoch': 0.8}

 27%|██▋       | 3559/13317 [36:40<1:39:49,  1.63it/s]
 27%|██▋       | 3560/13317 [36:41<1:39:47,  1.63it/s]
                                                      
{'loss': 0.3668, 'grad_norm': 0.6665012836456299, 'learning_rate': 9.171266796164377e-06, 'epoch': 0.8}

 27%|██▋       | 3560/13317 [36:41<1:39:47,  1.63it/s]
 27%|██▋       | 3561/13317 [36:42<1:39:48,  1.63it/s]
                                                      
{'loss': 0.2354, 'grad_norm': 0.4509719908237457, 'learning_rate': 9.170543993115333e-06, 'epoch': 0.8}

 27%|██▋       | 3561/13317 [36:42<1:39:48,  1.63it/s]
 27%|██▋       | 3562/13317 [36:42<1:39:51,  1.63it/s]
                                                      
{'loss': 0.2338, 'grad_norm': 0.4968297481536865, 'learning_rate': 9.169820903505758e-06, 'epoch': 0.8}

 27%|██▋       | 3562/13317 [36:42<1:39:51,  1.63it/s]
 27%|██▋       | 3563/13317 [36:43<1:39:45,  1.63it/s]
                                                      
{'loss': 0.2421, 'grad_norm': 0.4830007255077362, 'learning_rate': 9.169097527385337e-06, 'epoch': 0.8}

 27%|██▋       | 3563/13317 [36:43<1:39:45,  1.63it/s]
 27%|██▋       | 3564/13317 [36:43<1:39:43,  1.63it/s]
                                                      
{'loss': 0.2134, 'grad_norm': 0.4783182442188263, 'learning_rate': 9.168373864803773e-06, 'epoch': 0.8}

 27%|██▋       | 3564/13317 [36:43<1:39:43,  1.63it/s]
 27%|██▋       | 3565/13317 [36:44<1:39:42,  1.63it/s]
                                                      
{'loss': 0.2089, 'grad_norm': 0.5204567313194275, 'learning_rate': 9.167649915810787e-06, 'epoch': 0.8}

 27%|██▋       | 3565/13317 [36:44<1:39:42,  1.63it/s]
 27%|██▋       | 3566/13317 [36:45<1:39:39,  1.63it/s]
                                                      
{'loss': 0.2562, 'grad_norm': 0.5748143196105957, 'learning_rate': 9.166925680456124e-06, 'epoch': 0.8}

 27%|██▋       | 3566/13317 [36:45<1:39:39,  1.63it/s]
 27%|██▋       | 3567/13317 [36:45<1:39:36,  1.63it/s]
                                                      
{'loss': 0.2384, 'grad_norm': 0.5005431771278381, 'learning_rate': 9.166201158789548e-06, 'epoch': 0.8}

 27%|██▋       | 3567/13317 [36:45<1:39:36,  1.63it/s]
 27%|██▋       | 3568/13317 [36:46<1:39:40,  1.63it/s]
                                                      
{'loss': 0.2482, 'grad_norm': 0.49233222007751465, 'learning_rate': 9.16547635086084e-06, 'epoch': 0.8}

 27%|██▋       | 3568/13317 [36:46<1:39:40,  1.63it/s]
 27%|██▋       | 3569/13317 [36:46<1:39:34,  1.63it/s]
                                                      
{'loss': 0.308, 'grad_norm': 0.6450995206832886, 'learning_rate': 9.164751256719801e-06, 'epoch': 0.8}

 27%|██▋       | 3569/13317 [36:47<1:39:34,  1.63it/s]
 27%|██▋       | 3570/13317 [36:47<1:39:39,  1.63it/s]
                                                      
{'loss': 0.2792, 'grad_norm': 0.5666971206665039, 'learning_rate': 9.164025876416255e-06, 'epoch': 0.8}

 27%|██▋       | 3570/13317 [36:47<1:39:39,  1.63it/s]
 27%|██▋       | 3571/13317 [36:48<1:39:38,  1.63it/s]
                                                      
{'loss': 0.2173, 'grad_norm': 0.4843234121799469, 'learning_rate': 9.163300210000041e-06, 'epoch': 0.8}

 27%|██▋       | 3571/13317 [36:48<1:39:38,  1.63it/s]
 27%|██▋       | 3572/13317 [36:48<1:39:36,  1.63it/s]
                                                      
{'loss': 0.2069, 'grad_norm': 0.4497396945953369, 'learning_rate': 9.16257425752102e-06, 'epoch': 0.8}

 27%|██▋       | 3572/13317 [36:48<1:39:36,  1.63it/s]
 27%|██▋       | 3573/13317 [36:49<1:39:38,  1.63it/s]
                                                      
{'loss': 0.3057, 'grad_norm': 0.6122580170631409, 'learning_rate': 9.161848019029075e-06, 'epoch': 0.8}

 27%|██▋       | 3573/13317 [36:49<1:39:38,  1.63it/s]
 27%|██▋       | 3574/13317 [36:49<1:39:38,  1.63it/s]
                                                      
{'loss': 0.2066, 'grad_norm': 0.7420292496681213, 'learning_rate': 9.161121494574103e-06, 'epoch': 0.81}

 27%|██▋       | 3574/13317 [36:50<1:39:38,  1.63it/s]
 27%|██▋       | 3575/13317 [36:50<1:39:36,  1.63it/s]
                                                      
{'loss': 0.2641, 'grad_norm': 0.5763885378837585, 'learning_rate': 9.160394684206028e-06, 'epoch': 0.81}

 27%|██▋       | 3575/13317 [36:50<1:39:36,  1.63it/s]
 27%|██▋       | 3576/13317 [36:51<1:39:38,  1.63it/s]
                                                      
{'loss': 0.245, 'grad_norm': 0.44906216859817505, 'learning_rate': 9.159667587974786e-06, 'epoch': 0.81}

 27%|██▋       | 3576/13317 [36:51<1:39:38,  1.63it/s]
 27%|██▋       | 3577/13317 [36:51<1:39:36,  1.63it/s]
                                                      
{'loss': 0.2699, 'grad_norm': 0.5039014220237732, 'learning_rate': 9.158940205930338e-06, 'epoch': 0.81}

 27%|██▋       | 3577/13317 [36:51<1:39:36,  1.63it/s]
 27%|██▋       | 3578/13317 [36:52<1:39:33,  1.63it/s]
                                                      
{'loss': 0.1882, 'grad_norm': 0.44378548860549927, 'learning_rate': 9.15821253812266e-06, 'epoch': 0.81}

 27%|██▋       | 3578/13317 [36:52<1:39:33,  1.63it/s]
 27%|██▋       | 3579/13317 [36:53<1:39:31,  1.63it/s]
                                                      
{'loss': 0.2214, 'grad_norm': 0.46366018056869507, 'learning_rate': 9.157484584601755e-06, 'epoch': 0.81}

 27%|██▋       | 3579/13317 [36:53<1:39:31,  1.63it/s]
 27%|██▋       | 3580/13317 [36:53<1:39:31,  1.63it/s]
                                                      
{'loss': 0.2886, 'grad_norm': 0.5214093327522278, 'learning_rate': 9.156756345417639e-06, 'epoch': 0.81}

 27%|██▋       | 3580/13317 [36:53<1:39:31,  1.63it/s]
 27%|██▋       | 3581/13317 [36:54<1:39:30,  1.63it/s]
                                                      
{'loss': 0.2067, 'grad_norm': 0.4371277987957001, 'learning_rate': 9.156027820620348e-06, 'epoch': 0.81}

 27%|██▋       | 3581/13317 [36:54<1:39:30,  1.63it/s]
 27%|██▋       | 3582/13317 [36:54<1:39:33,  1.63it/s]
                                                      
{'loss': 0.2474, 'grad_norm': 0.4907263219356537, 'learning_rate': 9.155299010259942e-06, 'epoch': 0.81}

 27%|██▋       | 3582/13317 [36:55<1:39:33,  1.63it/s]
 27%|██▋       | 3583/13317 [36:55<1:39:35,  1.63it/s]
                                                      
{'loss': 0.2076, 'grad_norm': 0.5308104753494263, 'learning_rate': 9.154569914386497e-06, 'epoch': 0.81}

 27%|██▋       | 3583/13317 [36:55<1:39:35,  1.63it/s]
 27%|██▋       | 3584/13317 [36:56<1:39:30,  1.63it/s]
                                                      
{'loss': 0.2217, 'grad_norm': 0.527709424495697, 'learning_rate': 9.153840533050109e-06, 'epoch': 0.81}

 27%|██▋       | 3584/13317 [36:56<1:39:30,  1.63it/s]
 27%|██▋       | 3585/13317 [36:56<1:39:30,  1.63it/s]
                                                      
{'loss': 0.2666, 'grad_norm': 0.5982299447059631, 'learning_rate': 9.153110866300896e-06, 'epoch': 0.81}

 27%|██▋       | 3585/13317 [36:56<1:39:30,  1.63it/s]
 27%|██▋       | 3586/13317 [36:57<1:39:28,  1.63it/s]
                                                      
{'loss': 0.2826, 'grad_norm': 0.48263442516326904, 'learning_rate': 9.152380914188992e-06, 'epoch': 0.81}

 27%|██▋       | 3586/13317 [36:57<1:39:28,  1.63it/s]
 27%|██▋       | 3587/13317 [36:57<1:39:30,  1.63it/s]
                                                      
{'loss': 0.2359, 'grad_norm': 0.46296021342277527, 'learning_rate': 9.151650676764553e-06, 'epoch': 0.81}

 27%|██▋       | 3587/13317 [36:58<1:39:30,  1.63it/s]
 27%|██▋       | 3588/13317 [36:58<1:39:27,  1.63it/s]
                                                      
{'loss': 0.3363, 'grad_norm': 0.7155130505561829, 'learning_rate': 9.150920154077753e-06, 'epoch': 0.81}

 27%|██▋       | 3588/13317 [36:58<1:39:27,  1.63it/s]
 27%|██▋       | 3589/13317 [36:59<1:39:30,  1.63it/s]
                                                      
{'loss': 0.2474, 'grad_norm': 0.5254568457603455, 'learning_rate': 9.15018934617879e-06, 'epoch': 0.81}

 27%|██▋       | 3589/13317 [36:59<1:39:30,  1.63it/s]
 27%|██▋       | 3590/13317 [36:59<1:39:30,  1.63it/s]
                                                      
{'loss': 0.1985, 'grad_norm': 0.42762109637260437, 'learning_rate': 9.149458253117876e-06, 'epoch': 0.81}

 27%|██▋       | 3590/13317 [36:59<1:39:30,  1.63it/s]
 27%|██▋       | 3591/13317 [37:00<1:39:26,  1.63it/s]
                                                      
{'loss': 0.2867, 'grad_norm': 0.667864203453064, 'learning_rate': 9.148726874945241e-06, 'epoch': 0.81}

 27%|██▋       | 3591/13317 [37:00<1:39:26,  1.63it/s]
 27%|██▋       | 3592/13317 [37:01<1:39:26,  1.63it/s]
                                                      
{'loss': 0.2893, 'grad_norm': 0.7162400484085083, 'learning_rate': 9.147995211711148e-06, 'epoch': 0.81}

 27%|██▋       | 3592/13317 [37:01<1:39:26,  1.63it/s]
 27%|██▋       | 3593/13317 [37:01<1:39:24,  1.63it/s]
                                                      
{'loss': 0.3134, 'grad_norm': 0.5295892357826233, 'learning_rate': 9.14726326346586e-06, 'epoch': 0.81}

 27%|██▋       | 3593/13317 [37:01<1:39:24,  1.63it/s]
 27%|██▋       | 3594/13317 [37:02<1:45:00,  1.54it/s]
                                                      
{'loss': 0.3206, 'grad_norm': 0.6093184351921082, 'learning_rate': 9.146531030259675e-06, 'epoch': 0.81}

 27%|██▋       | 3594/13317 [37:02<1:45:00,  1.54it/s]
 27%|██▋       | 3595/13317 [37:02<1:43:48,  1.56it/s]
                                                      
{'loss': 0.1675, 'grad_norm': 0.4491075575351715, 'learning_rate': 9.145798512142906e-06, 'epoch': 0.81}

 27%|██▋       | 3595/13317 [37:03<1:43:48,  1.56it/s]
 27%|██▋       | 3596/13317 [37:03<1:42:28,  1.58it/s]
                                                      
{'loss': 0.2391, 'grad_norm': 0.5401059985160828, 'learning_rate': 9.14506570916588e-06, 'epoch': 0.81}

 27%|██▋       | 3596/13317 [37:03<1:42:28,  1.58it/s]
 27%|██▋       | 3597/13317 [37:04<1:42:03,  1.59it/s]
                                                      
{'loss': 0.2542, 'grad_norm': 0.5958550572395325, 'learning_rate': 9.144332621378951e-06, 'epoch': 0.81}

 27%|██▋       | 3597/13317 [37:04<1:42:03,  1.59it/s]
 27%|██▋       | 3598/13317 [37:04<1:41:14,  1.60it/s]
                                                      
{'loss': 0.2299, 'grad_norm': 0.5256568193435669, 'learning_rate': 9.143599248832492e-06, 'epoch': 0.81}

 27%|██▋       | 3598/13317 [37:04<1:41:14,  1.60it/s]
 27%|██▋       | 3599/13317 [37:05<1:40:50,  1.61it/s]
                                                      
{'loss': 0.2216, 'grad_norm': 0.467530757188797, 'learning_rate': 9.14286559157689e-06, 'epoch': 0.81}

 27%|██▋       | 3599/13317 [37:05<1:40:50,  1.61it/s]
 27%|██▋       | 3600/13317 [37:06<1:40:51,  1.61it/s]
                                                      
{'loss': 0.2641, 'grad_norm': 0.5004764199256897, 'learning_rate': 9.142131649662558e-06, 'epoch': 0.81}

 27%|██▋       | 3600/13317 [37:06<1:40:51,  1.61it/s]
 27%|██▋       | 3601/13317 [37:06<1:40:26,  1.61it/s]
                                                      
{'loss': 0.2791, 'grad_norm': 0.5501017570495605, 'learning_rate': 9.141397423139924e-06, 'epoch': 0.81}

 27%|██▋       | 3601/13317 [37:06<1:40:26,  1.61it/s]
 27%|██▋       | 3602/13317 [37:07<1:40:18,  1.61it/s]
                                                      
{'loss': 0.2544, 'grad_norm': 0.5114219784736633, 'learning_rate': 9.140662912059434e-06, 'epoch': 0.81}

 27%|██▋       | 3602/13317 [37:07<1:40:18,  1.61it/s]
 27%|██▋       | 3603/13317 [37:07<1:39:58,  1.62it/s]
                                                      
{'loss': 0.2088, 'grad_norm': 0.4436148405075073, 'learning_rate': 9.139928116471564e-06, 'epoch': 0.81}

 27%|██▋       | 3603/13317 [37:08<1:39:58,  1.62it/s]
 27%|██▋       | 3604/13317 [37:08<1:39:52,  1.62it/s]
                                                      
{'loss': 0.2043, 'grad_norm': 0.4584232270717621, 'learning_rate': 9.139193036426797e-06, 'epoch': 0.81}

 27%|██▋       | 3604/13317 [37:08<1:39:52,  1.62it/s]
 27%|██▋       | 3605/13317 [37:09<1:39:48,  1.62it/s]
                                                      
{'loss': 0.2768, 'grad_norm': 0.5533987283706665, 'learning_rate': 9.13845767197564e-06, 'epoch': 0.81}

 27%|██▋       | 3605/13317 [37:09<1:39:48,  1.62it/s]
 27%|██▋       | 3606/13317 [37:09<1:39:36,  1.62it/s]
                                                      
{'loss': 0.2956, 'grad_norm': 0.5612226724624634, 'learning_rate': 9.137722023168625e-06, 'epoch': 0.81}

 27%|██▋       | 3606/13317 [37:09<1:39:36,  1.62it/s]
 27%|██▋       | 3607/13317 [37:10<1:39:37,  1.62it/s]
                                                      
{'loss': 0.3206, 'grad_norm': 0.6233672499656677, 'learning_rate': 9.136986090056295e-06, 'epoch': 0.81}

 27%|██▋       | 3607/13317 [37:10<1:39:37,  1.62it/s]
 27%|██▋       | 3608/13317 [37:10<1:39:30,  1.63it/s]
                                                      
{'loss': 0.247, 'grad_norm': 0.5126433372497559, 'learning_rate': 9.136249872689216e-06, 'epoch': 0.81}

 27%|██▋       | 3608/13317 [37:11<1:39:30,  1.63it/s]
 27%|██▋       | 3609/13317 [37:11<1:39:30,  1.63it/s]
                                                      
{'loss': 0.2442, 'grad_norm': 0.5456997752189636, 'learning_rate': 9.135513371117977e-06, 'epoch': 0.81}

 27%|██▋       | 3609/13317 [37:11<1:39:30,  1.63it/s]
 27%|██▋       | 3610/13317 [37:12<1:39:29,  1.63it/s]
                                                      
{'loss': 0.1858, 'grad_norm': 0.43644949793815613, 'learning_rate': 9.134776585393181e-06, 'epoch': 0.81}

 27%|██▋       | 3610/13317 [37:12<1:39:29,  1.63it/s]
 27%|██▋       | 3611/13317 [37:12<1:39:22,  1.63it/s]
                                                      
{'loss': 0.2663, 'grad_norm': 0.5505449175834656, 'learning_rate': 9.134039515565454e-06, 'epoch': 0.81}

 27%|██▋       | 3611/13317 [37:12<1:39:22,  1.63it/s]
 27%|██▋       | 3612/13317 [37:13<1:39:27,  1.63it/s]
                                                      
{'loss': 0.2028, 'grad_norm': 0.48136353492736816, 'learning_rate': 9.133302161685442e-06, 'epoch': 0.81}

 27%|██▋       | 3612/13317 [37:13<1:39:27,  1.63it/s]
 27%|██▋       | 3613/13317 [37:14<1:39:21,  1.63it/s]
                                                      
{'loss': 0.2285, 'grad_norm': 0.4373517334461212, 'learning_rate': 9.132564523803807e-06, 'epoch': 0.81}

 27%|██▋       | 3613/13317 [37:14<1:39:21,  1.63it/s]
 27%|██▋       | 3614/13317 [37:14<1:39:18,  1.63it/s]
                                                      
{'loss': 0.2177, 'grad_norm': 0.4640796482563019, 'learning_rate': 9.131826601971232e-06, 'epoch': 0.81}

 27%|██▋       | 3614/13317 [37:14<1:39:18,  1.63it/s]
 27%|██▋       | 3615/13317 [37:15<1:39:18,  1.63it/s]
                                                      
{'loss': 0.1896, 'grad_norm': 0.46155673265457153, 'learning_rate': 9.13108839623842e-06, 'epoch': 0.81}

 27%|██▋       | 3615/13317 [37:15<1:39:18,  1.63it/s]
 27%|██▋       | 3616/13317 [37:15<1:39:16,  1.63it/s]
                                                      
{'loss': 0.3187, 'grad_norm': 0.6019287705421448, 'learning_rate': 9.130349906656095e-06, 'epoch': 0.81}

 27%|██▋       | 3616/13317 [37:16<1:39:16,  1.63it/s]
 27%|██▋       | 3617/13317 [37:16<1:39:13,  1.63it/s]
                                                      
{'loss': 0.2694, 'grad_norm': 0.5132286548614502, 'learning_rate': 9.129611133275e-06, 'epoch': 0.81}

 27%|██▋       | 3617/13317 [37:16<1:39:13,  1.63it/s]
 27%|██▋       | 3618/13317 [37:17<1:39:10,  1.63it/s]
                                                      
{'loss': 0.3079, 'grad_norm': 0.5837533473968506, 'learning_rate': 9.128872076145894e-06, 'epoch': 0.82}

 27%|██▋       | 3618/13317 [37:17<1:39:10,  1.63it/s]
 27%|██▋       | 3619/13317 [37:17<1:39:06,  1.63it/s]
                                                      
{'loss': 0.2494, 'grad_norm': 0.5140951871871948, 'learning_rate': 9.12813273531956e-06, 'epoch': 0.82}

 27%|██▋       | 3619/13317 [37:17<1:39:06,  1.63it/s]
 27%|██▋       | 3620/13317 [37:18<1:39:16,  1.63it/s]
                                                      
{'loss': 0.2104, 'grad_norm': 0.4481223225593567, 'learning_rate': 9.127393110846796e-06, 'epoch': 0.82}

 27%|██▋       | 3620/13317 [37:18<1:39:16,  1.63it/s]
 27%|██▋       | 3621/13317 [37:18<1:39:16,  1.63it/s]
                                                      
{'loss': 0.2496, 'grad_norm': 0.5161937475204468, 'learning_rate': 9.126653202778425e-06, 'epoch': 0.82}

 27%|██▋       | 3621/13317 [37:19<1:39:16,  1.63it/s]
 27%|██▋       | 3622/13317 [37:19<1:39:10,  1.63it/s]
                                                      
{'loss': 0.2282, 'grad_norm': 0.48959505558013916, 'learning_rate': 9.125913011165286e-06, 'epoch': 0.82}

 27%|██▋       | 3622/13317 [37:19<1:39:10,  1.63it/s]
 27%|██▋       | 3623/13317 [37:20<1:39:12,  1.63it/s]
                                                      
{'loss': 0.2375, 'grad_norm': 0.5642036199569702, 'learning_rate': 9.125172536058235e-06, 'epoch': 0.82}

 27%|██▋       | 3623/13317 [37:20<1:39:12,  1.63it/s]
 27%|██▋       | 3624/13317 [37:20<1:39:10,  1.63it/s]
                                                      
{'loss': 0.2312, 'grad_norm': 0.4610452950000763, 'learning_rate': 9.124431777508156e-06, 'epoch': 0.82}

 27%|██▋       | 3624/13317 [37:20<1:39:10,  1.63it/s]
 27%|██▋       | 3625/13317 [37:21<1:39:09,  1.63it/s]
                                                      
{'loss': 0.2568, 'grad_norm': 0.5851590633392334, 'learning_rate': 9.123690735565942e-06, 'epoch': 0.82}

 27%|██▋       | 3625/13317 [37:21<1:39:09,  1.63it/s]
 27%|██▋       | 3626/13317 [37:22<1:39:06,  1.63it/s]
                                                      
{'loss': 0.2903, 'grad_norm': 0.5878082513809204, 'learning_rate': 9.122949410282512e-06, 'epoch': 0.82}

 27%|██▋       | 3626/13317 [37:22<1:39:06,  1.63it/s]
 27%|██▋       | 3627/13317 [37:22<1:39:09,  1.63it/s]
                                                      
{'loss': 0.2203, 'grad_norm': 0.4964502155780792, 'learning_rate': 9.122207801708802e-06, 'epoch': 0.82}

 27%|██▋       | 3627/13317 [37:22<1:39:09,  1.63it/s]
 27%|██▋       | 3628/13317 [37:23<1:39:07,  1.63it/s]
                                                      
{'loss': 0.4815, 'grad_norm': 0.6538137793540955, 'learning_rate': 9.12146590989577e-06, 'epoch': 0.82}

 27%|██▋       | 3628/13317 [37:23<1:39:07,  1.63it/s]
 27%|██▋       | 3629/13317 [37:23<1:39:05,  1.63it/s]
                                                      
{'loss': 0.2822, 'grad_norm': 0.5568472743034363, 'learning_rate': 9.120723734894393e-06, 'epoch': 0.82}

 27%|██▋       | 3629/13317 [37:24<1:39:05,  1.63it/s]
 27%|██▋       | 3630/13317 [37:24<1:39:12,  1.63it/s]
                                                      
{'loss': 0.3684, 'grad_norm': 0.6642332673072815, 'learning_rate': 9.119981276755663e-06, 'epoch': 0.82}

 27%|██▋       | 3630/13317 [37:24<1:39:12,  1.63it/s]
 27%|██▋       | 3631/13317 [37:25<1:39:09,  1.63it/s]
                                                      
{'loss': 0.2778, 'grad_norm': 0.5453833341598511, 'learning_rate': 9.119238535530597e-06, 'epoch': 0.82}

 27%|██▋       | 3631/13317 [37:25<1:39:09,  1.63it/s]
 27%|██▋       | 3632/13317 [37:25<1:39:14,  1.63it/s]
                                                      
{'loss': 0.2604, 'grad_norm': 0.5985252261161804, 'learning_rate': 9.118495511270226e-06, 'epoch': 0.82}

 27%|██▋       | 3632/13317 [37:25<1:39:14,  1.63it/s]
 27%|██▋       | 3633/13317 [37:26<1:39:07,  1.63it/s]
                                                      
{'loss': 0.2933, 'grad_norm': 0.6114388108253479, 'learning_rate': 9.117752204025608e-06, 'epoch': 0.82}

 27%|██▋       | 3633/13317 [37:26<1:39:07,  1.63it/s]
 27%|██▋       | 3634/13317 [37:26<1:39:06,  1.63it/s]
                                                      
{'loss': 0.4409, 'grad_norm': 0.6235356330871582, 'learning_rate': 9.117008613847814e-06, 'epoch': 0.82}

 27%|██▋       | 3634/13317 [37:27<1:39:06,  1.63it/s]
 27%|██▋       | 3635/13317 [37:27<1:39:09,  1.63it/s]
                                                      
{'loss': 0.1914, 'grad_norm': 0.4863748252391815, 'learning_rate': 9.116264740787937e-06, 'epoch': 0.82}

 27%|██▋       | 3635/13317 [37:27<1:39:09,  1.63it/s]
 27%|██▋       | 3636/13317 [37:28<1:39:04,  1.63it/s]
                                                      
{'loss': 0.1629, 'grad_norm': 0.42066070437431335, 'learning_rate': 9.115520584897088e-06, 'epoch': 0.82}

 27%|██▋       | 3636/13317 [37:28<1:39:04,  1.63it/s]
 27%|██▋       | 3637/13317 [37:28<1:38:57,  1.63it/s]
                                                      
{'loss': 0.1999, 'grad_norm': 0.4847801923751831, 'learning_rate': 9.1147761462264e-06, 'epoch': 0.82}

 27%|██▋       | 3637/13317 [37:28<1:38:57,  1.63it/s]
 27%|██▋       | 3638/13317 [37:29<1:38:56,  1.63it/s]
                                                      
{'loss': 0.2114, 'grad_norm': 0.568091094493866, 'learning_rate': 9.114031424827021e-06, 'epoch': 0.82}

 27%|██▋       | 3638/13317 [37:29<1:38:56,  1.63it/s]
 27%|██▋       | 3639/13317 [37:30<1:38:55,  1.63it/s]
                                                      
{'loss': 0.2715, 'grad_norm': 0.5709511041641235, 'learning_rate': 9.113286420750125e-06, 'epoch': 0.82}

 27%|██▋       | 3639/13317 [37:30<1:38:55,  1.63it/s]
 27%|██▋       | 3640/13317 [37:30<1:39:02,  1.63it/s]
                                                      
{'loss': 0.2734, 'grad_norm': 0.5857031941413879, 'learning_rate': 9.112541134046898e-06, 'epoch': 0.82}

 27%|██▋       | 3640/13317 [37:30<1:39:02,  1.63it/s]
 27%|██▋       | 3641/13317 [37:31<1:38:58,  1.63it/s]
                                                      
{'loss': 0.2312, 'grad_norm': 0.5581566095352173, 'learning_rate': 9.111795564768551e-06, 'epoch': 0.82}

 27%|██▋       | 3641/13317 [37:31<1:38:58,  1.63it/s]
 27%|██▋       | 3642/13317 [37:31<1:38:55,  1.63it/s]
                                                      
{'loss': 0.2186, 'grad_norm': 0.45483657717704773, 'learning_rate': 9.111049712966313e-06, 'epoch': 0.82}

 27%|██▋       | 3642/13317 [37:32<1:38:55,  1.63it/s]
 27%|██▋       | 3643/13317 [37:32<1:38:59,  1.63it/s]
                                                      
{'loss': 0.2334, 'grad_norm': 0.48400983214378357, 'learning_rate': 9.110303578691432e-06, 'epoch': 0.82}

 27%|██▋       | 3643/13317 [37:32<1:38:59,  1.63it/s]
 27%|██▋       | 3644/13317 [37:33<1:38:58,  1.63it/s]
                                                      
{'loss': 0.2111, 'grad_norm': 0.47193464636802673, 'learning_rate': 9.109557161995172e-06, 'epoch': 0.82}

 27%|██▋       | 3644/13317 [37:33<1:38:58,  1.63it/s]
 27%|██▋       | 3645/13317 [37:33<1:39:00,  1.63it/s]
                                                      
{'loss': 0.2208, 'grad_norm': 0.5171938538551331, 'learning_rate': 9.108810462928826e-06, 'epoch': 0.82}

 27%|██▋       | 3645/13317 [37:33<1:39:00,  1.63it/s]
 27%|██▋       | 3646/13317 [37:34<1:38:59,  1.63it/s]
                                                      
{'loss': 0.2439, 'grad_norm': 0.5906844735145569, 'learning_rate': 9.108063481543694e-06, 'epoch': 0.82}

 27%|██▋       | 3646/13317 [37:34<1:38:59,  1.63it/s]
 27%|██▋       | 3647/13317 [37:34<1:38:56,  1.63it/s]
                                                      
{'loss': 0.2079, 'grad_norm': 0.474918931722641, 'learning_rate': 9.107316217891105e-06, 'epoch': 0.82}

 27%|██▋       | 3647/13317 [37:35<1:38:56,  1.63it/s]
 27%|██▋       | 3648/13317 [37:35<1:38:57,  1.63it/s]
                                                      
{'loss': 0.2266, 'grad_norm': 0.5035297870635986, 'learning_rate': 9.1065686720224e-06, 'epoch': 0.82}

 27%|██▋       | 3648/13317 [37:35<1:38:57,  1.63it/s]
 27%|██▋       | 3649/13317 [37:36<1:38:55,  1.63it/s]
                                                      
{'loss': 0.2784, 'grad_norm': 0.5192338824272156, 'learning_rate': 9.10582084398895e-06, 'epoch': 0.82}

 27%|██▋       | 3649/13317 [37:36<1:38:55,  1.63it/s]
 27%|██▋       | 3650/13317 [37:36<1:38:52,  1.63it/s]
                                                      
{'loss': 0.2731, 'grad_norm': 0.5656454563140869, 'learning_rate': 9.105072733842132e-06, 'epoch': 0.82}

 27%|██▋       | 3650/13317 [37:36<1:38:52,  1.63it/s]
 27%|██▋       | 3651/13317 [37:37<1:38:49,  1.63it/s]
                                                      
{'loss': 0.2958, 'grad_norm': 0.5907272100448608, 'learning_rate': 9.104324341633354e-06, 'epoch': 0.82}

 27%|██▋       | 3651/13317 [37:37<1:38:49,  1.63it/s]
 27%|██▋       | 3652/13317 [37:38<1:38:47,  1.63it/s]
                                                      
{'loss': 0.2047, 'grad_norm': 0.49980834126472473, 'learning_rate': 9.103575667414036e-06, 'epoch': 0.82}

 27%|██▋       | 3652/13317 [37:38<1:38:47,  1.63it/s]
 27%|██▋       | 3653/13317 [37:38<1:38:46,  1.63it/s]
                                                      
{'loss': 0.302, 'grad_norm': 0.5844130516052246, 'learning_rate': 9.102826711235619e-06, 'epoch': 0.82}

 27%|██▋       | 3653/13317 [37:38<1:38:46,  1.63it/s]
 27%|██▋       | 3654/13317 [37:39<1:38:46,  1.63it/s]
                                                      
{'loss': 0.2167, 'grad_norm': 0.5354143977165222, 'learning_rate': 9.102077473149566e-06, 'epoch': 0.82}

 27%|██▋       | 3654/13317 [37:39<1:38:46,  1.63it/s]
 27%|██▋       | 3655/13317 [37:39<1:38:46,  1.63it/s]
                                                      
{'loss': 0.2595, 'grad_norm': 0.5520585179328918, 'learning_rate': 9.10132795320736e-06, 'epoch': 0.82}

 27%|██▋       | 3655/13317 [37:39<1:38:46,  1.63it/s]
 27%|██▋       | 3656/13317 [37:40<1:38:46,  1.63it/s]
                                                      
{'loss': 0.1921, 'grad_norm': 0.4555610716342926, 'learning_rate': 9.100578151460494e-06, 'epoch': 0.82}

 27%|██▋       | 3656/13317 [37:40<1:38:46,  1.63it/s]
 27%|██▋       | 3657/13317 [37:41<1:38:45,  1.63it/s]
                                                      
{'loss': 0.2214, 'grad_norm': 0.48916900157928467, 'learning_rate': 9.099828067960493e-06, 'epoch': 0.82}

 27%|██▋       | 3657/13317 [37:41<1:38:45,  1.63it/s]
 27%|██▋       | 3658/13317 [37:41<1:38:42,  1.63it/s]
                                                      
{'loss': 0.239, 'grad_norm': 0.48995205760002136, 'learning_rate': 9.099077702758894e-06, 'epoch': 0.82}

 27%|██▋       | 3658/13317 [37:41<1:38:42,  1.63it/s]
 27%|██▋       | 3659/13317 [37:42<1:38:39,  1.63it/s]
                                                      
{'loss': 0.2426, 'grad_norm': 0.5179347395896912, 'learning_rate': 9.098327055907254e-06, 'epoch': 0.82}

 27%|██▋       | 3659/13317 [37:42<1:38:39,  1.63it/s]
 27%|██▋       | 3660/13317 [37:42<1:38:39,  1.63it/s]
                                                      
{'loss': 0.2171, 'grad_norm': 0.45009246468544006, 'learning_rate': 9.097576127457154e-06, 'epoch': 0.82}

 27%|██▋       | 3660/13317 [37:43<1:38:39,  1.63it/s]
 27%|██▋       | 3661/13317 [37:43<1:38:38,  1.63it/s]
                                                      
{'loss': 0.215, 'grad_norm': 0.4836113452911377, 'learning_rate': 9.096824917460186e-06, 'epoch': 0.82}

 27%|██▋       | 3661/13317 [37:43<1:38:38,  1.63it/s]
 27%|██▋       | 3662/13317 [37:44<1:38:38,  1.63it/s]
                                                      
{'loss': 0.2553, 'grad_norm': 0.5377118587493896, 'learning_rate': 9.09607342596797e-06, 'epoch': 0.82}

 27%|██▋       | 3662/13317 [37:44<1:38:38,  1.63it/s]
 28%|██▊       | 3663/13317 [37:44<1:38:37,  1.63it/s]
                                                      
{'loss': 0.2154, 'grad_norm': 0.5630577206611633, 'learning_rate': 9.09532165303214e-06, 'epoch': 0.83}

 28%|██▊       | 3663/13317 [37:44<1:38:37,  1.63it/s]
 28%|██▊       | 3664/13317 [37:45<1:38:36,  1.63it/s]
                                                      
{'loss': 0.2027, 'grad_norm': 0.492109477519989, 'learning_rate': 9.094569598704349e-06, 'epoch': 0.83}

 28%|██▊       | 3664/13317 [37:45<1:38:36,  1.63it/s]
 28%|██▊       | 3665/13317 [37:45<1:38:37,  1.63it/s]
                                                      
{'loss': 0.2247, 'grad_norm': 0.47018489241600037, 'learning_rate': 9.09381726303627e-06, 'epoch': 0.83}

 28%|██▊       | 3665/13317 [37:46<1:38:37,  1.63it/s]
 28%|██▊       | 3666/13317 [37:46<1:38:36,  1.63it/s]
                                                      
{'loss': 0.2424, 'grad_norm': 0.5520727634429932, 'learning_rate': 9.093064646079604e-06, 'epoch': 0.83}

 28%|██▊       | 3666/13317 [37:46<1:38:36,  1.63it/s]
 28%|██▊       | 3667/13317 [37:47<1:38:34,  1.63it/s]
                                                      
{'loss': 0.2921, 'grad_norm': 0.6167296171188354, 'learning_rate': 9.092311747886056e-06, 'epoch': 0.83}

 28%|██▊       | 3667/13317 [37:47<1:38:34,  1.63it/s]
 28%|██▊       | 3668/13317 [37:47<1:38:35,  1.63it/s]
                                                      
{'loss': 0.2318, 'grad_norm': 0.5029744505882263, 'learning_rate': 9.091558568507362e-06, 'epoch': 0.83}

 28%|██▊       | 3668/13317 [37:47<1:38:35,  1.63it/s]
 28%|██▊       | 3669/13317 [37:48<1:38:32,  1.63it/s]
                                                      
{'loss': 0.2054, 'grad_norm': 0.5710597038269043, 'learning_rate': 9.09080510799527e-06, 'epoch': 0.83}

 28%|██▊       | 3669/13317 [37:48<1:38:32,  1.63it/s]
 28%|██▊       | 3670/13317 [37:49<1:38:32,  1.63it/s]
                                                      
{'loss': 0.2331, 'grad_norm': 0.5189693570137024, 'learning_rate': 9.090051366401556e-06, 'epoch': 0.83}

 28%|██▊       | 3670/13317 [37:49<1:38:32,  1.63it/s]
 28%|██▊       | 3671/13317 [37:49<1:38:31,  1.63it/s]
                                                      
{'loss': 0.3344, 'grad_norm': 0.6132514476776123, 'learning_rate': 9.089297343778005e-06, 'epoch': 0.83}

 28%|██▊       | 3671/13317 [37:49<1:38:31,  1.63it/s]
 28%|██▊       | 3672/13317 [37:50<1:38:31,  1.63it/s]
                                                      
{'loss': 0.1783, 'grad_norm': 0.44221845269203186, 'learning_rate': 9.08854304017643e-06, 'epoch': 0.83}

 28%|██▊       | 3672/13317 [37:50<1:38:31,  1.63it/s]
 28%|██▊       | 3673/13317 [37:50<1:38:38,  1.63it/s]
                                                      
{'loss': 0.1984, 'grad_norm': 0.4644451141357422, 'learning_rate': 9.087788455648657e-06, 'epoch': 0.83}

 28%|██▊       | 3673/13317 [37:51<1:38:38,  1.63it/s]
 28%|██▊       | 3674/13317 [37:51<1:38:36,  1.63it/s]
                                                      
{'loss': 0.3107, 'grad_norm': 0.5494344234466553, 'learning_rate': 9.087033590246535e-06, 'epoch': 0.83}

 28%|██▊       | 3674/13317 [37:51<1:38:36,  1.63it/s]
 28%|██▊       | 3675/13317 [37:52<1:38:36,  1.63it/s]
                                                      
{'loss': 0.2558, 'grad_norm': 0.544126570224762, 'learning_rate': 9.08627844402193e-06, 'epoch': 0.83}

 28%|██▊       | 3675/13317 [37:52<1:38:36,  1.63it/s]
 28%|██▊       | 3676/13317 [37:52<1:38:32,  1.63it/s]
                                                      
{'loss': 0.2133, 'grad_norm': 0.4526440501213074, 'learning_rate': 9.08552301702673e-06, 'epoch': 0.83}

 28%|██▊       | 3676/13317 [37:52<1:38:32,  1.63it/s]
 28%|██▊       | 3677/13317 [37:53<1:38:30,  1.63it/s]
                                                      
{'loss': 0.2622, 'grad_norm': 0.5216955542564392, 'learning_rate': 9.084767309312842e-06, 'epoch': 0.83}

 28%|██▊       | 3677/13317 [37:53<1:38:30,  1.63it/s]
 28%|██▊       | 3678/13317 [37:53<1:38:32,  1.63it/s]
                                                      
{'loss': 0.2493, 'grad_norm': 0.5137353539466858, 'learning_rate': 9.08401132093219e-06, 'epoch': 0.83}

 28%|██▊       | 3678/13317 [37:54<1:38:32,  1.63it/s]
 28%|██▊       | 3679/13317 [37:54<1:38:27,  1.63it/s]
                                                      
{'loss': 0.2632, 'grad_norm': 0.5656010508537292, 'learning_rate': 9.083255051936715e-06, 'epoch': 0.83}

 28%|██▊       | 3679/13317 [37:54<1:38:27,  1.63it/s]
 28%|██▊       | 3680/13317 [37:55<1:38:25,  1.63it/s]
                                                      
{'loss': 0.2052, 'grad_norm': 0.4988196790218353, 'learning_rate': 9.082498502378385e-06, 'epoch': 0.83}

 28%|██▊       | 3680/13317 [37:55<1:38:25,  1.63it/s]
 28%|██▊       | 3681/13317 [37:55<1:38:28,  1.63it/s]
                                                      
{'loss': 0.3111, 'grad_norm': 0.7381272912025452, 'learning_rate': 9.081741672309183e-06, 'epoch': 0.83}

 28%|██▊       | 3681/13317 [37:55<1:38:28,  1.63it/s]
 28%|██▊       | 3682/13317 [37:56<1:38:29,  1.63it/s]
                                                      
{'loss': 0.2319, 'grad_norm': 0.5333594083786011, 'learning_rate': 9.08098456178111e-06, 'epoch': 0.83}

 28%|██▊       | 3682/13317 [37:56<1:38:29,  1.63it/s]
 28%|██▊       | 3683/13317 [37:57<1:38:26,  1.63it/s]
                                                      
{'loss': 0.3128, 'grad_norm': 0.592070460319519, 'learning_rate': 9.080227170846185e-06, 'epoch': 0.83}

 28%|██▊       | 3683/13317 [37:57<1:38:26,  1.63it/s]
 28%|██▊       | 3684/13317 [37:57<1:38:23,  1.63it/s]
                                                      
{'loss': 0.2642, 'grad_norm': 0.5541512966156006, 'learning_rate': 9.079469499556453e-06, 'epoch': 0.83}

 28%|██▊       | 3684/13317 [37:57<1:38:23,  1.63it/s]
 28%|██▊       | 3685/13317 [37:58<1:38:24,  1.63it/s]
                                                      
{'loss': 0.2824, 'grad_norm': 0.5401637554168701, 'learning_rate': 9.078711547963972e-06, 'epoch': 0.83}

 28%|██▊       | 3685/13317 [37:58<1:38:24,  1.63it/s]
 28%|██▊       | 3686/13317 [37:58<1:38:27,  1.63it/s]
                                                      
{'loss': 0.2555, 'grad_norm': 0.550029456615448, 'learning_rate': 9.077953316120822e-06, 'epoch': 0.83}

 28%|██▊       | 3686/13317 [37:58<1:38:27,  1.63it/s]
 28%|██▊       | 3687/13317 [37:59<1:38:29,  1.63it/s]
                                                      
{'loss': 0.3035, 'grad_norm': 0.5872386693954468, 'learning_rate': 9.077194804079102e-06, 'epoch': 0.83}

 28%|██▊       | 3687/13317 [37:59<1:38:29,  1.63it/s]
 28%|██▊       | 3688/13317 [38:00<1:38:33,  1.63it/s]
                                                      
{'loss': 0.2813, 'grad_norm': 0.5498243570327759, 'learning_rate': 9.076436011890928e-06, 'epoch': 0.83}

 28%|██▊       | 3688/13317 [38:00<1:38:33,  1.63it/s]
 28%|██▊       | 3689/13317 [38:00<1:38:30,  1.63it/s]
                                                      
{'loss': 0.2116, 'grad_norm': 0.474519282579422, 'learning_rate': 9.075676939608437e-06, 'epoch': 0.83}

 28%|██▊       | 3689/13317 [38:00<1:38:30,  1.63it/s]
 28%|██▊       | 3690/13317 [38:01<1:38:28,  1.63it/s]
                                                      
{'loss': 0.2151, 'grad_norm': 0.5005728006362915, 'learning_rate': 9.074917587283787e-06, 'epoch': 0.83}

 28%|██▊       | 3690/13317 [38:01<1:38:28,  1.63it/s]
 28%|██▊       | 3691/13317 [38:01<1:38:30,  1.63it/s]
                                                      
{'loss': 0.1887, 'grad_norm': 0.4793955981731415, 'learning_rate': 9.074157954969152e-06, 'epoch': 0.83}

 28%|██▊       | 3691/13317 [38:02<1:38:30,  1.63it/s]
 28%|██▊       | 3692/13317 [38:02<1:38:26,  1.63it/s]
                                                      
{'loss': 0.1373, 'grad_norm': 0.36044153571128845, 'learning_rate': 9.073398042716728e-06, 'epoch': 0.83}

 28%|██▊       | 3692/13317 [38:02<1:38:26,  1.63it/s]
 28%|██▊       | 3693/13317 [38:03<1:38:22,  1.63it/s]
                                                      
{'loss': 0.2535, 'grad_norm': 0.550033688545227, 'learning_rate': 9.07263785057873e-06, 'epoch': 0.83}

 28%|██▊       | 3693/13317 [38:03<1:38:22,  1.63it/s]
 28%|██▊       | 3694/13317 [38:03<1:38:23,  1.63it/s]
                                                      
{'loss': 0.2494, 'grad_norm': 0.4639136791229248, 'learning_rate': 9.07187737860739e-06, 'epoch': 0.83}

 28%|██▊       | 3694/13317 [38:03<1:38:23,  1.63it/s]
 28%|██▊       | 3695/13317 [38:04<1:38:22,  1.63it/s]
                                                      
{'loss': 0.2962, 'grad_norm': 0.5552821159362793, 'learning_rate': 9.071116626854958e-06, 'epoch': 0.83}

 28%|██▊       | 3695/13317 [38:04<1:38:22,  1.63it/s]
 28%|██▊       | 3696/13317 [38:04<1:38:21,  1.63it/s]
                                                      
{'loss': 0.2942, 'grad_norm': 0.7008855938911438, 'learning_rate': 9.07035559537371e-06, 'epoch': 0.83}

 28%|██▊       | 3696/13317 [38:05<1:38:21,  1.63it/s]
 28%|██▊       | 3697/13317 [38:05<1:38:20,  1.63it/s]
                                                      
{'loss': 0.2728, 'grad_norm': 0.51970374584198, 'learning_rate': 9.069594284215935e-06, 'epoch': 0.83}

 28%|██▊       | 3697/13317 [38:05<1:38:20,  1.63it/s]
 28%|██▊       | 3698/13317 [38:06<1:38:21,  1.63it/s]
                                                      
{'loss': 0.2345, 'grad_norm': 0.43950721621513367, 'learning_rate': 9.068832693433942e-06, 'epoch': 0.83}

 28%|██▊       | 3698/13317 [38:06<1:38:21,  1.63it/s]
 28%|██▊       | 3699/13317 [38:06<1:38:18,  1.63it/s]
                                                      
{'loss': 0.2204, 'grad_norm': 0.49050775170326233, 'learning_rate': 9.068070823080062e-06, 'epoch': 0.83}

 28%|██▊       | 3699/13317 [38:06<1:38:18,  1.63it/s]
 28%|██▊       | 3700/13317 [38:07<1:38:19,  1.63it/s]
                                                      
{'loss': 0.2779, 'grad_norm': 0.5273403525352478, 'learning_rate': 9.067308673206642e-06, 'epoch': 0.83}

 28%|██▊       | 3700/13317 [38:07<1:38:19,  1.63it/s]
 28%|██▊       | 3701/13317 [38:08<1:38:18,  1.63it/s]
                                                      
{'loss': 0.1927, 'grad_norm': 0.44787436723709106, 'learning_rate': 9.066546243866052e-06, 'epoch': 0.83}

 28%|██▊       | 3701/13317 [38:08<1:38:18,  1.63it/s]
 28%|██▊       | 3702/13317 [38:08<1:38:18,  1.63it/s]
                                                      
{'loss': 0.2973, 'grad_norm': 0.5325551629066467, 'learning_rate': 9.065783535110678e-06, 'epoch': 0.83}

 28%|██▊       | 3702/13317 [38:08<1:38:18,  1.63it/s]
 28%|██▊       | 3703/13317 [38:09<1:38:17,  1.63it/s]
                                                      
{'loss': 0.2359, 'grad_norm': 0.547703206539154, 'learning_rate': 9.065020546992927e-06, 'epoch': 0.83}

 28%|██▊       | 3703/13317 [38:09<1:38:17,  1.63it/s]
 28%|██▊       | 3704/13317 [38:09<1:38:14,  1.63it/s]
                                                      
{'loss': 0.2038, 'grad_norm': 0.5071679353713989, 'learning_rate': 9.064257279565221e-06, 'epoch': 0.83}

 28%|██▊       | 3704/13317 [38:10<1:38:14,  1.63it/s]
 28%|██▊       | 3705/13317 [38:10<1:38:16,  1.63it/s]
                                                      
{'loss': 0.2324, 'grad_norm': 0.5476030111312866, 'learning_rate': 9.063493732880007e-06, 'epoch': 0.83}

 28%|██▊       | 3705/13317 [38:10<1:38:16,  1.63it/s]
 28%|██▊       | 3706/13317 [38:11<1:38:17,  1.63it/s]
                                                      
{'loss': 0.3207, 'grad_norm': 0.5542616844177246, 'learning_rate': 9.062729906989749e-06, 'epoch': 0.83}

 28%|██▊       | 3706/13317 [38:11<1:38:17,  1.63it/s]
 28%|██▊       | 3707/13317 [38:11<1:38:15,  1.63it/s]
                                                      
{'loss': 0.2038, 'grad_norm': 0.4833027124404907, 'learning_rate': 9.06196580194693e-06, 'epoch': 0.84}

 28%|██▊       | 3707/13317 [38:11<1:38:15,  1.63it/s]
 28%|██▊       | 3708/13317 [38:12<1:38:11,  1.63it/s]
                                                      
{'loss': 0.2773, 'grad_norm': 0.5890411734580994, 'learning_rate': 9.061201417804051e-06, 'epoch': 0.84}

 28%|██▊       | 3708/13317 [38:12<1:38:11,  1.63it/s]
 28%|██▊       | 3709/13317 [38:12<1:38:09,  1.63it/s]
                                                      
{'loss': 0.2896, 'grad_norm': 0.5517049431800842, 'learning_rate': 9.060436754613635e-06, 'epoch': 0.84}

 28%|██▊       | 3709/13317 [38:13<1:38:09,  1.63it/s]
 28%|██▊       | 3710/13317 [38:13<1:38:07,  1.63it/s]
                                                      
{'loss': 0.2459, 'grad_norm': 0.593850314617157, 'learning_rate': 9.05967181242822e-06, 'epoch': 0.84}

 28%|██▊       | 3710/13317 [38:13<1:38:07,  1.63it/s]
 28%|██▊       | 3711/13317 [38:14<1:38:07,  1.63it/s]
                                                      
{'loss': 0.3382, 'grad_norm': 0.6697419285774231, 'learning_rate': 9.058906591300369e-06, 'epoch': 0.84}

 28%|██▊       | 3711/13317 [38:14<1:38:07,  1.63it/s]
 28%|██▊       | 3712/13317 [38:14<1:38:09,  1.63it/s]
                                                      
{'loss': 0.2726, 'grad_norm': 0.5075836777687073, 'learning_rate': 9.058141091282656e-06, 'epoch': 0.84}

 28%|██▊       | 3712/13317 [38:14<1:38:09,  1.63it/s]
 28%|██▊       | 3713/13317 [38:15<1:38:07,  1.63it/s]
                                                      
{'loss': 0.1606, 'grad_norm': 0.37603726983070374, 'learning_rate': 9.057375312427683e-06, 'epoch': 0.84}

 28%|██▊       | 3713/13317 [38:15<1:38:07,  1.63it/s]
 28%|██▊       | 3714/13317 [38:16<1:38:06,  1.63it/s]
                                                      
{'loss': 0.2795, 'grad_norm': 0.5168750882148743, 'learning_rate': 9.056609254788066e-06, 'epoch': 0.84}

 28%|██▊       | 3714/13317 [38:16<1:38:06,  1.63it/s]
 28%|██▊       | 3715/13317 [38:16<1:38:04,  1.63it/s]
                                                      
{'loss': 0.2387, 'grad_norm': 0.4838539659976959, 'learning_rate': 9.055842918416443e-06, 'epoch': 0.84}

 28%|██▊       | 3715/13317 [38:16<1:38:04,  1.63it/s]
 28%|██▊       | 3716/13317 [38:17<1:38:02,  1.63it/s]
                                                      
{'loss': 0.3387, 'grad_norm': 0.6066830158233643, 'learning_rate': 9.055076303365465e-06, 'epoch': 0.84}

 28%|██▊       | 3716/13317 [38:17<1:38:02,  1.63it/s]
 28%|██▊       | 3717/13317 [38:17<1:38:05,  1.63it/s]
                                                      
{'loss': 0.2413, 'grad_norm': 0.5331384539604187, 'learning_rate': 9.054309409687809e-06, 'epoch': 0.84}

 28%|██▊       | 3717/13317 [38:18<1:38:05,  1.63it/s]
 28%|██▊       | 3718/13317 [38:18<1:38:04,  1.63it/s]
                                                      
{'loss': 0.2642, 'grad_norm': 0.5625419616699219, 'learning_rate': 9.05354223743617e-06, 'epoch': 0.84}

 28%|██▊       | 3718/13317 [38:18<1:38:04,  1.63it/s]
 28%|██▊       | 3719/13317 [38:19<1:38:03,  1.63it/s]
                                                      
{'loss': 0.2326, 'grad_norm': 0.5137094855308533, 'learning_rate': 9.052774786663259e-06, 'epoch': 0.84}

 28%|██▊       | 3719/13317 [38:19<1:38:03,  1.63it/s]
 28%|██▊       | 3720/13317 [38:19<1:38:02,  1.63it/s]
                                                      
{'loss': 0.239, 'grad_norm': 0.4414961338043213, 'learning_rate': 9.05200705742181e-06, 'epoch': 0.84}

 28%|██▊       | 3720/13317 [38:19<1:38:02,  1.63it/s]
 28%|██▊       | 3721/13317 [38:20<1:38:06,  1.63it/s]
                                                      
{'loss': 0.2675, 'grad_norm': 0.5107757449150085, 'learning_rate': 9.05123904976457e-06, 'epoch': 0.84}

 28%|██▊       | 3721/13317 [38:20<1:38:06,  1.63it/s]
 28%|██▊       | 3722/13317 [38:20<1:38:03,  1.63it/s]
                                                      
{'loss': 0.2661, 'grad_norm': 0.5981664061546326, 'learning_rate': 9.050470763744315e-06, 'epoch': 0.84}

 28%|██▊       | 3722/13317 [38:21<1:38:03,  1.63it/s]
 28%|██▊       | 3723/13317 [38:21<1:38:03,  1.63it/s]
                                                      
{'loss': 0.2636, 'grad_norm': 0.5606467127799988, 'learning_rate': 9.04970219941383e-06, 'epoch': 0.84}

 28%|██▊       | 3723/13317 [38:21<1:38:03,  1.63it/s]
 28%|██▊       | 3724/13317 [38:22<1:38:00,  1.63it/s]
                                                      
{'loss': 0.2198, 'grad_norm': 0.4739685356616974, 'learning_rate': 9.048933356825927e-06, 'epoch': 0.84}

 28%|██▊       | 3724/13317 [38:22<1:38:00,  1.63it/s]
 28%|██▊       | 3725/13317 [38:22<1:38:01,  1.63it/s]
                                                      
{'loss': 0.2156, 'grad_norm': 0.44038262963294983, 'learning_rate': 9.04816423603343e-06, 'epoch': 0.84}

 28%|██▊       | 3725/13317 [38:22<1:38:01,  1.63it/s]
 28%|██▊       | 3726/13317 [38:23<1:38:01,  1.63it/s]
                                                      
{'loss': 0.2501, 'grad_norm': 0.5570729970932007, 'learning_rate': 9.047394837089186e-06, 'epoch': 0.84}

 28%|██▊       | 3726/13317 [38:23<1:38:01,  1.63it/s]
 28%|██▊       | 3727/13317 [38:24<1:38:00,  1.63it/s]
                                                      
{'loss': 0.3149, 'grad_norm': 0.5543330311775208, 'learning_rate': 9.046625160046063e-06, 'epoch': 0.84}

 28%|██▊       | 3727/13317 [38:24<1:38:00,  1.63it/s]
 28%|██▊       | 3728/13317 [38:24<1:38:00,  1.63it/s]
                                                      
{'loss': 0.2474, 'grad_norm': 0.520582377910614, 'learning_rate': 9.045855204956946e-06, 'epoch': 0.84}

 28%|██▊       | 3728/13317 [38:24<1:38:00,  1.63it/s]
 28%|██▊       | 3729/13317 [38:25<1:38:02,  1.63it/s]
                                                      
{'loss': 0.2895, 'grad_norm': 0.5698521733283997, 'learning_rate': 9.045084971874738e-06, 'epoch': 0.84}

 28%|██▊       | 3729/13317 [38:25<1:38:02,  1.63it/s]
 28%|██▊       | 3730/13317 [38:25<1:38:02,  1.63it/s]
                                                      
{'loss': 0.2826, 'grad_norm': 0.5456900596618652, 'learning_rate': 9.044314460852362e-06, 'epoch': 0.84}

 28%|██▊       | 3730/13317 [38:25<1:38:02,  1.63it/s]
 28%|██▊       | 3731/13317 [38:26<1:38:02,  1.63it/s]
                                                      
{'loss': 0.2293, 'grad_norm': 0.49746522307395935, 'learning_rate': 9.04354367194276e-06, 'epoch': 0.84}

 28%|██▊       | 3731/13317 [38:26<1:38:02,  1.63it/s]
 28%|██▊       | 3732/13317 [38:27<1:38:03,  1.63it/s]
                                                      
{'loss': 0.2237, 'grad_norm': 0.5935807228088379, 'learning_rate': 9.042772605198896e-06, 'epoch': 0.84}

 28%|██▊       | 3732/13317 [38:27<1:38:03,  1.63it/s]
 28%|██▊       | 3733/13317 [38:27<1:37:59,  1.63it/s]
                                                      
{'loss': 0.1808, 'grad_norm': 0.47666335105895996, 'learning_rate': 9.042001260673748e-06, 'epoch': 0.84}

 28%|██▊       | 3733/13317 [38:27<1:37:59,  1.63it/s]
 28%|██▊       | 3734/13317 [38:28<1:37:59,  1.63it/s]
                                                      
{'loss': 0.2209, 'grad_norm': 0.5000836849212646, 'learning_rate': 9.041229638420315e-06, 'epoch': 0.84}

 28%|██▊       | 3734/13317 [38:28<1:37:59,  1.63it/s]
 28%|██▊       | 3735/13317 [38:28<1:37:55,  1.63it/s]
                                                      
{'loss': 0.3154, 'grad_norm': 0.522454559803009, 'learning_rate': 9.040457738491618e-06, 'epoch': 0.84}

 28%|██▊       | 3735/13317 [38:29<1:37:55,  1.63it/s]
 28%|██▊       | 3736/13317 [38:29<1:37:57,  1.63it/s]
                                                      
{'loss': 0.3039, 'grad_norm': 0.584898054599762, 'learning_rate': 9.039685560940691e-06, 'epoch': 0.84}

 28%|██▊       | 3736/13317 [38:29<1:37:57,  1.63it/s]
 28%|██▊       | 3737/13317 [38:30<1:37:57,  1.63it/s]
                                                      
{'loss': 0.1631, 'grad_norm': 0.43257206678390503, 'learning_rate': 9.038913105820595e-06, 'epoch': 0.84}

 28%|██▊       | 3737/13317 [38:30<1:37:57,  1.63it/s]
 28%|██▊       | 3738/13317 [38:30<1:37:57,  1.63it/s]
                                                      
{'loss': 0.2668, 'grad_norm': 0.6046870946884155, 'learning_rate': 9.038140373184404e-06, 'epoch': 0.84}

 28%|██▊       | 3738/13317 [38:30<1:37:57,  1.63it/s]
 28%|██▊       | 3739/13317 [38:31<1:37:54,  1.63it/s]
                                                      
{'loss': 0.2122, 'grad_norm': 0.5380891561508179, 'learning_rate': 9.03736736308521e-06, 'epoch': 0.84}

 28%|██▊       | 3739/13317 [38:31<1:37:54,  1.63it/s]
 28%|██▊       | 3740/13317 [38:31<1:37:55,  1.63it/s]
                                                      
{'loss': 0.2394, 'grad_norm': 0.5338062644004822, 'learning_rate': 9.036594075576133e-06, 'epoch': 0.84}

 28%|██▊       | 3740/13317 [38:32<1:37:55,  1.63it/s]
 28%|██▊       | 3741/13317 [38:32<1:37:53,  1.63it/s]
                                                      
{'loss': 0.198, 'grad_norm': 0.48414620757102966, 'learning_rate': 9.035820510710302e-06, 'epoch': 0.84}

 28%|██▊       | 3741/13317 [38:32<1:37:53,  1.63it/s]
 28%|██▊       | 3742/13317 [38:33<1:37:54,  1.63it/s]
                                                      
{'loss': 0.2053, 'grad_norm': 0.46334394812583923, 'learning_rate': 9.03504666854087e-06, 'epoch': 0.84}

 28%|██▊       | 3742/13317 [38:33<1:37:54,  1.63it/s]
 28%|██▊       | 3743/13317 [38:33<1:37:50,  1.63it/s]
                                                      
{'loss': 0.219, 'grad_norm': 0.5152288675308228, 'learning_rate': 9.034272549121007e-06, 'epoch': 0.84}

 28%|██▊       | 3743/13317 [38:33<1:37:50,  1.63it/s]
 28%|██▊       | 3744/13317 [38:34<1:37:48,  1.63it/s]
                                                      
{'loss': 0.2224, 'grad_norm': 0.45669984817504883, 'learning_rate': 9.033498152503905e-06, 'epoch': 0.84}

 28%|██▊       | 3744/13317 [38:34<1:37:48,  1.63it/s]
 28%|██▊       | 3745/13317 [38:35<1:40:24,  1.59it/s]
                                                      
{'loss': 0.2488, 'grad_norm': 0.5332451462745667, 'learning_rate': 9.032723478742773e-06, 'epoch': 0.84}

 28%|██▊       | 3745/13317 [38:35<1:40:24,  1.59it/s]
 28%|██▊       | 3746/13317 [38:35<1:39:36,  1.60it/s]
                                                      
{'loss': 0.2071, 'grad_norm': 0.5101264715194702, 'learning_rate': 9.03194852789084e-06, 'epoch': 0.84}

 28%|██▊       | 3746/13317 [38:35<1:39:36,  1.60it/s]
 28%|██▊       | 3747/13317 [38:36<1:39:07,  1.61it/s]
                                                      
{'loss': 0.2276, 'grad_norm': 0.5311393737792969, 'learning_rate': 9.03117330000135e-06, 'epoch': 0.84}

 28%|██▊       | 3747/13317 [38:36<1:39:07,  1.61it/s]
 28%|██▊       | 3748/13317 [38:36<1:38:43,  1.62it/s]
                                                      
{'loss': 0.315, 'grad_norm': 0.5586031675338745, 'learning_rate': 9.030397795127572e-06, 'epoch': 0.84}

 28%|██▊       | 3748/13317 [38:37<1:38:43,  1.62it/s]
 28%|██▊       | 3749/13317 [38:37<1:38:25,  1.62it/s]
                                                      
{'loss': 0.2098, 'grad_norm': 0.5027840733528137, 'learning_rate': 9.029622013322792e-06, 'epoch': 0.84}

 28%|██▊       | 3749/13317 [38:37<1:38:25,  1.62it/s]
 28%|██▊       | 3750/13317 [38:38<1:38:18,  1.62it/s]
                                                      
{'loss': 0.1696, 'grad_norm': 0.45469796657562256, 'learning_rate': 9.028845954640311e-06, 'epoch': 0.84}

 28%|██▊       | 3750/13317 [38:38<1:38:18,  1.62it/s]
 28%|██▊       | 3751/13317 [38:38<1:38:10,  1.62it/s]
                                                      
{'loss': 0.2606, 'grad_norm': 0.5320020914077759, 'learning_rate': 9.028069619133457e-06, 'epoch': 0.85}

 28%|██▊       | 3751/13317 [38:38<1:38:10,  1.62it/s]
 28%|██▊       | 3752/13317 [38:39<1:38:11,  1.62it/s]
                                                      
{'loss': 0.2135, 'grad_norm': 0.47719937562942505, 'learning_rate': 9.027293006855571e-06, 'epoch': 0.85}

 28%|██▊       | 3752/13317 [38:39<1:38:11,  1.62it/s]
 28%|██▊       | 3753/13317 [38:40<1:38:02,  1.63it/s]
                                                      
{'loss': 0.1882, 'grad_norm': 0.5216204524040222, 'learning_rate': 9.026516117860011e-06, 'epoch': 0.85}

 28%|██▊       | 3753/13317 [38:40<1:38:02,  1.63it/s]
 28%|██▊       | 3754/13317 [38:40<1:38:03,  1.63it/s]
                                                      
{'loss': 0.2734, 'grad_norm': 0.6294114589691162, 'learning_rate': 9.025738952200163e-06, 'epoch': 0.85}

 28%|██▊       | 3754/13317 [38:40<1:38:03,  1.63it/s]
 28%|██▊       | 3755/13317 [38:41<1:37:58,  1.63it/s]
                                                      
{'loss': 0.2079, 'grad_norm': 0.49671339988708496, 'learning_rate': 9.024961509929422e-06, 'epoch': 0.85}

 28%|██▊       | 3755/13317 [38:41<1:37:58,  1.63it/s]
 28%|██▊       | 3756/13317 [38:41<1:37:55,  1.63it/s]
                                                      
{'loss': 0.1795, 'grad_norm': 0.45797982811927795, 'learning_rate': 9.024183791101208e-06, 'epoch': 0.85}

 28%|██▊       | 3756/13317 [38:41<1:37:55,  1.63it/s]
 28%|██▊       | 3757/13317 [38:42<1:37:50,  1.63it/s]
                                                      
{'loss': 0.2477, 'grad_norm': 0.5546942353248596, 'learning_rate': 9.023405795768961e-06, 'epoch': 0.85}

 28%|██▊       | 3757/13317 [38:42<1:37:50,  1.63it/s]
 28%|██▊       | 3758/13317 [38:43<1:37:48,  1.63it/s]
                                                      
{'loss': 0.2652, 'grad_norm': 0.4955917298793793, 'learning_rate': 9.022627523986134e-06, 'epoch': 0.85}

 28%|██▊       | 3758/13317 [38:43<1:37:48,  1.63it/s]
 28%|██▊       | 3759/13317 [38:43<1:37:49,  1.63it/s]
                                                      
{'loss': 0.2893, 'grad_norm': 0.5319892168045044, 'learning_rate': 9.021848975806203e-06, 'epoch': 0.85}

 28%|██▊       | 3759/13317 [38:43<1:37:49,  1.63it/s]
 28%|██▊       | 3760/13317 [38:44<1:37:53,  1.63it/s]
                                                      
{'loss': 0.2016, 'grad_norm': 0.5155942440032959, 'learning_rate': 9.021070151282665e-06, 'epoch': 0.85}

 28%|██▊       | 3760/13317 [38:44<1:37:53,  1.63it/s]
 28%|██▊       | 3761/13317 [38:44<1:37:48,  1.63it/s]
                                                      
{'loss': 0.235, 'grad_norm': 0.5684719681739807, 'learning_rate': 9.02029105046903e-06, 'epoch': 0.85}

 28%|██▊       | 3761/13317 [38:45<1:37:48,  1.63it/s]
 28%|██▊       | 3762/13317 [38:45<1:37:45,  1.63it/s]
                                                      
{'loss': 0.244, 'grad_norm': 0.5879352688789368, 'learning_rate': 9.019511673418834e-06, 'epoch': 0.85}

 28%|██▊       | 3762/13317 [38:45<1:37:45,  1.63it/s]
 28%|██▊       | 3763/13317 [38:46<1:37:44,  1.63it/s]
                                                      
{'loss': 0.3048, 'grad_norm': 0.5107309818267822, 'learning_rate': 9.018732020185625e-06, 'epoch': 0.85}

 28%|██▊       | 3763/13317 [38:46<1:37:44,  1.63it/s]
 28%|██▊       | 3764/13317 [38:46<1:37:45,  1.63it/s]
                                                      
{'loss': 0.2545, 'grad_norm': 0.5025999546051025, 'learning_rate': 9.017952090822977e-06, 'epoch': 0.85}

 28%|██▊       | 3764/13317 [38:46<1:37:45,  1.63it/s]
 28%|██▊       | 3765/13317 [38:47<1:37:47,  1.63it/s]
                                                      
{'loss': 0.2608, 'grad_norm': 0.5614901185035706, 'learning_rate': 9.017171885384478e-06, 'epoch': 0.85}

 28%|██▊       | 3765/13317 [38:47<1:37:47,  1.63it/s]
 28%|██▊       | 3766/13317 [38:47<1:37:47,  1.63it/s]
                                                      
{'loss': 0.2584, 'grad_norm': 0.5112230181694031, 'learning_rate': 9.016391403923735e-06, 'epoch': 0.85}

 28%|██▊       | 3766/13317 [38:48<1:37:47,  1.63it/s]
 28%|██▊       | 3767/13317 [38:48<1:37:41,  1.63it/s]
                                                      
{'loss': 0.2459, 'grad_norm': 0.49876508116722107, 'learning_rate': 9.015610646494375e-06, 'epoch': 0.85}

 28%|██▊       | 3767/13317 [38:48<1:37:41,  1.63it/s]
 28%|██▊       | 3768/13317 [38:49<1:37:37,  1.63it/s]
                                                      
{'loss': 0.2105, 'grad_norm': 0.4751169979572296, 'learning_rate': 9.014829613150047e-06, 'epoch': 0.85}

 28%|██▊       | 3768/13317 [38:49<1:37:37,  1.63it/s]
 28%|██▊       | 3769/13317 [38:49<1:37:46,  1.63it/s]
                                                      
{'loss': 0.2618, 'grad_norm': 0.597651481628418, 'learning_rate': 9.014048303944415e-06, 'epoch': 0.85}

 28%|██▊       | 3769/13317 [38:49<1:37:46,  1.63it/s]
 28%|██▊       | 3770/13317 [38:50<1:37:47,  1.63it/s]
                                                      
{'loss': 0.2559, 'grad_norm': 0.581855833530426, 'learning_rate': 9.013266718931161e-06, 'epoch': 0.85}

 28%|██▊       | 3770/13317 [38:50<1:37:47,  1.63it/s]
 28%|██▊       | 3771/13317 [38:51<1:37:43,  1.63it/s]
                                                      
{'loss': 0.2158, 'grad_norm': 0.5161266922950745, 'learning_rate': 9.012484858163994e-06, 'epoch': 0.85}

 28%|██▊       | 3771/13317 [38:51<1:37:43,  1.63it/s]
 28%|██▊       | 3772/13317 [38:51<1:37:51,  1.63it/s]
                                                      
{'loss': 0.26, 'grad_norm': 0.5080208778381348, 'learning_rate': 9.01170272169663e-06, 'epoch': 0.85}

 28%|██▊       | 3772/13317 [38:51<1:37:51,  1.63it/s]
 28%|██▊       | 3773/13317 [38:52<1:37:47,  1.63it/s]
                                                      
{'loss': 0.2332, 'grad_norm': 0.4990759491920471, 'learning_rate': 9.010920309582812e-06, 'epoch': 0.85}

 28%|██▊       | 3773/13317 [38:52<1:37:47,  1.63it/s]
 28%|██▊       | 3774/13317 [38:52<1:37:55,  1.62it/s]
                                                      
{'loss': 0.2006, 'grad_norm': 0.4703560173511505, 'learning_rate': 9.010137621876303e-06, 'epoch': 0.85}

 28%|██▊       | 3774/13317 [38:53<1:37:55,  1.62it/s]
 28%|██▊       | 3775/13317 [38:53<1:37:52,  1.62it/s]
                                                      
{'loss': 0.3023, 'grad_norm': 0.56620854139328, 'learning_rate': 9.009354658630876e-06, 'epoch': 0.85}

 28%|██▊       | 3775/13317 [38:53<1:37:52,  1.62it/s]
 28%|██▊       | 3776/13317 [38:54<1:37:44,  1.63it/s]
                                                      
{'loss': 0.3278, 'grad_norm': 0.5675899386405945, 'learning_rate': 9.008571419900334e-06, 'epoch': 0.85}

 28%|██▊       | 3776/13317 [38:54<1:37:44,  1.63it/s]
 28%|██▊       | 3777/13317 [38:54<1:37:44,  1.63it/s]
                                                      
{'loss': 0.2273, 'grad_norm': 0.5066867470741272, 'learning_rate': 9.007787905738492e-06, 'epoch': 0.85}

 28%|██▊       | 3777/13317 [38:54<1:37:44,  1.63it/s]
 28%|██▊       | 3778/13317 [38:55<1:37:38,  1.63it/s]
                                                      
{'loss': 0.2266, 'grad_norm': 0.4510524570941925, 'learning_rate': 9.007004116199184e-06, 'epoch': 0.85}

 28%|██▊       | 3778/13317 [38:55<1:37:38,  1.63it/s]
 28%|██▊       | 3779/13317 [38:55<1:37:57,  1.62it/s]
                                                      
{'loss': 0.2122, 'grad_norm': 0.45192140340805054, 'learning_rate': 9.00622005133627e-06, 'epoch': 0.85}

 28%|██▊       | 3779/13317 [38:56<1:37:57,  1.62it/s]
 28%|██▊       | 3780/13317 [38:56<1:37:51,  1.62it/s]
                                                      
{'loss': 0.2541, 'grad_norm': 0.5519589185714722, 'learning_rate': 9.005435711203619e-06, 'epoch': 0.85}

 28%|██▊       | 3780/13317 [38:56<1:37:51,  1.62it/s]
 28%|██▊       | 3781/13317 [38:57<1:37:42,  1.63it/s]
                                                      
{'loss': 0.2273, 'grad_norm': 0.5730631351470947, 'learning_rate': 9.004651095855122e-06, 'epoch': 0.85}

 28%|██▊       | 3781/13317 [38:57<1:37:42,  1.63it/s]
 28%|██▊       | 3782/13317 [38:57<1:37:44,  1.63it/s]
                                                      
{'loss': 0.1939, 'grad_norm': 0.4623742699623108, 'learning_rate': 9.003866205344694e-06, 'epoch': 0.85}

 28%|██▊       | 3782/13317 [38:57<1:37:44,  1.63it/s]
 28%|██▊       | 3783/13317 [38:58<1:37:38,  1.63it/s]
                                                      
{'loss': 0.2151, 'grad_norm': 0.4594777524471283, 'learning_rate': 9.003081039726265e-06, 'epoch': 0.85}

 28%|██▊       | 3783/13317 [38:58<1:37:38,  1.63it/s]
 28%|██▊       | 3784/13317 [38:59<1:37:41,  1.63it/s]
                                                      
{'loss': 0.2524, 'grad_norm': 0.542252242565155, 'learning_rate': 9.00229559905378e-06, 'epoch': 0.85}

 28%|██▊       | 3784/13317 [38:59<1:37:41,  1.63it/s]
 28%|██▊       | 3785/13317 [38:59<1:37:37,  1.63it/s]
                                                      
{'loss': 0.2778, 'grad_norm': 0.6238046288490295, 'learning_rate': 9.001509883381214e-06, 'epoch': 0.85}

 28%|██▊       | 3785/13317 [38:59<1:37:37,  1.63it/s]
 28%|██▊       | 3786/13317 [39:00<1:37:36,  1.63it/s]
                                                      
{'loss': 0.2909, 'grad_norm': 0.5919274687767029, 'learning_rate': 9.00072389276255e-06, 'epoch': 0.85}

 28%|██▊       | 3786/13317 [39:00<1:37:36,  1.63it/s]
 28%|██▊       | 3787/13317 [39:00<1:37:42,  1.63it/s]
                                                      
{'loss': 0.2414, 'grad_norm': 0.5101300477981567, 'learning_rate': 8.999937627251793e-06, 'epoch': 0.85}

 28%|██▊       | 3787/13317 [39:01<1:37:42,  1.63it/s]
 28%|██▊       | 3788/13317 [39:01<1:37:38,  1.63it/s]
                                                      
{'loss': 0.2435, 'grad_norm': 0.5268042683601379, 'learning_rate': 8.99915108690297e-06, 'epoch': 0.85}

 28%|██▊       | 3788/13317 [39:01<1:37:38,  1.63it/s]
 28%|██▊       | 3789/13317 [39:02<1:37:34,  1.63it/s]
                                                      
{'loss': 0.2415, 'grad_norm': 0.530182421207428, 'learning_rate': 8.998364271770121e-06, 'epoch': 0.85}

 28%|██▊       | 3789/13317 [39:02<1:37:34,  1.63it/s]
 28%|██▊       | 3790/13317 [39:02<1:37:29,  1.63it/s]
                                                      
{'loss': 0.2849, 'grad_norm': 0.6220847368240356, 'learning_rate': 8.997577181907314e-06, 'epoch': 0.85}

 28%|██▊       | 3790/13317 [39:02<1:37:29,  1.63it/s]
 28%|██▊       | 3791/13317 [39:03<1:37:29,  1.63it/s]
                                                      
{'loss': 0.1872, 'grad_norm': 0.487877756357193, 'learning_rate': 8.996789817368627e-06, 'epoch': 0.85}

 28%|██▊       | 3791/13317 [39:03<1:37:29,  1.63it/s]
 28%|██▊       | 3792/13317 [39:03<1:37:34,  1.63it/s]
                                                      
{'loss': 0.2392, 'grad_norm': 0.5017850995063782, 'learning_rate': 8.99600217820816e-06, 'epoch': 0.85}

 28%|██▊       | 3792/13317 [39:04<1:37:34,  1.63it/s]
 28%|██▊       | 3793/13317 [39:04<1:37:30,  1.63it/s]
                                                      
{'loss': 0.2628, 'grad_norm': 0.5276907682418823, 'learning_rate': 8.995214264480034e-06, 'epoch': 0.85}

 28%|██▊       | 3793/13317 [39:04<1:37:30,  1.63it/s]
 28%|██▊       | 3794/13317 [39:05<1:37:28,  1.63it/s]
                                                      
{'loss': 0.1989, 'grad_norm': 0.4730173945426941, 'learning_rate': 8.994426076238386e-06, 'epoch': 0.85}

 28%|██▊       | 3794/13317 [39:05<1:37:28,  1.63it/s]
 28%|██▊       | 3795/13317 [39:05<1:37:28,  1.63it/s]
                                                      
{'loss': 0.2897, 'grad_norm': 0.5736653804779053, 'learning_rate': 8.993637613537373e-06, 'epoch': 0.85}

 28%|██▊       | 3795/13317 [39:05<1:37:28,  1.63it/s]
 29%|██▊       | 3796/13317 [39:06<1:37:24,  1.63it/s]
                                                      
{'loss': 0.313, 'grad_norm': 0.6154019832611084, 'learning_rate': 8.99284887643117e-06, 'epoch': 0.86}

 29%|██▊       | 3796/13317 [39:06<1:37:24,  1.63it/s]
 29%|██▊       | 3797/13317 [39:07<1:37:22,  1.63it/s]
                                                      
{'loss': 0.2257, 'grad_norm': 0.5241266489028931, 'learning_rate': 8.992059864973972e-06, 'epoch': 0.86}

 29%|██▊       | 3797/13317 [39:07<1:37:22,  1.63it/s]
 29%|██▊       | 3798/13317 [39:07<1:37:20,  1.63it/s]
                                                      
{'loss': 0.2551, 'grad_norm': 0.5116270780563354, 'learning_rate': 8.991270579219992e-06, 'epoch': 0.86}

 29%|██▊       | 3798/13317 [39:07<1:37:20,  1.63it/s]
 29%|██▊       | 3799/13317 [39:08<1:37:19,  1.63it/s]
                                                      
{'loss': 0.2301, 'grad_norm': 0.484589159488678, 'learning_rate': 8.990481019223465e-06, 'epoch': 0.86}

 29%|██▊       | 3799/13317 [39:08<1:37:19,  1.63it/s]
 29%|██▊       | 3800/13317 [39:08<1:37:20,  1.63it/s]
                                                      
{'loss': 0.2388, 'grad_norm': 0.4996248483657837, 'learning_rate': 8.989691185038638e-06, 'epoch': 0.86}

 29%|██▊       | 3800/13317 [39:09<1:37:20,  1.63it/s]
 29%|██▊       | 3801/13317 [39:09<1:37:19,  1.63it/s]
                                                      
{'loss': 0.2007, 'grad_norm': 0.5036982893943787, 'learning_rate': 8.988901076719786e-06, 'epoch': 0.86}

 29%|██▊       | 3801/13317 [39:09<1:37:19,  1.63it/s]
 29%|██▊       | 3802/13317 [39:10<1:37:19,  1.63it/s]
                                                      
{'loss': 0.2139, 'grad_norm': 0.4915751516819, 'learning_rate': 8.988110694321193e-06, 'epoch': 0.86}

 29%|██▊       | 3802/13317 [39:10<1:37:19,  1.63it/s]
 29%|██▊       | 3803/13317 [39:10<1:37:16,  1.63it/s]
                                                      
{'loss': 0.2662, 'grad_norm': 0.4840783476829529, 'learning_rate': 8.98732003789717e-06, 'epoch': 0.86}

 29%|██▊       | 3803/13317 [39:10<1:37:16,  1.63it/s]
 29%|██▊       | 3804/13317 [39:11<1:37:15,  1.63it/s]
                                                      
{'loss': 0.2225, 'grad_norm': 0.49125221371650696, 'learning_rate': 8.986529107502039e-06, 'epoch': 0.86}

 29%|██▊       | 3804/13317 [39:11<1:37:15,  1.63it/s]
 29%|██▊       | 3805/13317 [39:11<1:37:15,  1.63it/s]
                                                      
{'loss': 0.256, 'grad_norm': 0.5122411251068115, 'learning_rate': 8.98573790319015e-06, 'epoch': 0.86}

 29%|██▊       | 3805/13317 [39:12<1:37:15,  1.63it/s]
 29%|██▊       | 3806/13317 [39:12<1:37:14,  1.63it/s]
                                                      
{'loss': 0.2641, 'grad_norm': 0.551936149597168, 'learning_rate': 8.984946425015864e-06, 'epoch': 0.86}

 29%|██▊       | 3806/13317 [39:12<1:37:14,  1.63it/s]
 29%|██▊       | 3807/13317 [39:13<1:37:14,  1.63it/s]
                                                      
{'loss': 0.2471, 'grad_norm': 0.5287801027297974, 'learning_rate': 8.984154673033568e-06, 'epoch': 0.86}

 29%|██▊       | 3807/13317 [39:13<1:37:14,  1.63it/s]
 29%|██▊       | 3808/13317 [39:13<1:37:11,  1.63it/s]
                                                      
{'loss': 0.2572, 'grad_norm': 0.523393988609314, 'learning_rate': 8.98336264729766e-06, 'epoch': 0.86}

 29%|██▊       | 3808/13317 [39:13<1:37:11,  1.63it/s]
 29%|██▊       | 3809/13317 [39:14<1:37:08,  1.63it/s]
                                                      
{'loss': 0.2276, 'grad_norm': 0.48852285742759705, 'learning_rate': 8.982570347862561e-06, 'epoch': 0.86}

 29%|██▊       | 3809/13317 [39:14<1:37:08,  1.63it/s]
 29%|██▊       | 3810/13317 [39:15<1:37:11,  1.63it/s]
                                                      
{'loss': 0.2596, 'grad_norm': 0.5245879888534546, 'learning_rate': 8.98177777478271e-06, 'epoch': 0.86}

 29%|██▊       | 3810/13317 [39:15<1:37:11,  1.63it/s]
 29%|██▊       | 3811/13317 [39:15<1:37:12,  1.63it/s]
                                                      
{'loss': 0.3034, 'grad_norm': 0.5521055459976196, 'learning_rate': 8.980984928112568e-06, 'epoch': 0.86}

 29%|██▊       | 3811/13317 [39:15<1:37:12,  1.63it/s]
 29%|██▊       | 3812/13317 [39:16<1:37:13,  1.63it/s]
                                                      
{'loss': 0.2898, 'grad_norm': 0.5726510286331177, 'learning_rate': 8.980191807906608e-06, 'epoch': 0.86}

 29%|██▊       | 3812/13317 [39:16<1:37:13,  1.63it/s]
 29%|██▊       | 3813/13317 [39:16<1:37:11,  1.63it/s]
                                                      
{'loss': 0.2558, 'grad_norm': 0.5537142157554626, 'learning_rate': 8.97939841421933e-06, 'epoch': 0.86}

 29%|██▊       | 3813/13317 [39:16<1:37:11,  1.63it/s]
 29%|██▊       | 3814/13317 [39:17<1:37:13,  1.63it/s]
                                                      
{'loss': 0.1807, 'grad_norm': 0.3932594656944275, 'learning_rate': 8.978604747105247e-06, 'epoch': 0.86}

 29%|██▊       | 3814/13317 [39:17<1:37:13,  1.63it/s]
 29%|██▊       | 3815/13317 [39:18<1:37:13,  1.63it/s]
                                                      
{'loss': 0.2161, 'grad_norm': 0.45899197459220886, 'learning_rate': 8.977810806618889e-06, 'epoch': 0.86}

 29%|██▊       | 3815/13317 [39:18<1:37:13,  1.63it/s]
 29%|██▊       | 3816/13317 [39:18<1:37:16,  1.63it/s]
                                                      
{'loss': 0.194, 'grad_norm': 0.44541800022125244, 'learning_rate': 8.977016592814812e-06, 'epoch': 0.86}

 29%|██▊       | 3816/13317 [39:18<1:37:16,  1.63it/s]
 29%|██▊       | 3817/13317 [39:19<1:37:21,  1.63it/s]
                                                      
{'loss': 0.2186, 'grad_norm': 0.5081738233566284, 'learning_rate': 8.976222105747587e-06, 'epoch': 0.86}

 29%|██▊       | 3817/13317 [39:19<1:37:21,  1.63it/s]
 29%|██▊       | 3818/13317 [39:19<1:37:17,  1.63it/s]
                                                      
{'loss': 0.2537, 'grad_norm': 0.5056613087654114, 'learning_rate': 8.9754273454718e-06, 'epoch': 0.86}

 29%|██▊       | 3818/13317 [39:20<1:37:17,  1.63it/s]
 29%|██▊       | 3819/13317 [39:20<1:37:14,  1.63it/s]
                                                      
{'loss': 0.2509, 'grad_norm': 0.490288645029068, 'learning_rate': 8.974632312042064e-06, 'epoch': 0.86}

 29%|██▊       | 3819/13317 [39:20<1:37:14,  1.63it/s]
 29%|██▊       | 3820/13317 [39:21<1:37:15,  1.63it/s]
                                                      
{'loss': 0.2584, 'grad_norm': 0.4861788749694824, 'learning_rate': 8.973837005513004e-06, 'epoch': 0.86}

 29%|██▊       | 3820/13317 [39:21<1:37:15,  1.63it/s]
 29%|██▊       | 3821/13317 [39:21<1:37:09,  1.63it/s]
                                                      
{'loss': 0.2635, 'grad_norm': 0.5785199999809265, 'learning_rate': 8.973041425939265e-06, 'epoch': 0.86}

 29%|██▊       | 3821/13317 [39:21<1:37:09,  1.63it/s]
 29%|██▊       | 3822/13317 [39:22<1:37:21,  1.63it/s]
                                                      
{'loss': 0.2484, 'grad_norm': 0.48005053400993347, 'learning_rate': 8.972245573375513e-06, 'epoch': 0.86}

 29%|██▊       | 3822/13317 [39:22<1:37:21,  1.63it/s]
 29%|██▊       | 3823/13317 [39:23<1:37:12,  1.63it/s]
                                                      
{'loss': 0.2623, 'grad_norm': 0.5540967583656311, 'learning_rate': 8.97144944787643e-06, 'epoch': 0.86}

 29%|██▊       | 3823/13317 [39:23<1:37:12,  1.63it/s]
 29%|██▊       | 3824/13317 [39:23<1:37:08,  1.63it/s]
                                                      
{'loss': 0.2967, 'grad_norm': 0.49530845880508423, 'learning_rate': 8.970653049496723e-06, 'epoch': 0.86}

 29%|██▊       | 3824/13317 [39:23<1:37:08,  1.63it/s]
 29%|██▊       | 3825/13317 [39:24<1:37:16,  1.63it/s]
                                                      
{'loss': 0.2862, 'grad_norm': 0.5828845500946045, 'learning_rate': 8.969856378291106e-06, 'epoch': 0.86}

 29%|██▊       | 3825/13317 [39:24<1:37:16,  1.63it/s]
 29%|██▊       | 3826/13317 [39:24<1:37:12,  1.63it/s]
                                                      
{'loss': 0.2327, 'grad_norm': 0.7151734232902527, 'learning_rate': 8.969059434314324e-06, 'epoch': 0.86}

 29%|██▊       | 3826/13317 [39:24<1:37:12,  1.63it/s]
 29%|██▊       | 3827/13317 [39:25<1:37:17,  1.63it/s]
                                                      
{'loss': 0.2222, 'grad_norm': 0.4951602518558502, 'learning_rate': 8.968262217621134e-06, 'epoch': 0.86}

 29%|██▊       | 3827/13317 [39:25<1:37:17,  1.63it/s]
 29%|██▊       | 3828/13317 [39:26<1:37:11,  1.63it/s]
                                                      
{'loss': 0.2219, 'grad_norm': 0.45786428451538086, 'learning_rate': 8.967464728266312e-06, 'epoch': 0.86}

 29%|██▊       | 3828/13317 [39:26<1:37:11,  1.63it/s]
 29%|██▉       | 3829/13317 [39:26<1:37:09,  1.63it/s]
                                                      
{'loss': 0.2146, 'grad_norm': 0.5378241539001465, 'learning_rate': 8.966666966304656e-06, 'epoch': 0.86}

 29%|██▉       | 3829/13317 [39:26<1:37:09,  1.63it/s]
 29%|██▉       | 3830/13317 [39:27<1:37:12,  1.63it/s]
                                                      
{'loss': 0.2226, 'grad_norm': 0.514809250831604, 'learning_rate': 8.96586893179098e-06, 'epoch': 0.86}

 29%|██▉       | 3830/13317 [39:27<1:37:12,  1.63it/s]
 29%|██▉       | 3831/13317 [39:27<1:37:09,  1.63it/s]
                                                      
{'loss': 0.2228, 'grad_norm': 0.48639029264450073, 'learning_rate': 8.965070624780117e-06, 'epoch': 0.86}

 29%|██▉       | 3831/13317 [39:28<1:37:09,  1.63it/s]
 29%|██▉       | 3832/13317 [39:28<1:37:09,  1.63it/s]
                                                      
{'loss': 0.3124, 'grad_norm': 0.6533536911010742, 'learning_rate': 8.964272045326918e-06, 'epoch': 0.86}

 29%|██▉       | 3832/13317 [39:28<1:37:09,  1.63it/s]
 29%|██▉       | 3833/13317 [39:29<1:37:03,  1.63it/s]
                                                      
{'loss': 0.3086, 'grad_norm': 0.5463089346885681, 'learning_rate': 8.963473193486257e-06, 'epoch': 0.86}

 29%|██▉       | 3833/13317 [39:29<1:37:03,  1.63it/s]
 29%|██▉       | 3834/13317 [39:29<1:37:02,  1.63it/s]
                                                      
{'loss': 0.2835, 'grad_norm': 0.5694723129272461, 'learning_rate': 8.96267406931302e-06, 'epoch': 0.86}

 29%|██▉       | 3834/13317 [39:29<1:37:02,  1.63it/s]
 29%|██▉       | 3835/13317 [39:30<1:37:02,  1.63it/s]
                                                      
{'loss': 0.2404, 'grad_norm': 0.5430892705917358, 'learning_rate': 8.961874672862116e-06, 'epoch': 0.86}

 29%|██▉       | 3835/13317 [39:30<1:37:02,  1.63it/s]
 29%|██▉       | 3836/13317 [39:30<1:36:57,  1.63it/s]
                                                      
{'loss': 0.2236, 'grad_norm': 0.43991202116012573, 'learning_rate': 8.961075004188476e-06, 'epoch': 0.86}

 29%|██▉       | 3836/13317 [39:31<1:36:57,  1.63it/s]
 29%|██▉       | 3837/13317 [39:31<1:37:07,  1.63it/s]
                                                      
{'loss': 0.2362, 'grad_norm': 0.5500754714012146, 'learning_rate': 8.96027506334704e-06, 'epoch': 0.86}

 29%|██▉       | 3837/13317 [39:31<1:37:07,  1.63it/s]
 29%|██▉       | 3838/13317 [39:32<1:37:03,  1.63it/s]
                                                      
{'loss': 0.2543, 'grad_norm': 0.534909725189209, 'learning_rate': 8.959474850392776e-06, 'epoch': 0.86}

 29%|██▉       | 3838/13317 [39:32<1:37:03,  1.63it/s]
 29%|██▉       | 3839/13317 [39:32<1:36:58,  1.63it/s]
                                                      
{'loss': 0.2109, 'grad_norm': 0.5273274183273315, 'learning_rate': 8.958674365380665e-06, 'epoch': 0.86}

 29%|██▉       | 3839/13317 [39:32<1:36:58,  1.63it/s]
 29%|██▉       | 3840/13317 [39:33<1:36:56,  1.63it/s]
                                                      
{'loss': 0.2078, 'grad_norm': 0.5374495387077332, 'learning_rate': 8.957873608365713e-06, 'epoch': 0.87}

 29%|██▉       | 3840/13317 [39:33<1:36:56,  1.63it/s]
 29%|██▉       | 3841/13317 [39:34<1:36:55,  1.63it/s]
                                                      
{'loss': 0.2981, 'grad_norm': 0.5508188009262085, 'learning_rate': 8.957072579402936e-06, 'epoch': 0.87}

 29%|██▉       | 3841/13317 [39:34<1:36:55,  1.63it/s]
 29%|██▉       | 3842/13317 [39:34<1:36:53,  1.63it/s]
                                                      
{'loss': 0.1854, 'grad_norm': 0.4174865484237671, 'learning_rate': 8.956271278547374e-06, 'epoch': 0.87}

 29%|██▉       | 3842/13317 [39:34<1:36:53,  1.63it/s]
 29%|██▉       | 3843/13317 [39:35<1:36:55,  1.63it/s]
                                                      
{'loss': 0.1724, 'grad_norm': 0.4770667254924774, 'learning_rate': 8.955469705854086e-06, 'epoch': 0.87}

 29%|██▉       | 3843/13317 [39:35<1:36:55,  1.63it/s]
 29%|██▉       | 3844/13317 [39:35<1:36:54,  1.63it/s]
                                                      
{'loss': 0.24, 'grad_norm': 0.548923671245575, 'learning_rate': 8.954667861378149e-06, 'epoch': 0.87}

 29%|██▉       | 3844/13317 [39:36<1:36:54,  1.63it/s]
 29%|██▉       | 3845/13317 [39:36<1:36:57,  1.63it/s]
                                                      
{'loss': 0.2512, 'grad_norm': 0.5722547769546509, 'learning_rate': 8.953865745174655e-06, 'epoch': 0.87}

 29%|██▉       | 3845/13317 [39:36<1:36:57,  1.63it/s]
 29%|██▉       | 3846/13317 [39:37<1:36:54,  1.63it/s]
                                                      
{'loss': 0.3, 'grad_norm': 0.5889806747436523, 'learning_rate': 8.953063357298722e-06, 'epoch': 0.87}

 29%|██▉       | 3846/13317 [39:37<1:36:54,  1.63it/s]
 29%|██▉       | 3847/13317 [39:37<1:37:01,  1.63it/s]
                                                      
{'loss': 0.2212, 'grad_norm': 0.44120675325393677, 'learning_rate': 8.952260697805479e-06, 'epoch': 0.87}

 29%|██▉       | 3847/13317 [39:37<1:37:01,  1.63it/s]
 29%|██▉       | 3848/13317 [39:38<1:36:54,  1.63it/s]
                                                      
{'loss': 0.2688, 'grad_norm': 0.5304579138755798, 'learning_rate': 8.95145776675008e-06, 'epoch': 0.87}

 29%|██▉       | 3848/13317 [39:38<1:36:54,  1.63it/s]
 29%|██▉       | 3849/13317 [39:38<1:36:53,  1.63it/s]
                                                      
{'loss': 0.3268, 'grad_norm': 0.5186736583709717, 'learning_rate': 8.950654564187694e-06, 'epoch': 0.87}

 29%|██▉       | 3849/13317 [39:39<1:36:53,  1.63it/s]
 29%|██▉       | 3850/13317 [39:39<1:36:51,  1.63it/s]
                                                      
{'loss': 0.2684, 'grad_norm': 0.5610089302062988, 'learning_rate': 8.949851090173508e-06, 'epoch': 0.87}

 29%|██▉       | 3850/13317 [39:39<1:36:51,  1.63it/s]
 29%|██▉       | 3851/13317 [39:40<1:36:48,  1.63it/s]
                                                      
{'loss': 0.2024, 'grad_norm': 0.4369642436504364, 'learning_rate': 8.949047344762732e-06, 'epoch': 0.87}

 29%|██▉       | 3851/13317 [39:40<1:36:48,  1.63it/s]
 29%|██▉       | 3852/13317 [39:40<1:36:53,  1.63it/s]
                                                      
{'loss': 0.2202, 'grad_norm': 0.5981151461601257, 'learning_rate': 8.948243328010588e-06, 'epoch': 0.87}

 29%|██▉       | 3852/13317 [39:40<1:36:53,  1.63it/s]
 29%|██▉       | 3853/13317 [39:41<1:36:50,  1.63it/s]
                                                      
{'loss': 0.2333, 'grad_norm': 0.4765323996543884, 'learning_rate': 8.947439039972325e-06, 'epoch': 0.87}

 29%|██▉       | 3853/13317 [39:41<1:36:50,  1.63it/s]
 29%|██▉       | 3854/13317 [39:42<1:36:54,  1.63it/s]
                                                      
{'loss': 0.2206, 'grad_norm': 0.5506442785263062, 'learning_rate': 8.946634480703203e-06, 'epoch': 0.87}

 29%|██▉       | 3854/13317 [39:42<1:36:54,  1.63it/s]
 29%|██▉       | 3855/13317 [39:42<1:36:54,  1.63it/s]
                                                      
{'loss': 0.248, 'grad_norm': 0.48346105217933655, 'learning_rate': 8.945829650258504e-06, 'epoch': 0.87}

 29%|██▉       | 3855/13317 [39:42<1:36:54,  1.63it/s]
 29%|██▉       | 3856/13317 [39:43<1:36:50,  1.63it/s]
                                                      
{'loss': 0.2116, 'grad_norm': 0.46016833186149597, 'learning_rate': 8.945024548693529e-06, 'epoch': 0.87}

 29%|██▉       | 3856/13317 [39:43<1:36:50,  1.63it/s]
 29%|██▉       | 3857/13317 [39:43<1:36:53,  1.63it/s]
                                                      
{'loss': 0.2521, 'grad_norm': 0.47860246896743774, 'learning_rate': 8.944219176063598e-06, 'epoch': 0.87}

 29%|██▉       | 3857/13317 [39:44<1:36:53,  1.63it/s]
 29%|██▉       | 3858/13317 [39:44<1:36:50,  1.63it/s]
                                                      
{'loss': 0.2245, 'grad_norm': 0.4839986562728882, 'learning_rate': 8.943413532424047e-06, 'epoch': 0.87}

 29%|██▉       | 3858/13317 [39:44<1:36:50,  1.63it/s]
 29%|██▉       | 3859/13317 [39:45<1:36:51,  1.63it/s]
                                                      
{'loss': 0.2993, 'grad_norm': 0.5488479733467102, 'learning_rate': 8.942607617830232e-06, 'epoch': 0.87}

 29%|██▉       | 3859/13317 [39:45<1:36:51,  1.63it/s]
 29%|██▉       | 3860/13317 [39:45<1:36:49,  1.63it/s]
                                                      
{'loss': 0.2433, 'grad_norm': 0.5372480750083923, 'learning_rate': 8.94180143233753e-06, 'epoch': 0.87}

 29%|██▉       | 3860/13317 [39:45<1:36:49,  1.63it/s]
 29%|██▉       | 3861/13317 [39:46<1:36:48,  1.63it/s]
                                                      
{'loss': 0.274, 'grad_norm': 0.6051788330078125, 'learning_rate': 8.940994976001332e-06, 'epoch': 0.87}

 29%|██▉       | 3861/13317 [39:46<1:36:48,  1.63it/s]
 29%|██▉       | 3862/13317 [39:46<1:36:56,  1.63it/s]
                                                      
{'loss': 0.2829, 'grad_norm': 0.6002283692359924, 'learning_rate': 8.94018824887705e-06, 'epoch': 0.87}

 29%|██▉       | 3862/13317 [39:47<1:36:56,  1.63it/s]
 29%|██▉       | 3863/13317 [39:47<1:36:52,  1.63it/s]
                                                      
{'loss': 0.2905, 'grad_norm': 0.528643012046814, 'learning_rate': 8.939381251020119e-06, 'epoch': 0.87}

 29%|██▉       | 3863/13317 [39:47<1:36:52,  1.63it/s]
 29%|██▉       | 3864/13317 [39:48<1:36:51,  1.63it/s]
                                                      
{'loss': 0.2069, 'grad_norm': 0.5357279181480408, 'learning_rate': 8.938573982485983e-06, 'epoch': 0.87}

 29%|██▉       | 3864/13317 [39:48<1:36:51,  1.63it/s]
 29%|██▉       | 3865/13317 [39:48<1:36:52,  1.63it/s]
                                                      
{'loss': 0.3178, 'grad_norm': 0.6062705516815186, 'learning_rate': 8.937766443330114e-06, 'epoch': 0.87}

 29%|██▉       | 3865/13317 [39:48<1:36:52,  1.63it/s]
 29%|██▉       | 3866/13317 [39:49<1:36:45,  1.63it/s]
                                                      
{'loss': 0.2402, 'grad_norm': 0.6676257848739624, 'learning_rate': 8.936958633607995e-06, 'epoch': 0.87}

 29%|██▉       | 3866/13317 [39:49<1:36:45,  1.63it/s]
 29%|██▉       | 3867/13317 [39:50<1:36:51,  1.63it/s]
                                                      
{'loss': 0.2563, 'grad_norm': 0.5988389849662781, 'learning_rate': 8.936150553375133e-06, 'epoch': 0.87}

 29%|██▉       | 3867/13317 [39:50<1:36:51,  1.63it/s]
 29%|██▉       | 3868/13317 [39:50<1:36:46,  1.63it/s]
                                                      
{'loss': 0.2055, 'grad_norm': 0.5141773819923401, 'learning_rate': 8.935342202687052e-06, 'epoch': 0.87}

 29%|██▉       | 3868/13317 [39:50<1:36:46,  1.63it/s]
 29%|██▉       | 3869/13317 [39:51<1:36:44,  1.63it/s]
                                                      
{'loss': 0.1881, 'grad_norm': 0.43252596259117126, 'learning_rate': 8.934533581599292e-06, 'epoch': 0.87}

 29%|██▉       | 3869/13317 [39:51<1:36:44,  1.63it/s]
 29%|██▉       | 3870/13317 [39:51<1:36:46,  1.63it/s]
                                                      
{'loss': 0.2076, 'grad_norm': 0.44391223788261414, 'learning_rate': 8.933724690167417e-06, 'epoch': 0.87}

 29%|██▉       | 3870/13317 [39:52<1:36:46,  1.63it/s]
 29%|██▉       | 3871/13317 [39:52<1:36:41,  1.63it/s]
                                                      
{'loss': 0.2639, 'grad_norm': 0.5532019138336182, 'learning_rate': 8.932915528447003e-06, 'epoch': 0.87}

 29%|██▉       | 3871/13317 [39:52<1:36:41,  1.63it/s]
 29%|██▉       | 3872/13317 [39:53<1:36:44,  1.63it/s]
                                                      
{'loss': 0.2162, 'grad_norm': 0.4720420837402344, 'learning_rate': 8.932106096493653e-06, 'epoch': 0.87}

 29%|██▉       | 3872/13317 [39:53<1:36:44,  1.63it/s]
 29%|██▉       | 3873/13317 [39:53<1:36:41,  1.63it/s]
                                                      
{'loss': 0.2643, 'grad_norm': 0.5574502348899841, 'learning_rate': 8.931296394362978e-06, 'epoch': 0.87}

 29%|██▉       | 3873/13317 [39:53<1:36:41,  1.63it/s]
 29%|██▉       | 3874/13317 [39:54<1:36:42,  1.63it/s]
                                                      
{'loss': 0.2261, 'grad_norm': 0.4514000713825226, 'learning_rate': 8.930486422110616e-06, 'epoch': 0.87}

 29%|██▉       | 3874/13317 [39:54<1:36:42,  1.63it/s]
 29%|██▉       | 3875/13317 [39:54<1:36:40,  1.63it/s]
                                                      
{'loss': 0.258, 'grad_norm': 0.5020679831504822, 'learning_rate': 8.929676179792221e-06, 'epoch': 0.87}

 29%|██▉       | 3875/13317 [39:55<1:36:40,  1.63it/s]
 29%|██▉       | 3876/13317 [39:55<1:36:35,  1.63it/s]
                                                      
{'loss': 0.2686, 'grad_norm': 0.534700870513916, 'learning_rate': 8.928865667463463e-06, 'epoch': 0.87}

 29%|██▉       | 3876/13317 [39:55<1:36:35,  1.63it/s]
 29%|██▉       | 3877/13317 [39:56<1:36:46,  1.63it/s]
                                                      
{'loss': 0.1973, 'grad_norm': 0.4921933114528656, 'learning_rate': 8.928054885180036e-06, 'epoch': 0.87}

 29%|██▉       | 3877/13317 [39:56<1:36:46,  1.63it/s]
 29%|██▉       | 3878/13317 [39:56<1:36:40,  1.63it/s]
                                                      
{'loss': 0.2311, 'grad_norm': 0.5167208909988403, 'learning_rate': 8.927243832997646e-06, 'epoch': 0.87}

 29%|██▉       | 3878/13317 [39:56<1:36:40,  1.63it/s]
 29%|██▉       | 3879/13317 [39:57<1:36:34,  1.63it/s]
                                                      
{'loss': 0.2159, 'grad_norm': 0.47279107570648193, 'learning_rate': 8.926432510972021e-06, 'epoch': 0.87}

 29%|██▉       | 3879/13317 [39:57<1:36:34,  1.63it/s]
 29%|██▉       | 3880/13317 [39:58<1:36:31,  1.63it/s]
                                                      
{'loss': 0.2648, 'grad_norm': 0.5388237237930298, 'learning_rate': 8.92562091915891e-06, 'epoch': 0.87}

 29%|██▉       | 3880/13317 [39:58<1:36:31,  1.63it/s]
 29%|██▉       | 3881/13317 [39:58<1:36:28,  1.63it/s]
                                                      
{'loss': 0.2554, 'grad_norm': 0.5338147878646851, 'learning_rate': 8.924809057614078e-06, 'epoch': 0.87}

 29%|██▉       | 3881/13317 [39:58<1:36:28,  1.63it/s]
 29%|██▉       | 3882/13317 [39:59<1:36:33,  1.63it/s]
                                                      
{'loss': 0.2152, 'grad_norm': 0.4844064712524414, 'learning_rate': 8.923996926393306e-06, 'epoch': 0.87}

 29%|██▉       | 3882/13317 [39:59<1:36:33,  1.63it/s]
 29%|██▉       | 3883/13317 [39:59<1:36:28,  1.63it/s]
                                                      
{'loss': 0.2855, 'grad_norm': 0.48578596115112305, 'learning_rate': 8.923184525552398e-06, 'epoch': 0.87}

 29%|██▉       | 3883/13317 [39:59<1:36:28,  1.63it/s]
 29%|██▉       | 3884/13317 [40:00<1:36:32,  1.63it/s]
                                                      
{'loss': 0.2465, 'grad_norm': 0.4964396059513092, 'learning_rate': 8.922371855147172e-06, 'epoch': 0.87}

 29%|██▉       | 3884/13317 [40:00<1:36:32,  1.63it/s]
 29%|██▉       | 3885/13317 [40:01<1:36:35,  1.63it/s]
                                                      
{'loss': 0.279, 'grad_norm': 0.5451133251190186, 'learning_rate': 8.92155891523347e-06, 'epoch': 0.88}

 29%|██▉       | 3885/13317 [40:01<1:36:35,  1.63it/s]
 29%|██▉       | 3886/13317 [40:01<1:36:31,  1.63it/s]
                                                      
{'loss': 0.2758, 'grad_norm': 0.5745965838432312, 'learning_rate': 8.920745705867148e-06, 'epoch': 0.88}

 29%|██▉       | 3886/13317 [40:01<1:36:31,  1.63it/s]
 29%|██▉       | 3887/13317 [40:02<1:36:36,  1.63it/s]
                                                      
{'loss': 0.3063, 'grad_norm': 0.577737033367157, 'learning_rate': 8.919932227104083e-06, 'epoch': 0.88}

 29%|██▉       | 3887/13317 [40:02<1:36:36,  1.63it/s]
 29%|██▉       | 3888/13317 [40:02<1:36:33,  1.63it/s]
                                                      
{'loss': 0.2735, 'grad_norm': 0.5546910166740417, 'learning_rate': 8.919118479000169e-06, 'epoch': 0.88}

 29%|██▉       | 3888/13317 [40:03<1:36:33,  1.63it/s]
 29%|██▉       | 3889/13317 [40:03<1:37:00,  1.62it/s]
                                                      
{'loss': 0.1776, 'grad_norm': 0.4227936863899231, 'learning_rate': 8.91830446161132e-06, 'epoch': 0.88}

 29%|██▉       | 3889/13317 [40:03<1:37:00,  1.62it/s]
 29%|██▉       | 3890/13317 [40:04<1:36:55,  1.62it/s]
                                                      
{'loss': 0.2911, 'grad_norm': 0.5672867894172668, 'learning_rate': 8.917490174993466e-06, 'epoch': 0.88}

 29%|██▉       | 3890/13317 [40:04<1:36:55,  1.62it/s]
 29%|██▉       | 3891/13317 [40:04<1:36:45,  1.62it/s]
                                                      
{'loss': 0.225, 'grad_norm': 0.4641851782798767, 'learning_rate': 8.916675619202557e-06, 'epoch': 0.88}

 29%|██▉       | 3891/13317 [40:04<1:36:45,  1.62it/s]
 29%|██▉       | 3892/13317 [40:05<1:37:16,  1.61it/s]
                                                      
{'loss': 0.2546, 'grad_norm': 0.5420972108840942, 'learning_rate': 8.915860794294565e-06, 'epoch': 0.88}

 29%|██▉       | 3892/13317 [40:05<1:37:16,  1.61it/s]
 29%|██▉       | 3893/13317 [40:06<1:37:00,  1.62it/s]
                                                      
{'loss': 0.2924, 'grad_norm': 0.5265830159187317, 'learning_rate': 8.91504570032547e-06, 'epoch': 0.88}

 29%|██▉       | 3893/13317 [40:06<1:37:00,  1.62it/s]
 29%|██▉       | 3894/13317 [40:06<1:36:48,  1.62it/s]
                                                      
{'loss': 0.2692, 'grad_norm': 0.5749432444572449, 'learning_rate': 8.914230337351287e-06, 'epoch': 0.88}

 29%|██▉       | 3894/13317 [40:06<1:36:48,  1.62it/s]
 29%|██▉       | 3895/13317 [40:07<1:36:45,  1.62it/s]
                                                      
{'loss': 0.2583, 'grad_norm': 0.6035924553871155, 'learning_rate': 8.913414705428034e-06, 'epoch': 0.88}

 29%|██▉       | 3895/13317 [40:07<1:36:45,  1.62it/s]
 29%|██▉       | 3896/13317 [40:07<1:36:37,  1.62it/s]
                                                      
{'loss': 0.2231, 'grad_norm': 0.5204378962516785, 'learning_rate': 8.912598804611755e-06, 'epoch': 0.88}

 29%|██▉       | 3896/13317 [40:08<1:36:37,  1.62it/s]
 29%|██▉       | 3897/13317 [40:08<1:36:37,  1.62it/s]
                                                      
{'loss': 0.1834, 'grad_norm': 0.4399634897708893, 'learning_rate': 8.911782634958509e-06, 'epoch': 0.88}

 29%|██▉       | 3897/13317 [40:08<1:36:37,  1.62it/s]
 29%|██▉       | 3898/13317 [40:09<1:36:31,  1.63it/s]
                                                      
{'loss': 0.202, 'grad_norm': 0.49772581458091736, 'learning_rate': 8.910966196524378e-06, 'epoch': 0.88}

 29%|██▉       | 3898/13317 [40:09<1:36:31,  1.63it/s]
 29%|██▉       | 3899/13317 [40:09<1:36:31,  1.63it/s]
                                                      
{'loss': 0.255, 'grad_norm': 0.5930992364883423, 'learning_rate': 8.91014948936546e-06, 'epoch': 0.88}

 29%|██▉       | 3899/13317 [40:09<1:36:31,  1.63it/s]
 29%|██▉       | 3900/13317 [40:10<1:36:32,  1.63it/s]
                                                      
{'loss': 0.2126, 'grad_norm': 0.534127950668335, 'learning_rate': 8.909332513537872e-06, 'epoch': 0.88}

 29%|██▉       | 3900/13317 [40:10<1:36:32,  1.63it/s]
 29%|██▉       | 3901/13317 [40:10<1:36:28,  1.63it/s]
                                                      
{'loss': 0.3309, 'grad_norm': 0.5794639587402344, 'learning_rate': 8.908515269097746e-06, 'epoch': 0.88}

 29%|██▉       | 3901/13317 [40:11<1:36:28,  1.63it/s]
 29%|██▉       | 3902/13317 [40:11<1:36:31,  1.63it/s]
                                                      
{'loss': 0.2437, 'grad_norm': 0.4924762547016144, 'learning_rate': 8.907697756101238e-06, 'epoch': 0.88}

 29%|██▉       | 3902/13317 [40:11<1:36:31,  1.63it/s]
 29%|██▉       | 3903/13317 [40:12<1:36:28,  1.63it/s]
                                                      
{'loss': 0.2757, 'grad_norm': 0.6507858633995056, 'learning_rate': 8.906879974604517e-06, 'epoch': 0.88}

 29%|██▉       | 3903/13317 [40:12<1:36:28,  1.63it/s]
 29%|██▉       | 3904/13317 [40:12<1:36:23,  1.63it/s]
                                                      
{'loss': 0.2472, 'grad_norm': 0.5882624387741089, 'learning_rate': 8.906061924663775e-06, 'epoch': 0.88}

 29%|██▉       | 3904/13317 [40:12<1:36:23,  1.63it/s]
 29%|██▉       | 3905/13317 [40:13<1:36:23,  1.63it/s]
                                                      
{'loss': 0.292, 'grad_norm': 0.820288896560669, 'learning_rate': 8.905243606335223e-06, 'epoch': 0.88}

 29%|██▉       | 3905/13317 [40:13<1:36:23,  1.63it/s]
 29%|██▉       | 3906/13317 [40:14<1:36:16,  1.63it/s]
                                                      
{'loss': 0.2388, 'grad_norm': 0.5692503452301025, 'learning_rate': 8.904425019675085e-06, 'epoch': 0.88}

 29%|██▉       | 3906/13317 [40:14<1:36:16,  1.63it/s]
 29%|██▉       | 3907/13317 [40:14<1:36:19,  1.63it/s]
                                                      
{'loss': 0.2067, 'grad_norm': 0.43211182951927185, 'learning_rate': 8.903606164739606e-06, 'epoch': 0.88}

 29%|██▉       | 3907/13317 [40:14<1:36:19,  1.63it/s]
 29%|██▉       | 3908/13317 [40:15<1:36:17,  1.63it/s]
                                                      
{'loss': 0.2879, 'grad_norm': 0.5721641182899475, 'learning_rate': 8.902787041585054e-06, 'epoch': 0.88}

 29%|██▉       | 3908/13317 [40:15<1:36:17,  1.63it/s]
 29%|██▉       | 3909/13317 [40:15<1:36:13,  1.63it/s]
                                                      
{'loss': 0.3543, 'grad_norm': 0.508274257183075, 'learning_rate': 8.901967650267707e-06, 'epoch': 0.88}

 29%|██▉       | 3909/13317 [40:15<1:36:13,  1.63it/s]
 29%|██▉       | 3910/13317 [40:16<1:36:19,  1.63it/s]
                                                      
{'loss': 0.2708, 'grad_norm': 0.5368648767471313, 'learning_rate': 8.901147990843868e-06, 'epoch': 0.88}

 29%|██▉       | 3910/13317 [40:16<1:36:19,  1.63it/s]
 29%|██▉       | 3911/13317 [40:17<1:36:14,  1.63it/s]
                                                      
{'loss': 0.2031, 'grad_norm': 0.47742700576782227, 'learning_rate': 8.900328063369857e-06, 'epoch': 0.88}

 29%|██▉       | 3911/13317 [40:17<1:36:14,  1.63it/s]
 29%|██▉       | 3912/13317 [40:17<1:36:09,  1.63it/s]
                                                      
{'loss': 0.3047, 'grad_norm': 0.5214318633079529, 'learning_rate': 8.899507867902009e-06, 'epoch': 0.88}

 29%|██▉       | 3912/13317 [40:17<1:36:09,  1.63it/s]
 29%|██▉       | 3913/13317 [40:18<1:36:08,  1.63it/s]
                                                      
{'loss': 0.2935, 'grad_norm': 0.5991809964179993, 'learning_rate': 8.898687404496684e-06, 'epoch': 0.88}

 29%|██▉       | 3913/13317 [40:18<1:36:08,  1.63it/s]
 29%|██▉       | 3914/13317 [40:18<1:36:11,  1.63it/s]
                                                      
{'loss': 0.2588, 'grad_norm': 0.538353443145752, 'learning_rate': 8.897866673210252e-06, 'epoch': 0.88}

 29%|██▉       | 3914/13317 [40:19<1:36:11,  1.63it/s]
 29%|██▉       | 3915/13317 [40:19<1:36:13,  1.63it/s]
                                                      
{'loss': 0.2843, 'grad_norm': 0.678791880607605, 'learning_rate': 8.897045674099109e-06, 'epoch': 0.88}

 29%|██▉       | 3915/13317 [40:19<1:36:13,  1.63it/s]
 29%|██▉       | 3916/13317 [40:20<1:36:11,  1.63it/s]
                                                      
{'loss': 0.2309, 'grad_norm': 0.538266122341156, 'learning_rate': 8.896224407219667e-06, 'epoch': 0.88}

 29%|██▉       | 3916/13317 [40:20<1:36:11,  1.63it/s]
 29%|██▉       | 3917/13317 [40:20<1:36:09,  1.63it/s]
                                                      
{'loss': 0.2758, 'grad_norm': 0.49902358651161194, 'learning_rate': 8.895402872628352e-06, 'epoch': 0.88}

 29%|██▉       | 3917/13317 [40:20<1:36:09,  1.63it/s]
 29%|██▉       | 3918/13317 [40:21<1:36:08,  1.63it/s]
                                                      
{'loss': 0.2171, 'grad_norm': 0.5062902569770813, 'learning_rate': 8.894581070381617e-06, 'epoch': 0.88}

 29%|██▉       | 3918/13317 [40:21<1:36:08,  1.63it/s]
 29%|██▉       | 3919/13317 [40:21<1:36:05,  1.63it/s]
                                                      
{'loss': 0.2228, 'grad_norm': 0.5913341045379639, 'learning_rate': 8.893759000535923e-06, 'epoch': 0.88}

 29%|██▉       | 3919/13317 [40:22<1:36:05,  1.63it/s]
 29%|██▉       | 3920/13317 [40:22<1:36:13,  1.63it/s]
                                                      
{'loss': 0.2933, 'grad_norm': 0.6568008065223694, 'learning_rate': 8.892936663147762e-06, 'epoch': 0.88}

 29%|██▉       | 3920/13317 [40:22<1:36:13,  1.63it/s]
 29%|██▉       | 3921/13317 [40:23<1:36:11,  1.63it/s]
                                                      
{'loss': 0.3872, 'grad_norm': 0.6556405425071716, 'learning_rate': 8.89211405827363e-06, 'epoch': 0.88}

 29%|██▉       | 3921/13317 [40:23<1:36:11,  1.63it/s]
 29%|██▉       | 3922/13317 [40:23<1:36:08,  1.63it/s]
                                                      
{'loss': 0.1732, 'grad_norm': 0.4216349720954895, 'learning_rate': 8.891291185970052e-06, 'epoch': 0.88}

 29%|██▉       | 3922/13317 [40:23<1:36:08,  1.63it/s]
 29%|██▉       | 3923/13317 [40:24<1:36:06,  1.63it/s]
                                                      
{'loss': 0.2225, 'grad_norm': 0.4565683603286743, 'learning_rate': 8.890468046293569e-06, 'epoch': 0.88}

 29%|██▉       | 3923/13317 [40:24<1:36:06,  1.63it/s]
 29%|██▉       | 3924/13317 [40:25<1:36:05,  1.63it/s]
                                                      
{'loss': 0.2658, 'grad_norm': 0.5702142119407654, 'learning_rate': 8.889644639300738e-06, 'epoch': 0.88}

 29%|██▉       | 3924/13317 [40:25<1:36:05,  1.63it/s]
 29%|██▉       | 3925/13317 [40:25<1:36:07,  1.63it/s]
                                                      
{'loss': 0.2701, 'grad_norm': 0.4700146019458771, 'learning_rate': 8.888820965048135e-06, 'epoch': 0.88}

 29%|██▉       | 3925/13317 [40:25<1:36:07,  1.63it/s]
 29%|██▉       | 3926/13317 [40:26<1:36:05,  1.63it/s]
                                                      
{'loss': 0.319, 'grad_norm': 0.5180373787879944, 'learning_rate': 8.887997023592358e-06, 'epoch': 0.88}

 29%|██▉       | 3926/13317 [40:26<1:36:05,  1.63it/s]
 29%|██▉       | 3927/13317 [40:26<1:36:04,  1.63it/s]
                                                      
{'loss': 0.2516, 'grad_norm': 0.5107998847961426, 'learning_rate': 8.887172814990016e-06, 'epoch': 0.88}

 29%|██▉       | 3927/13317 [40:27<1:36:04,  1.63it/s]
 29%|██▉       | 3928/13317 [40:27<1:35:59,  1.63it/s]
                                                      
{'loss': 0.3256, 'grad_norm': 0.5457088351249695, 'learning_rate': 8.886348339297747e-06, 'epoch': 0.88}

 29%|██▉       | 3928/13317 [40:27<1:35:59,  1.63it/s]
 30%|██▉       | 3929/13317 [40:28<1:36:00,  1.63it/s]
                                                      
{'loss': 0.2503, 'grad_norm': 0.5051091909408569, 'learning_rate': 8.885523596572196e-06, 'epoch': 0.89}

 30%|██▉       | 3929/13317 [40:28<1:36:00,  1.63it/s]
 30%|██▉       | 3930/13317 [40:28<1:36:05,  1.63it/s]
                                                      
{'loss': 0.2318, 'grad_norm': 0.5788974761962891, 'learning_rate': 8.884698586870033e-06, 'epoch': 0.89}

 30%|██▉       | 3930/13317 [40:28<1:36:05,  1.63it/s]
 30%|██▉       | 3931/13317 [40:29<1:36:01,  1.63it/s]
                                                      
{'loss': 0.2292, 'grad_norm': 0.46733254194259644, 'learning_rate': 8.883873310247946e-06, 'epoch': 0.89}

 30%|██▉       | 3931/13317 [40:29<1:36:01,  1.63it/s]
 30%|██▉       | 3932/13317 [40:29<1:36:03,  1.63it/s]
                                                      
{'loss': 0.2235, 'grad_norm': 0.5135341882705688, 'learning_rate': 8.88304776676264e-06, 'epoch': 0.89}

 30%|██▉       | 3932/13317 [40:30<1:36:03,  1.63it/s]
 30%|██▉       | 3933/13317 [40:30<1:36:01,  1.63it/s]
                                                      
{'loss': 0.2507, 'grad_norm': 0.5372889041900635, 'learning_rate': 8.882221956470838e-06, 'epoch': 0.89}

 30%|██▉       | 3933/13317 [40:30<1:36:01,  1.63it/s]
 30%|██▉       | 3934/13317 [40:31<1:35:56,  1.63it/s]
                                                      
{'loss': 0.2593, 'grad_norm': 0.5622549057006836, 'learning_rate': 8.88139587942928e-06, 'epoch': 0.89}

 30%|██▉       | 3934/13317 [40:31<1:35:56,  1.63it/s]
 30%|██▉       | 3935/13317 [40:31<1:35:57,  1.63it/s]
                                                      
{'loss': 0.2227, 'grad_norm': 0.48431867361068726, 'learning_rate': 8.88056953569473e-06, 'epoch': 0.89}

 30%|██▉       | 3935/13317 [40:31<1:35:57,  1.63it/s]
 30%|██▉       | 3936/13317 [40:32<1:35:59,  1.63it/s]
                                                      
{'loss': 0.2311, 'grad_norm': 0.5020008087158203, 'learning_rate': 8.879742925323964e-06, 'epoch': 0.89}

 30%|██▉       | 3936/13317 [40:32<1:35:59,  1.63it/s]
 30%|██▉       | 3937/13317 [40:33<1:35:57,  1.63it/s]
                                                      
{'loss': 0.2138, 'grad_norm': 0.47178226709365845, 'learning_rate': 8.87891604837378e-06, 'epoch': 0.89}

 30%|██▉       | 3937/13317 [40:33<1:35:57,  1.63it/s]
 30%|██▉       | 3938/13317 [40:33<1:35:55,  1.63it/s]
                                                      
{'loss': 0.2649, 'grad_norm': 0.5596247315406799, 'learning_rate': 8.878088904900992e-06, 'epoch': 0.89}

 30%|██▉       | 3938/13317 [40:33<1:35:55,  1.63it/s]
 30%|██▉       | 3939/13317 [40:34<1:35:54,  1.63it/s]
                                                      
{'loss': 0.2923, 'grad_norm': 0.563075602054596, 'learning_rate': 8.877261494962434e-06, 'epoch': 0.89}

 30%|██▉       | 3939/13317 [40:34<1:35:54,  1.63it/s]
 30%|██▉       | 3940/13317 [40:34<1:35:56,  1.63it/s]
                                                      
{'loss': 0.2341, 'grad_norm': 0.5340157747268677, 'learning_rate': 8.87643381861496e-06, 'epoch': 0.89}

 30%|██▉       | 3940/13317 [40:35<1:35:56,  1.63it/s]
 30%|██▉       | 3941/13317 [40:35<1:35:53,  1.63it/s]
                                                      
{'loss': 0.2729, 'grad_norm': 0.5870695114135742, 'learning_rate': 8.875605875915438e-06, 'epoch': 0.89}

 30%|██▉       | 3941/13317 [40:35<1:35:53,  1.63it/s]
 30%|██▉       | 3942/13317 [40:36<1:35:52,  1.63it/s]
                                                      
{'loss': 0.2313, 'grad_norm': 0.4588182270526886, 'learning_rate': 8.874777666920756e-06, 'epoch': 0.89}

 30%|██▉       | 3942/13317 [40:36<1:35:52,  1.63it/s]
 30%|██▉       | 3943/13317 [40:36<1:35:51,  1.63it/s]
                                                      
{'loss': 0.2611, 'grad_norm': 0.5445171594619751, 'learning_rate': 8.873949191687822e-06, 'epoch': 0.89}

 30%|██▉       | 3943/13317 [40:36<1:35:51,  1.63it/s]
 30%|██▉       | 3944/13317 [40:37<1:35:51,  1.63it/s]
                                                      
{'loss': 0.2533, 'grad_norm': 0.5297250151634216, 'learning_rate': 8.87312045027356e-06, 'epoch': 0.89}

 30%|██▉       | 3944/13317 [40:37<1:35:51,  1.63it/s]
 30%|██▉       | 3945/13317 [40:37<1:35:53,  1.63it/s]
                                                      
{'loss': 0.2294, 'grad_norm': 0.4912565052509308, 'learning_rate': 8.872291442734914e-06, 'epoch': 0.89}

 30%|██▉       | 3945/13317 [40:38<1:35:53,  1.63it/s]
 30%|██▉       | 3946/13317 [40:38<1:35:53,  1.63it/s]
                                                      
{'loss': 0.3126, 'grad_norm': 0.6201950907707214, 'learning_rate': 8.871462169128846e-06, 'epoch': 0.89}

 30%|██▉       | 3946/13317 [40:38<1:35:53,  1.63it/s]
 30%|██▉       | 3947/13317 [40:39<1:35:53,  1.63it/s]
                                                      
{'loss': 0.2119, 'grad_norm': 0.6348736882209778, 'learning_rate': 8.870632629512335e-06, 'epoch': 0.89}

 30%|██▉       | 3947/13317 [40:39<1:35:53,  1.63it/s]
 30%|██▉       | 3948/13317 [40:39<1:35:50,  1.63it/s]
                                                      
{'loss': 0.2444, 'grad_norm': 0.555500864982605, 'learning_rate': 8.86980282394238e-06, 'epoch': 0.89}

 30%|██▉       | 3948/13317 [40:39<1:35:50,  1.63it/s]
 30%|██▉       | 3949/13317 [40:40<1:35:49,  1.63it/s]
                                                      
{'loss': 0.255, 'grad_norm': 0.5122093558311462, 'learning_rate': 8.868972752475996e-06, 'epoch': 0.89}

 30%|██▉       | 3949/13317 [40:40<1:35:49,  1.63it/s]
 30%|██▉       | 3950/13317 [40:41<1:35:52,  1.63it/s]
                                                      
{'loss': 0.2569, 'grad_norm': 0.5597578883171082, 'learning_rate': 8.868142415170218e-06, 'epoch': 0.89}

 30%|██▉       | 3950/13317 [40:41<1:35:52,  1.63it/s]
 30%|██▉       | 3951/13317 [40:41<1:35:49,  1.63it/s]
                                                      
{'loss': 0.2387, 'grad_norm': 0.5102441310882568, 'learning_rate': 8.8673118120821e-06, 'epoch': 0.89}

 30%|██▉       | 3951/13317 [40:41<1:35:49,  1.63it/s]
 30%|██▉       | 3952/13317 [40:42<1:35:47,  1.63it/s]
                                                      
{'loss': 0.3247, 'grad_norm': 0.5513226985931396, 'learning_rate': 8.866480943268713e-06, 'epoch': 0.89}

 30%|██▉       | 3952/13317 [40:42<1:35:47,  1.63it/s]
 30%|██▉       | 3953/13317 [40:42<1:35:47,  1.63it/s]
                                                      
{'loss': 0.2654, 'grad_norm': 0.5221174955368042, 'learning_rate': 8.865649808787146e-06, 'epoch': 0.89}

 30%|██▉       | 3953/13317 [40:42<1:35:47,  1.63it/s]
 30%|██▉       | 3954/13317 [40:43<1:35:45,  1.63it/s]
                                                      
{'loss': 0.2886, 'grad_norm': 0.5468663573265076, 'learning_rate': 8.864818408694506e-06, 'epoch': 0.89}

 30%|██▉       | 3954/13317 [40:43<1:35:45,  1.63it/s]
 30%|██▉       | 3955/13317 [40:44<1:35:51,  1.63it/s]
                                                      
{'loss': 0.2627, 'grad_norm': 0.5353173613548279, 'learning_rate': 8.863986743047922e-06, 'epoch': 0.89}

 30%|██▉       | 3955/13317 [40:44<1:35:51,  1.63it/s]
 30%|██▉       | 3956/13317 [40:44<1:35:46,  1.63it/s]
                                                      
{'loss': 0.2469, 'grad_norm': 0.5211354494094849, 'learning_rate': 8.863154811904535e-06, 'epoch': 0.89}

 30%|██▉       | 3956/13317 [40:44<1:35:46,  1.63it/s]
 30%|██▉       | 3957/13317 [40:45<1:35:45,  1.63it/s]
                                                      
{'loss': 0.2322, 'grad_norm': 0.5042508244514465, 'learning_rate': 8.862322615321511e-06, 'epoch': 0.89}

 30%|██▉       | 3957/13317 [40:45<1:35:45,  1.63it/s]
 30%|██▉       | 3958/13317 [40:45<1:35:44,  1.63it/s]
                                                      
{'loss': 0.288, 'grad_norm': 0.522577166557312, 'learning_rate': 8.861490153356026e-06, 'epoch': 0.89}

 30%|██▉       | 3958/13317 [40:46<1:35:44,  1.63it/s]
 30%|██▉       | 3959/13317 [40:46<1:35:41,  1.63it/s]
                                                      
{'loss': 0.2407, 'grad_norm': 0.5045750141143799, 'learning_rate': 8.86065742606528e-06, 'epoch': 0.89}

 30%|██▉       | 3959/13317 [40:46<1:35:41,  1.63it/s]
 30%|██▉       | 3960/13317 [40:47<1:35:50,  1.63it/s]
                                                      
{'loss': 0.3053, 'grad_norm': 0.5057578682899475, 'learning_rate': 8.859824433506494e-06, 'epoch': 0.89}

 30%|██▉       | 3960/13317 [40:47<1:35:50,  1.63it/s]
 30%|██▉       | 3961/13317 [40:47<1:35:46,  1.63it/s]
                                                      
{'loss': 0.2479, 'grad_norm': 0.5016666650772095, 'learning_rate': 8.8589911757369e-06, 'epoch': 0.89}

 30%|██▉       | 3961/13317 [40:47<1:35:46,  1.63it/s]
 30%|██▉       | 3962/13317 [40:48<1:35:43,  1.63it/s]
                                                      
{'loss': 0.2142, 'grad_norm': 0.4557982087135315, 'learning_rate': 8.858157652813752e-06, 'epoch': 0.89}

 30%|██▉       | 3962/13317 [40:48<1:35:43,  1.63it/s]
 30%|██▉       | 3963/13317 [40:49<1:35:40,  1.63it/s]
                                                      
{'loss': 0.2958, 'grad_norm': 0.6379660367965698, 'learning_rate': 8.857323864794326e-06, 'epoch': 0.89}

 30%|██▉       | 3963/13317 [40:49<1:35:40,  1.63it/s]
 30%|██▉       | 3964/13317 [40:49<1:35:40,  1.63it/s]
                                                      
{'loss': 0.2157, 'grad_norm': 0.4777471423149109, 'learning_rate': 8.856489811735904e-06, 'epoch': 0.89}

 30%|██▉       | 3964/13317 [40:49<1:35:40,  1.63it/s]
 30%|██▉       | 3965/13317 [40:50<1:35:48,  1.63it/s]
                                                      
{'loss': 0.2159, 'grad_norm': 0.5474370718002319, 'learning_rate': 8.8556554936958e-06, 'epoch': 0.89}

 30%|██▉       | 3965/13317 [40:50<1:35:48,  1.63it/s]
 30%|██▉       | 3966/13317 [40:50<1:35:45,  1.63it/s]
                                                      
{'loss': 0.3133, 'grad_norm': 0.5763237476348877, 'learning_rate': 8.854820910731341e-06, 'epoch': 0.89}

 30%|██▉       | 3966/13317 [40:50<1:35:45,  1.63it/s]
 30%|██▉       | 3967/13317 [40:51<1:35:44,  1.63it/s]
                                                      
{'loss': 0.2637, 'grad_norm': 0.5141853094100952, 'learning_rate': 8.853986062899869e-06, 'epoch': 0.89}

 30%|██▉       | 3967/13317 [40:51<1:35:44,  1.63it/s]
 30%|██▉       | 3968/13317 [40:52<1:35:41,  1.63it/s]
                                                      
{'loss': 0.2574, 'grad_norm': 0.5569793581962585, 'learning_rate': 8.853150950258747e-06, 'epoch': 0.89}

 30%|██▉       | 3968/13317 [40:52<1:35:41,  1.63it/s]
 30%|██▉       | 3969/13317 [40:52<1:35:37,  1.63it/s]
                                                      
{'loss': 0.1545, 'grad_norm': 0.4449329674243927, 'learning_rate': 8.852315572865358e-06, 'epoch': 0.89}

 30%|██▉       | 3969/13317 [40:52<1:35:37,  1.63it/s]
 30%|██▉       | 3970/13317 [40:53<1:35:41,  1.63it/s]
                                                      
{'loss': 0.3957, 'grad_norm': 0.5967901349067688, 'learning_rate': 8.851479930777098e-06, 'epoch': 0.89}

 30%|██▉       | 3970/13317 [40:53<1:35:41,  1.63it/s]
 30%|██▉       | 3971/13317 [40:53<1:35:35,  1.63it/s]
                                                      
{'loss': 0.2693, 'grad_norm': 0.5539321899414062, 'learning_rate': 8.850644024051389e-06, 'epoch': 0.89}

 30%|██▉       | 3971/13317 [40:54<1:35:35,  1.63it/s]
 30%|██▉       | 3972/13317 [40:54<1:35:34,  1.63it/s]
                                                      
{'loss': 0.2897, 'grad_norm': 0.4730742275714874, 'learning_rate': 8.849807852745661e-06, 'epoch': 0.89}

 30%|██▉       | 3972/13317 [40:54<1:35:34,  1.63it/s]
 30%|██▉       | 3973/13317 [40:55<1:35:35,  1.63it/s]
                                                      
{'loss': 0.1843, 'grad_norm': 0.566017746925354, 'learning_rate': 8.848971416917376e-06, 'epoch': 0.9}

 30%|██▉       | 3973/13317 [40:55<1:35:35,  1.63it/s]
 30%|██▉       | 3974/13317 [40:55<1:35:33,  1.63it/s]
                                                      
{'loss': 0.23, 'grad_norm': 0.544158935546875, 'learning_rate': 8.848134716624e-06, 'epoch': 0.9}

 30%|██▉       | 3974/13317 [40:55<1:35:33,  1.63it/s]
 30%|██▉       | 3975/13317 [40:56<1:35:32,  1.63it/s]
                                                      
{'loss': 0.2957, 'grad_norm': 0.6026687026023865, 'learning_rate': 8.84729775192302e-06, 'epoch': 0.9}

 30%|██▉       | 3975/13317 [40:56<1:35:32,  1.63it/s]
 30%|██▉       | 3976/13317 [40:56<1:35:33,  1.63it/s]
                                                      
{'loss': 0.2276, 'grad_norm': 0.5026265978813171, 'learning_rate': 8.846460522871954e-06, 'epoch': 0.9}

 30%|██▉       | 3976/13317 [40:57<1:35:33,  1.63it/s]
 30%|██▉       | 3977/13317 [40:57<1:35:32,  1.63it/s]
                                                      
{'loss': 0.2648, 'grad_norm': 0.5991185307502747, 'learning_rate': 8.845623029528319e-06, 'epoch': 0.9}

 30%|██▉       | 3977/13317 [40:57<1:35:32,  1.63it/s]
 30%|██▉       | 3978/13317 [40:58<1:35:29,  1.63it/s]
                                                      
{'loss': 0.1795, 'grad_norm': 0.4679270386695862, 'learning_rate': 8.844785271949665e-06, 'epoch': 0.9}

 30%|██▉       | 3978/13317 [40:58<1:35:29,  1.63it/s]
 30%|██▉       | 3979/13317 [40:58<1:35:30,  1.63it/s]
                                                      
{'loss': 0.2349, 'grad_norm': 0.46141108870506287, 'learning_rate': 8.843947250193555e-06, 'epoch': 0.9}

 30%|██▉       | 3979/13317 [40:58<1:35:30,  1.63it/s]
 30%|██▉       | 3980/13317 [40:59<1:35:30,  1.63it/s]
                                                      
{'loss': 0.2349, 'grad_norm': 0.5298835039138794, 'learning_rate': 8.843108964317568e-06, 'epoch': 0.9}

 30%|██▉       | 3980/13317 [40:59<1:35:30,  1.63it/s]
 30%|██▉       | 3981/13317 [41:00<1:35:30,  1.63it/s]
                                                      
{'loss': 0.2575, 'grad_norm': 0.5686982870101929, 'learning_rate': 8.842270414379302e-06, 'epoch': 0.9}

 30%|██▉       | 3981/13317 [41:00<1:35:30,  1.63it/s]
 30%|██▉       | 3982/13317 [41:00<1:35:28,  1.63it/s]
                                                      
{'loss': 0.2201, 'grad_norm': 0.5368353128433228, 'learning_rate': 8.841431600436377e-06, 'epoch': 0.9}

 30%|██▉       | 3982/13317 [41:00<1:35:28,  1.63it/s]
 30%|██▉       | 3983/13317 [41:01<1:35:28,  1.63it/s]
                                                      
{'loss': 0.2153, 'grad_norm': 0.4568006992340088, 'learning_rate': 8.840592522546429e-06, 'epoch': 0.9}

 30%|██▉       | 3983/13317 [41:01<1:35:28,  1.63it/s]
 30%|██▉       | 3984/13317 [41:01<1:35:25,  1.63it/s]
                                                      
{'loss': 0.3159, 'grad_norm': 0.5737602710723877, 'learning_rate': 8.839753180767108e-06, 'epoch': 0.9}

 30%|██▉       | 3984/13317 [41:02<1:35:25,  1.63it/s]
 30%|██▉       | 3985/13317 [41:02<1:35:49,  1.62it/s]
                                                      
{'loss': 0.3009, 'grad_norm': 0.5637494325637817, 'learning_rate': 8.838913575156088e-06, 'epoch': 0.9}

 30%|██▉       | 3985/13317 [41:02<1:35:49,  1.62it/s]
 30%|██▉       | 3986/13317 [41:03<1:35:42,  1.62it/s]
                                                      
{'loss': 0.2443, 'grad_norm': 0.5551648736000061, 'learning_rate': 8.838073705771057e-06, 'epoch': 0.9}

 30%|██▉       | 3986/13317 [41:03<1:35:42,  1.62it/s]
 30%|██▉       | 3987/13317 [41:03<1:35:43,  1.62it/s]
                                                      
{'loss': 0.2147, 'grad_norm': 0.46810564398765564, 'learning_rate': 8.837233572669725e-06, 'epoch': 0.9}

 30%|██▉       | 3987/13317 [41:03<1:35:43,  1.62it/s]
 30%|██▉       | 3988/13317 [41:04<1:35:42,  1.62it/s]
                                                      
{'loss': 0.2462, 'grad_norm': 0.5012428164482117, 'learning_rate': 8.836393175909819e-06, 'epoch': 0.9}

 30%|██▉       | 3988/13317 [41:04<1:35:42,  1.62it/s]
 30%|██▉       | 3989/13317 [41:04<1:35:36,  1.63it/s]
                                                      
{'loss': 0.2296, 'grad_norm': 0.5189476609230042, 'learning_rate': 8.83555251554908e-06, 'epoch': 0.9}

 30%|██▉       | 3989/13317 [41:05<1:35:36,  1.63it/s]
 30%|██▉       | 3990/13317 [41:05<1:35:33,  1.63it/s]
                                                      
{'loss': 0.2022, 'grad_norm': 0.43410560488700867, 'learning_rate': 8.83471159164527e-06, 'epoch': 0.9}

 30%|██▉       | 3990/13317 [41:05<1:35:33,  1.63it/s]
 30%|██▉       | 3991/13317 [41:06<1:35:28,  1.63it/s]
                                                      
{'loss': 0.191, 'grad_norm': 0.44419848918914795, 'learning_rate': 8.833870404256174e-06, 'epoch': 0.9}

 30%|██▉       | 3991/13317 [41:06<1:35:28,  1.63it/s]
 30%|██▉       | 3992/13317 [41:06<1:35:31,  1.63it/s]
                                                      
{'loss': 0.2728, 'grad_norm': 0.6042153239250183, 'learning_rate': 8.833028953439586e-06, 'epoch': 0.9}

 30%|██▉       | 3992/13317 [41:06<1:35:31,  1.63it/s]
 30%|██▉       | 3993/13317 [41:07<1:35:27,  1.63it/s]
                                                      
{'loss': 0.2372, 'grad_norm': 0.4906280040740967, 'learning_rate': 8.832187239253325e-06, 'epoch': 0.9}

 30%|██▉       | 3993/13317 [41:07<1:35:27,  1.63it/s]
 30%|██▉       | 3994/13317 [41:08<1:35:25,  1.63it/s]
                                                      
{'loss': 0.2375, 'grad_norm': 0.6253077387809753, 'learning_rate': 8.831345261755225e-06, 'epoch': 0.9}

 30%|██▉       | 3994/13317 [41:08<1:35:25,  1.63it/s]
 30%|██▉       | 3995/13317 [41:08<1:35:29,  1.63it/s]
                                                      
{'loss': 0.2032, 'grad_norm': 0.48869195580482483, 'learning_rate': 8.830503021003137e-06, 'epoch': 0.9}

 30%|██▉       | 3995/13317 [41:08<1:35:29,  1.63it/s]
 30%|███       | 3996/13317 [41:09<1:35:24,  1.63it/s]
                                                      
{'loss': 0.2316, 'grad_norm': 0.4613193869590759, 'learning_rate': 8.829660517054934e-06, 'epoch': 0.9}

 30%|███       | 3996/13317 [41:09<1:35:24,  1.63it/s]
 30%|███       | 3997/13317 [41:09<1:35:22,  1.63it/s]
                                                      
{'loss': 0.1944, 'grad_norm': 0.4673929512500763, 'learning_rate': 8.828817749968505e-06, 'epoch': 0.9}

 30%|███       | 3997/13317 [41:10<1:35:22,  1.63it/s]
 30%|███       | 3998/13317 [41:10<1:35:18,  1.63it/s]
                                                      
{'loss': 0.2281, 'grad_norm': 0.5460035800933838, 'learning_rate': 8.827974719801758e-06, 'epoch': 0.9}

 30%|███       | 3998/13317 [41:10<1:35:18,  1.63it/s]
 30%|███       | 3999/13317 [41:11<1:35:16,  1.63it/s]
                                                      
{'loss': 0.2357, 'grad_norm': 0.5051767230033875, 'learning_rate': 8.827131426612613e-06, 'epoch': 0.9}

 30%|███       | 3999/13317 [41:11<1:35:16,  1.63it/s]
 30%|███       | 4000/13317 [41:11<1:35:24,  1.63it/s]
                                                      
{'loss': 0.2741, 'grad_norm': 0.5997334122657776, 'learning_rate': 8.826287870459019e-06, 'epoch': 0.9}

 30%|███       | 4000/13317 [41:11<1:35:24,  1.63it/s]

  0%|          | 0/13 [00:00<?, ?it/s]

 15%|█▌        | 2/13 [00:00<00:00, 13.36it/s]

 31%|███       | 4/13 [00:00<00:01,  8.40it/s]

 38%|███▊      | 5/13 [00:00<00:01,  7.80it/s]

 46%|████▌     | 6/13 [00:00<00:00,  7.43it/s]

 54%|█████▍    | 7/13 [00:00<00:00,  7.18it/s]

 62%|██████▏   | 8/13 [00:01<00:00,  7.02it/s]

 69%|██████▉   | 9/13 [00:01<00:00,  6.91it/s]

 77%|███████▋  | 10/13 [00:01<00:00,  6.83it/s]

 85%|████████▍ | 11/13 [00:01<00:00,  6.78it/s]

 92%|█████████▏| 12/13 [00:01<00:00,  6.75it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]
                                                      

                                               
{'eval_loss': 0.642302930355072, 'eval_runtime': 1.9524, 'eval_samples_per_second': 102.436, 'eval_steps_per_second': 6.658, 'epoch': 0.9}

 30%|███       | 4000/13317 [41:13<1:35:24,  1.63it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]

                                               
 30%|███       | 4001/13317 [41:14<3:06:33,  1.20s/it]
                                                      
{'loss': 0.2567, 'grad_norm': 0.5598866939544678, 'learning_rate': 8.825444051398936e-06, 'epoch': 0.9}

 30%|███       | 4001/13317 [41:14<3:06:33,  1.20s/it]
 30%|███       | 4002/13317 [41:14<2:39:08,  1.03s/it]
                                                      
{'loss': 0.1899, 'grad_norm': 0.44073486328125, 'learning_rate': 8.82459996949034e-06, 'epoch': 0.9}

 30%|███       | 4002/13317 [41:15<2:39:08,  1.03s/it]
 30%|███       | 4003/13317 [41:15<2:19:56,  1.11it/s]
                                                      
{'loss': 0.3173, 'grad_norm': 0.6477890610694885, 'learning_rate': 8.82375562479123e-06, 'epoch': 0.9}

 30%|███       | 4003/13317 [41:15<2:19:56,  1.11it/s]
 30%|███       | 4004/13317 [41:16<2:06:30,  1.23it/s]
                                                      
{'loss': 0.3382, 'grad_norm': 0.5794007778167725, 'learning_rate': 8.822911017359623e-06, 'epoch': 0.9}

 30%|███       | 4004/13317 [41:16<2:06:30,  1.23it/s]
 30%|███       | 4005/13317 [41:16<1:57:04,  1.33it/s]
                                                      
{'loss': 0.3383, 'grad_norm': 0.6036810278892517, 'learning_rate': 8.822066147253552e-06, 'epoch': 0.9}

 30%|███       | 4005/13317 [41:16<1:57:04,  1.33it/s]
 30%|███       | 4006/13317 [41:17<1:50:27,  1.40it/s]
                                                      
{'loss': 0.2537, 'grad_norm': 0.5710217952728271, 'learning_rate': 8.821221014531067e-06, 'epoch': 0.9}

 30%|███       | 4006/13317 [41:17<1:50:27,  1.40it/s]
 30%|███       | 4007/13317 [41:17<1:46:04,  1.46it/s]
                                                      
{'loss': 0.2737, 'grad_norm': 0.4776287376880646, 'learning_rate': 8.820375619250239e-06, 'epoch': 0.9}

 30%|███       | 4007/13317 [41:18<1:46:04,  1.46it/s]
 30%|███       | 4008/13317 [41:18<1:42:50,  1.51it/s]
                                                      
{'loss': 0.2348, 'grad_norm': 0.4924545884132385, 'learning_rate': 8.819529961469155e-06, 'epoch': 0.9}

 30%|███       | 4008/13317 [41:18<1:42:50,  1.51it/s]
 30%|███       | 4009/13317 [41:19<1:40:31,  1.54it/s]
                                                      
{'loss': 0.2228, 'grad_norm': 0.512880265712738, 'learning_rate': 8.818684041245922e-06, 'epoch': 0.9}

 30%|███       | 4009/13317 [41:19<1:40:31,  1.54it/s]
 30%|███       | 4010/13317 [41:19<1:38:56,  1.57it/s]
                                                      
{'loss': 0.2435, 'grad_norm': 0.5384522676467896, 'learning_rate': 8.81783785863866e-06, 'epoch': 0.9}

 30%|███       | 4010/13317 [41:19<1:38:56,  1.57it/s]
 30%|███       | 4011/13317 [41:20<1:37:48,  1.59it/s]
                                                      
{'loss': 0.2814, 'grad_norm': 0.5525196194648743, 'learning_rate': 8.816991413705515e-06, 'epoch': 0.9}

 30%|███       | 4011/13317 [41:20<1:37:48,  1.59it/s]
 30%|███       | 4012/13317 [41:21<1:37:07,  1.60it/s]
                                                      
{'loss': 0.2155, 'grad_norm': 0.5749390125274658, 'learning_rate': 8.816144706504646e-06, 'epoch': 0.9}

 30%|███       | 4012/13317 [41:21<1:37:07,  1.60it/s]
 30%|███       | 4013/13317 [41:21<1:36:34,  1.61it/s]
                                                      
{'loss': 0.1766, 'grad_norm': 0.5054144263267517, 'learning_rate': 8.815297737094229e-06, 'epoch': 0.9}

 30%|███       | 4013/13317 [41:21<1:36:34,  1.61it/s]
 30%|███       | 4014/13317 [41:22<1:36:08,  1.61it/s]
                                                      
{'loss': 0.2477, 'grad_norm': 0.500525951385498, 'learning_rate': 8.81445050553246e-06, 'epoch': 0.9}

 30%|███       | 4014/13317 [41:22<1:36:08,  1.61it/s]
 30%|███       | 4015/13317 [41:22<1:35:48,  1.62it/s]
                                                      
{'loss': 0.2669, 'grad_norm': 0.5724359750747681, 'learning_rate': 8.813603011877551e-06, 'epoch': 0.9}

 30%|███       | 4015/13317 [41:23<1:35:48,  1.62it/s]
 30%|███       | 4016/13317 [41:23<1:35:38,  1.62it/s]
                                                      
{'loss': 0.1746, 'grad_norm': 0.4407348036766052, 'learning_rate': 8.812755256187739e-06, 'epoch': 0.9}

 30%|███       | 4016/13317 [41:23<1:35:38,  1.62it/s]
 30%|███       | 4017/13317 [41:24<1:35:36,  1.62it/s]
                                                      
{'loss': 0.2556, 'grad_norm': 0.5599730014801025, 'learning_rate': 8.81190723852127e-06, 'epoch': 0.9}

 30%|███       | 4017/13317 [41:24<1:35:36,  1.62it/s]
 30%|███       | 4018/13317 [41:24<1:35:27,  1.62it/s]
                                                      
{'loss': 0.2282, 'grad_norm': 0.43869850039482117, 'learning_rate': 8.81105895893641e-06, 'epoch': 0.91}

 30%|███       | 4018/13317 [41:24<1:35:27,  1.62it/s]
 30%|███       | 4019/13317 [41:25<1:35:18,  1.63it/s]
                                                      
{'loss': 0.2482, 'grad_norm': 0.5750786066055298, 'learning_rate': 8.810210417491452e-06, 'epoch': 0.91}

 30%|███       | 4019/13317 [41:25<1:35:18,  1.63it/s]
 30%|███       | 4020/13317 [41:25<1:35:25,  1.62it/s]
                                                      
{'loss': 0.2299, 'grad_norm': 0.55069899559021, 'learning_rate': 8.809361614244694e-06, 'epoch': 0.91}

 30%|███       | 4020/13317 [41:26<1:35:25,  1.62it/s]
 30%|███       | 4021/13317 [41:26<1:35:16,  1.63it/s]
                                                      
{'loss': 0.1829, 'grad_norm': 0.4956595003604889, 'learning_rate': 8.808512549254457e-06, 'epoch': 0.91}

 30%|███       | 4021/13317 [41:26<1:35:16,  1.63it/s]
 30%|███       | 4022/13317 [41:27<1:35:13,  1.63it/s]
                                                      
{'loss': 0.2613, 'grad_norm': 0.5729961395263672, 'learning_rate': 8.807663222579084e-06, 'epoch': 0.91}

 30%|███       | 4022/13317 [41:27<1:35:13,  1.63it/s]
 30%|███       | 4023/13317 [41:27<1:35:07,  1.63it/s]
                                                      
{'loss': 0.2336, 'grad_norm': 0.5313163995742798, 'learning_rate': 8.80681363427693e-06, 'epoch': 0.91}

 30%|███       | 4023/13317 [41:27<1:35:07,  1.63it/s]
 30%|███       | 4024/13317 [41:28<1:35:04,  1.63it/s]
                                                      
{'loss': 0.2186, 'grad_norm': 0.512514591217041, 'learning_rate': 8.805963784406374e-06, 'epoch': 0.91}

 30%|███       | 4024/13317 [41:28<1:35:04,  1.63it/s]
 30%|███       | 4025/13317 [41:29<1:35:06,  1.63it/s]
                                                      
{'loss': 0.2132, 'grad_norm': 0.4863186478614807, 'learning_rate': 8.805113673025807e-06, 'epoch': 0.91}

 30%|███       | 4025/13317 [41:29<1:35:06,  1.63it/s]
 30%|███       | 4026/13317 [41:29<1:35:04,  1.63it/s]
                                                      
{'loss': 0.2025, 'grad_norm': 0.48528745770454407, 'learning_rate': 8.80426330019364e-06, 'epoch': 0.91}

 30%|███       | 4026/13317 [41:29<1:35:04,  1.63it/s]
 30%|███       | 4027/13317 [41:30<1:35:28,  1.62it/s]
                                                      
{'loss': 0.2573, 'grad_norm': 0.5068069696426392, 'learning_rate': 8.803412665968307e-06, 'epoch': 0.91}

 30%|███       | 4027/13317 [41:30<1:35:28,  1.62it/s]
 30%|███       | 4028/13317 [41:30<1:35:21,  1.62it/s]
                                                      
{'loss': 0.3794, 'grad_norm': 0.6572163105010986, 'learning_rate': 8.80256177040825e-06, 'epoch': 0.91}

 30%|███       | 4028/13317 [41:31<1:35:21,  1.62it/s]
 30%|███       | 4029/13317 [41:31<1:35:14,  1.63it/s]
                                                      
{'loss': 0.286, 'grad_norm': 0.5493881106376648, 'learning_rate': 8.801710613571939e-06, 'epoch': 0.91}

 30%|███       | 4029/13317 [41:31<1:35:14,  1.63it/s]
 30%|███       | 4030/13317 [41:32<1:35:14,  1.63it/s]
                                                      
{'loss': 0.2829, 'grad_norm': 0.6724544763565063, 'learning_rate': 8.800859195517856e-06, 'epoch': 0.91}

 30%|███       | 4030/13317 [41:32<1:35:14,  1.63it/s]
 30%|███       | 4031/13317 [41:32<1:35:10,  1.63it/s]
                                                      
{'loss': 0.2301, 'grad_norm': 0.45262041687965393, 'learning_rate': 8.800007516304501e-06, 'epoch': 0.91}

 30%|███       | 4031/13317 [41:32<1:35:10,  1.63it/s]
 30%|███       | 4032/13317 [41:33<1:35:07,  1.63it/s]
                                                      
{'loss': 0.2864, 'grad_norm': 0.6116564869880676, 'learning_rate': 8.799155575990394e-06, 'epoch': 0.91}

 30%|███       | 4032/13317 [41:33<1:35:07,  1.63it/s]
 30%|███       | 4033/13317 [41:33<1:35:00,  1.63it/s]
                                                      
{'loss': 0.1657, 'grad_norm': 0.36845284700393677, 'learning_rate': 8.798303374634073e-06, 'epoch': 0.91}

 30%|███       | 4033/13317 [41:34<1:35:00,  1.63it/s]
 30%|███       | 4034/13317 [41:34<1:34:58,  1.63it/s]
                                                      
{'loss': 0.2909, 'grad_norm': 0.5481066107749939, 'learning_rate': 8.797450912294095e-06, 'epoch': 0.91}

 30%|███       | 4034/13317 [41:34<1:34:58,  1.63it/s]
 30%|███       | 4035/13317 [41:35<1:35:04,  1.63it/s]
                                                      
{'loss': 0.3145, 'grad_norm': 0.594548225402832, 'learning_rate': 8.79659818902903e-06, 'epoch': 0.91}

 30%|███       | 4035/13317 [41:35<1:35:04,  1.63it/s]
 30%|███       | 4036/13317 [41:35<1:34:57,  1.63it/s]
                                                      
{'loss': 0.2907, 'grad_norm': 0.5038217902183533, 'learning_rate': 8.79574520489747e-06, 'epoch': 0.91}

 30%|███       | 4036/13317 [41:35<1:34:57,  1.63it/s]
 30%|███       | 4037/13317 [41:36<1:34:57,  1.63it/s]
                                                      
{'loss': 0.2842, 'grad_norm': 0.5770299434661865, 'learning_rate': 8.794891959958024e-06, 'epoch': 0.91}

 30%|███       | 4037/13317 [41:36<1:34:57,  1.63it/s]
 30%|███       | 4038/13317 [41:37<1:34:55,  1.63it/s]
                                                      
{'loss': 0.3103, 'grad_norm': 0.7390388250350952, 'learning_rate': 8.79403845426932e-06, 'epoch': 0.91}

 30%|███       | 4038/13317 [41:37<1:34:55,  1.63it/s]
 30%|███       | 4039/13317 [41:37<1:34:52,  1.63it/s]
                                                      
{'loss': 0.28, 'grad_norm': 0.5863255858421326, 'learning_rate': 8.793184687890004e-06, 'epoch': 0.91}

 30%|███       | 4039/13317 [41:37<1:34:52,  1.63it/s]
 30%|███       | 4040/13317 [41:38<1:34:59,  1.63it/s]
                                                      
{'loss': 0.2246, 'grad_norm': 0.4899351894855499, 'learning_rate': 8.792330660878735e-06, 'epoch': 0.91}

 30%|███       | 4040/13317 [41:38<1:34:59,  1.63it/s]
 30%|███       | 4041/13317 [41:38<1:34:58,  1.63it/s]
                                                      
{'loss': 0.2286, 'grad_norm': 0.5388619303703308, 'learning_rate': 8.791476373294196e-06, 'epoch': 0.91}

 30%|███       | 4041/13317 [41:39<1:34:58,  1.63it/s]
 30%|███       | 4042/13317 [41:39<1:34:54,  1.63it/s]
                                                      
{'loss': 0.2338, 'grad_norm': 0.476406991481781, 'learning_rate': 8.790621825195085e-06, 'epoch': 0.91}

 30%|███       | 4042/13317 [41:39<1:34:54,  1.63it/s]
 30%|███       | 4043/13317 [41:40<1:34:55,  1.63it/s]
                                                      
{'loss': 0.2558, 'grad_norm': 0.5069482326507568, 'learning_rate': 8.789767016640121e-06, 'epoch': 0.91}

 30%|███       | 4043/13317 [41:40<1:34:55,  1.63it/s]
 30%|███       | 4044/13317 [41:40<1:34:55,  1.63it/s]
                                                      
{'loss': 0.2095, 'grad_norm': 0.45869141817092896, 'learning_rate': 8.788911947688034e-06, 'epoch': 0.91}

 30%|███       | 4044/13317 [41:40<1:34:55,  1.63it/s]
 30%|███       | 4045/13317 [41:41<1:34:59,  1.63it/s]
                                                      
{'loss': 0.1953, 'grad_norm': 0.4905432164669037, 'learning_rate': 8.788056618397578e-06, 'epoch': 0.91}

 30%|███       | 4045/13317 [41:41<1:34:59,  1.63it/s]
 30%|███       | 4046/13317 [41:41<1:34:52,  1.63it/s]
                                                      
{'loss': 0.1451, 'grad_norm': 0.39281323552131653, 'learning_rate': 8.787201028827527e-06, 'epoch': 0.91}

 30%|███       | 4046/13317 [41:42<1:34:52,  1.63it/s]
 30%|███       | 4047/13317 [41:42<1:34:53,  1.63it/s]
                                                      
{'loss': 0.2221, 'grad_norm': 0.5506792664527893, 'learning_rate': 8.786345179036664e-06, 'epoch': 0.91}

 30%|███       | 4047/13317 [41:42<1:34:53,  1.63it/s]
 30%|███       | 4048/13317 [41:43<1:34:50,  1.63it/s]
                                                      
{'loss': 0.2147, 'grad_norm': 0.4366404116153717, 'learning_rate': 8.785489069083796e-06, 'epoch': 0.91}

 30%|███       | 4048/13317 [41:43<1:34:50,  1.63it/s]
 30%|███       | 4049/13317 [41:43<1:34:46,  1.63it/s]
                                                      
{'loss': 0.2417, 'grad_norm': 0.5659778714179993, 'learning_rate': 8.784632699027748e-06, 'epoch': 0.91}

 30%|███       | 4049/13317 [41:43<1:34:46,  1.63it/s]
 30%|███       | 4050/13317 [41:44<1:35:21,  1.62it/s]
                                                      
{'loss': 0.26, 'grad_norm': 0.5649353861808777, 'learning_rate': 8.783776068927362e-06, 'epoch': 0.91}

 30%|███       | 4050/13317 [41:44<1:35:21,  1.62it/s]
 30%|███       | 4051/13317 [41:45<1:35:08,  1.62it/s]
                                                      
{'loss': 0.3386, 'grad_norm': 0.551205575466156, 'learning_rate': 8.782919178841495e-06, 'epoch': 0.91}

 30%|███       | 4051/13317 [41:45<1:35:08,  1.62it/s]
 30%|███       | 4052/13317 [41:45<1:35:03,  1.62it/s]
                                                      
{'loss': 0.2019, 'grad_norm': 0.446676105260849, 'learning_rate': 8.782062028829028e-06, 'epoch': 0.91}

 30%|███       | 4052/13317 [41:45<1:35:03,  1.62it/s]
 30%|███       | 4053/13317 [41:46<1:34:56,  1.63it/s]
                                                      
{'loss': 0.2124, 'grad_norm': 0.504501461982727, 'learning_rate': 8.781204618948853e-06, 'epoch': 0.91}

 30%|███       | 4053/13317 [41:46<1:34:56,  1.63it/s]
 30%|███       | 4054/13317 [41:46<1:34:56,  1.63it/s]
                                                      
{'loss': 0.2673, 'grad_norm': 0.487269788980484, 'learning_rate': 8.780346949259885e-06, 'epoch': 0.91}

 30%|███       | 4054/13317 [41:47<1:34:56,  1.63it/s]
 30%|███       | 4055/13317 [41:47<1:35:00,  1.62it/s]
                                                      
{'loss': 0.1975, 'grad_norm': 0.409564346075058, 'learning_rate': 8.779489019821057e-06, 'epoch': 0.91}

 30%|███       | 4055/13317 [41:47<1:35:00,  1.62it/s]
 30%|███       | 4056/13317 [41:48<1:34:55,  1.63it/s]
                                                      
{'loss': 0.2493, 'grad_norm': 0.5151880979537964, 'learning_rate': 8.778630830691313e-06, 'epoch': 0.91}

 30%|███       | 4056/13317 [41:48<1:34:55,  1.63it/s]
 30%|███       | 4057/13317 [41:48<1:34:55,  1.63it/s]
                                                      
{'loss': 0.2164, 'grad_norm': 0.510475218296051, 'learning_rate': 8.777772381929622e-06, 'epoch': 0.91}

 30%|███       | 4057/13317 [41:48<1:34:55,  1.63it/s]
 30%|███       | 4058/13317 [41:49<1:34:48,  1.63it/s]
                                                      
{'loss': 0.2264, 'grad_norm': 0.47544676065444946, 'learning_rate': 8.776913673594968e-06, 'epoch': 0.91}

 30%|███       | 4058/13317 [41:49<1:34:48,  1.63it/s]
 30%|███       | 4059/13317 [41:49<1:34:44,  1.63it/s]
                                                      
{'loss': 0.22, 'grad_norm': 0.4581185281276703, 'learning_rate': 8.776054705746357e-06, 'epoch': 0.91}

 30%|███       | 4059/13317 [41:50<1:34:44,  1.63it/s]
 30%|███       | 4060/13317 [41:50<1:34:48,  1.63it/s]
                                                      
{'loss': 0.2215, 'grad_norm': 0.4965716004371643, 'learning_rate': 8.775195478442804e-06, 'epoch': 0.91}

 30%|███       | 4060/13317 [41:50<1:34:48,  1.63it/s]
 30%|███       | 4061/13317 [41:51<1:34:44,  1.63it/s]
                                                      
{'loss': 0.2729, 'grad_norm': 0.48958346247673035, 'learning_rate': 8.77433599174335e-06, 'epoch': 0.91}

 30%|███       | 4061/13317 [41:51<1:34:44,  1.63it/s]
 31%|███       | 4062/13317 [41:51<1:34:51,  1.63it/s]
                                                      
{'loss': 0.2302, 'grad_norm': 0.5194380879402161, 'learning_rate': 8.77347624570705e-06, 'epoch': 0.92}

 31%|███       | 4062/13317 [41:51<1:34:51,  1.63it/s]
 31%|███       | 4063/13317 [41:52<1:34:47,  1.63it/s]
                                                      
{'loss': 0.262, 'grad_norm': 0.5403594970703125, 'learning_rate': 8.772616240392979e-06, 'epoch': 0.92}

 31%|███       | 4063/13317 [41:52<1:34:47,  1.63it/s]
 31%|███       | 4064/13317 [41:53<1:34:51,  1.63it/s]
                                                      
{'loss': 0.2817, 'grad_norm': 0.5616006255149841, 'learning_rate': 8.771755975860226e-06, 'epoch': 0.92}

 31%|███       | 4064/13317 [41:53<1:34:51,  1.63it/s]
 31%|███       | 4065/13317 [41:53<1:34:53,  1.62it/s]
                                                      
{'loss': 0.3083, 'grad_norm': 0.7022930979728699, 'learning_rate': 8.7708954521679e-06, 'epoch': 0.92}

 31%|███       | 4065/13317 [41:53<1:34:53,  1.62it/s]
 31%|███       | 4066/13317 [41:54<1:34:49,  1.63it/s]
                                                      
{'loss': 0.191, 'grad_norm': 0.42451250553131104, 'learning_rate': 8.77003466937513e-06, 'epoch': 0.92}

 31%|███       | 4066/13317 [41:54<1:34:49,  1.63it/s]
 31%|███       | 4067/13317 [41:54<1:34:43,  1.63it/s]
                                                      
{'loss': 0.2266, 'grad_norm': 0.4983607530593872, 'learning_rate': 8.769173627541063e-06, 'epoch': 0.92}

 31%|███       | 4067/13317 [41:54<1:34:43,  1.63it/s]
 31%|███       | 4068/13317 [41:55<1:34:41,  1.63it/s]
                                                      
{'loss': 0.2937, 'grad_norm': 0.5934345722198486, 'learning_rate': 8.768312326724856e-06, 'epoch': 0.92}

 31%|███       | 4068/13317 [41:55<1:34:41,  1.63it/s]
 31%|███       | 4069/13317 [41:56<1:34:45,  1.63it/s]
                                                      
{'loss': 0.3136, 'grad_norm': 0.5125113129615784, 'learning_rate': 8.767450766985695e-06, 'epoch': 0.92}

 31%|███       | 4069/13317 [41:56<1:34:45,  1.63it/s]
 31%|███       | 4070/13317 [41:56<1:34:50,  1.62it/s]
                                                      
{'loss': 0.33, 'grad_norm': 0.6946620345115662, 'learning_rate': 8.766588948382774e-06, 'epoch': 0.92}

 31%|███       | 4070/13317 [41:56<1:34:50,  1.62it/s]
 31%|███       | 4071/13317 [41:57<1:34:44,  1.63it/s]
                                                      
{'loss': 0.2423, 'grad_norm': 0.5411499738693237, 'learning_rate': 8.765726870975314e-06, 'epoch': 0.92}

 31%|███       | 4071/13317 [41:57<1:34:44,  1.63it/s]
 31%|███       | 4072/13317 [41:57<1:34:46,  1.63it/s]
                                                      
{'loss': 0.2533, 'grad_norm': 0.6122368574142456, 'learning_rate': 8.764864534822542e-06, 'epoch': 0.92}

 31%|███       | 4072/13317 [41:58<1:34:46,  1.63it/s]
 31%|███       | 4073/13317 [41:58<1:34:37,  1.63it/s]
                                                      
{'loss': 0.2472, 'grad_norm': 0.6923509240150452, 'learning_rate': 8.764001939983716e-06, 'epoch': 0.92}

 31%|███       | 4073/13317 [41:58<1:34:37,  1.63it/s]
 31%|███       | 4074/13317 [41:59<1:34:36,  1.63it/s]
                                                      
{'loss': 0.2751, 'grad_norm': 0.5493807792663574, 'learning_rate': 8.763139086518103e-06, 'epoch': 0.92}

 31%|███       | 4074/13317 [41:59<1:34:36,  1.63it/s]
 31%|███       | 4075/13317 [41:59<1:34:44,  1.63it/s]
                                                      
{'loss': 0.2542, 'grad_norm': 0.5987316966056824, 'learning_rate': 8.762275974484988e-06, 'epoch': 0.92}

 31%|███       | 4075/13317 [41:59<1:34:44,  1.63it/s]
 31%|███       | 4076/13317 [42:00<1:34:38,  1.63it/s]
                                                      
{'loss': 0.1992, 'grad_norm': 0.47123926877975464, 'learning_rate': 8.761412603943678e-06, 'epoch': 0.92}

 31%|███       | 4076/13317 [42:00<1:34:38,  1.63it/s]
 31%|███       | 4077/13317 [42:01<1:34:36,  1.63it/s]
                                                      
{'loss': 0.2827, 'grad_norm': 0.5451420545578003, 'learning_rate': 8.760548974953496e-06, 'epoch': 0.92}

 31%|███       | 4077/13317 [42:01<1:34:36,  1.63it/s]
 31%|███       | 4078/13317 [42:01<1:34:33,  1.63it/s]
                                                      
{'loss': 0.2668, 'grad_norm': 0.5446875095367432, 'learning_rate': 8.759685087573783e-06, 'epoch': 0.92}

 31%|███       | 4078/13317 [42:01<1:34:33,  1.63it/s]
 31%|███       | 4079/13317 [42:02<1:34:30,  1.63it/s]
                                                      
{'loss': 0.2128, 'grad_norm': 0.5191280841827393, 'learning_rate': 8.758820941863898e-06, 'epoch': 0.92}

 31%|███       | 4079/13317 [42:02<1:34:30,  1.63it/s]
 31%|███       | 4080/13317 [42:02<1:34:36,  1.63it/s]
                                                      
{'loss': 0.2791, 'grad_norm': 0.5823637843132019, 'learning_rate': 8.757956537883212e-06, 'epoch': 0.92}

 31%|███       | 4080/13317 [42:02<1:34:36,  1.63it/s]
 31%|███       | 4081/13317 [42:03<1:34:31,  1.63it/s]
                                                      
{'loss': 0.2627, 'grad_norm': 0.46480652689933777, 'learning_rate': 8.757091875691123e-06, 'epoch': 0.92}

 31%|███       | 4081/13317 [42:03<1:34:31,  1.63it/s]
 31%|███       | 4082/13317 [42:04<1:34:34,  1.63it/s]
                                                      
{'loss': 0.1986, 'grad_norm': 0.4384443759918213, 'learning_rate': 8.756226955347042e-06, 'epoch': 0.92}

 31%|███       | 4082/13317 [42:04<1:34:34,  1.63it/s]
 31%|███       | 4083/13317 [42:04<1:34:31,  1.63it/s]
                                                      
{'loss': 0.1846, 'grad_norm': 0.3908243477344513, 'learning_rate': 8.755361776910396e-06, 'epoch': 0.92}

 31%|███       | 4083/13317 [42:04<1:34:31,  1.63it/s]
 31%|███       | 4084/13317 [42:05<1:34:28,  1.63it/s]
                                                      
{'loss': 0.3047, 'grad_norm': 0.5698086023330688, 'learning_rate': 8.754496340440635e-06, 'epoch': 0.92}

 31%|███       | 4084/13317 [42:05<1:34:28,  1.63it/s]
 31%|███       | 4085/13317 [42:05<1:34:26,  1.63it/s]
                                                      
{'loss': 0.3645, 'grad_norm': 0.6541506052017212, 'learning_rate': 8.753630645997222e-06, 'epoch': 0.92}

 31%|███       | 4085/13317 [42:06<1:34:26,  1.63it/s]
 31%|███       | 4086/13317 [42:06<1:34:23,  1.63it/s]
                                                      
{'loss': 0.2946, 'grad_norm': 0.5320442318916321, 'learning_rate': 8.75276469363964e-06, 'epoch': 0.92}

 31%|███       | 4086/13317 [42:06<1:34:23,  1.63it/s]
 31%|███       | 4087/13317 [42:07<1:34:26,  1.63it/s]
                                                      
{'loss': 0.3072, 'grad_norm': 0.5717067122459412, 'learning_rate': 8.751898483427388e-06, 'epoch': 0.92}

 31%|███       | 4087/13317 [42:07<1:34:26,  1.63it/s]
 31%|███       | 4088/13317 [42:07<1:34:23,  1.63it/s]
                                                      
{'loss': 0.2932, 'grad_norm': 0.5272173285484314, 'learning_rate': 8.751032015419985e-06, 'epoch': 0.92}

 31%|███       | 4088/13317 [42:07<1:34:23,  1.63it/s]
 31%|███       | 4089/13317 [42:08<1:34:20,  1.63it/s]
                                                      
{'loss': 0.2766, 'grad_norm': 0.5589012503623962, 'learning_rate': 8.750165289676964e-06, 'epoch': 0.92}

 31%|███       | 4089/13317 [42:08<1:34:20,  1.63it/s]
 31%|███       | 4090/13317 [42:08<1:34:17,  1.63it/s]
                                                      
{'loss': 0.1941, 'grad_norm': 0.4354220926761627, 'learning_rate': 8.749298306257883e-06, 'epoch': 0.92}

 31%|███       | 4090/13317 [42:09<1:34:17,  1.63it/s]
 31%|███       | 4091/13317 [42:09<1:34:19,  1.63it/s]
                                                      
{'loss': 0.2612, 'grad_norm': 0.5471813082695007, 'learning_rate': 8.748431065222308e-06, 'epoch': 0.92}

 31%|███       | 4091/13317 [42:09<1:34:19,  1.63it/s]
 31%|███       | 4092/13317 [42:10<1:34:18,  1.63it/s]
                                                      
{'loss': 0.3505, 'grad_norm': 0.616605281829834, 'learning_rate': 8.74756356662983e-06, 'epoch': 0.92}

 31%|███       | 4092/13317 [42:10<1:34:18,  1.63it/s]
 31%|███       | 4093/13317 [42:10<1:34:20,  1.63it/s]
                                                      
{'loss': 0.2244, 'grad_norm': 0.48438799381256104, 'learning_rate': 8.746695810540056e-06, 'epoch': 0.92}

 31%|███       | 4093/13317 [42:10<1:34:20,  1.63it/s]
 31%|███       | 4094/13317 [42:11<1:34:17,  1.63it/s]
                                                      
{'loss': 0.2939, 'grad_norm': 0.5684486031532288, 'learning_rate': 8.74582779701261e-06, 'epoch': 0.92}

 31%|███       | 4094/13317 [42:11<1:34:17,  1.63it/s]
 31%|███       | 4095/13317 [42:12<1:34:14,  1.63it/s]
                                                      
{'loss': 0.294, 'grad_norm': 0.49753233790397644, 'learning_rate': 8.74495952610713e-06, 'epoch': 0.92}

 31%|███       | 4095/13317 [42:12<1:34:14,  1.63it/s]
 31%|███       | 4096/13317 [42:12<1:34:15,  1.63it/s]
                                                      
{'loss': 0.2275, 'grad_norm': 0.5174517035484314, 'learning_rate': 8.744090997883282e-06, 'epoch': 0.92}

 31%|███       | 4096/13317 [42:12<1:34:15,  1.63it/s]
 31%|███       | 4097/13317 [42:13<1:34:15,  1.63it/s]
                                                      
{'loss': 0.1992, 'grad_norm': 0.45169150829315186, 'learning_rate': 8.743222212400736e-06, 'epoch': 0.92}

 31%|███       | 4097/13317 [42:13<1:34:15,  1.63it/s]
 31%|███       | 4098/13317 [42:13<1:34:12,  1.63it/s]
                                                      
{'loss': 0.3141, 'grad_norm': 0.6647041440010071, 'learning_rate': 8.742353169719191e-06, 'epoch': 0.92}

 31%|███       | 4098/13317 [42:14<1:34:12,  1.63it/s]
 31%|███       | 4099/13317 [42:14<1:34:13,  1.63it/s]
                                                      
{'loss': 0.2499, 'grad_norm': 0.5310629606246948, 'learning_rate': 8.74148386989836e-06, 'epoch': 0.92}

 31%|███       | 4099/13317 [42:14<1:34:13,  1.63it/s]
 31%|███       | 4100/13317 [42:15<1:34:13,  1.63it/s]
                                                      
{'loss': 0.2314, 'grad_norm': 0.5498949885368347, 'learning_rate': 8.740614312997972e-06, 'epoch': 0.92}

 31%|███       | 4100/13317 [42:15<1:34:13,  1.63it/s]
 31%|███       | 4101/13317 [42:15<1:34:13,  1.63it/s]
                                                      
{'loss': 0.2018, 'grad_norm': 0.5267596244812012, 'learning_rate': 8.739744499077773e-06, 'epoch': 0.92}

 31%|███       | 4101/13317 [42:15<1:34:13,  1.63it/s]
 31%|███       | 4102/13317 [42:16<1:34:13,  1.63it/s]
                                                      
{'loss': 0.277, 'grad_norm': 0.5886213779449463, 'learning_rate': 8.738874428197531e-06, 'epoch': 0.92}

 31%|███       | 4102/13317 [42:16<1:34:13,  1.63it/s]
 31%|███       | 4103/13317 [42:16<1:34:12,  1.63it/s]
                                                      
{'loss': 0.24, 'grad_norm': 0.4439221918582916, 'learning_rate': 8.738004100417026e-06, 'epoch': 0.92}

 31%|███       | 4103/13317 [42:17<1:34:12,  1.63it/s]
 31%|███       | 4104/13317 [42:17<1:34:10,  1.63it/s]
                                                      
{'loss': 0.244, 'grad_norm': 0.49388664960861206, 'learning_rate': 8.737133515796061e-06, 'epoch': 0.92}

 31%|███       | 4104/13317 [42:17<1:34:10,  1.63it/s]
 31%|███       | 4105/13317 [42:18<1:34:10,  1.63it/s]
                                                      
{'loss': 0.2284, 'grad_norm': 0.4649561047554016, 'learning_rate': 8.736262674394455e-06, 'epoch': 0.92}

 31%|███       | 4105/13317 [42:18<1:34:10,  1.63it/s]
 31%|███       | 4106/13317 [42:18<1:34:12,  1.63it/s]
                                                      
{'loss': 0.2633, 'grad_norm': 0.5566365122795105, 'learning_rate': 8.735391576272043e-06, 'epoch': 0.92}

 31%|███       | 4106/13317 [42:18<1:34:12,  1.63it/s]
 31%|███       | 4107/13317 [42:19<1:34:11,  1.63it/s]
                                                      
{'loss': 0.2343, 'grad_norm': 0.46882161498069763, 'learning_rate': 8.734520221488679e-06, 'epoch': 0.93}

 31%|███       | 4107/13317 [42:19<1:34:11,  1.63it/s]
 31%|███       | 4108/13317 [42:20<1:34:12,  1.63it/s]
                                                      
{'loss': 0.2663, 'grad_norm': 0.6040326356887817, 'learning_rate': 8.733648610104234e-06, 'epoch': 0.93}

 31%|███       | 4108/13317 [42:20<1:34:12,  1.63it/s]
 31%|███       | 4109/13317 [42:20<1:34:13,  1.63it/s]
                                                      
{'loss': 0.2066, 'grad_norm': 0.4500402510166168, 'learning_rate': 8.732776742178596e-06, 'epoch': 0.93}

 31%|███       | 4109/13317 [42:20<1:34:13,  1.63it/s]
 31%|███       | 4110/13317 [42:21<1:34:08,  1.63it/s]
                                                      
{'loss': 0.2191, 'grad_norm': 0.5412492752075195, 'learning_rate': 8.731904617771674e-06, 'epoch': 0.93}

 31%|███       | 4110/13317 [42:21<1:34:08,  1.63it/s]
 31%|███       | 4111/13317 [42:21<1:34:07,  1.63it/s]
                                                      
{'loss': 0.1809, 'grad_norm': 0.4552905261516571, 'learning_rate': 8.73103223694339e-06, 'epoch': 0.93}

 31%|███       | 4111/13317 [42:21<1:34:07,  1.63it/s]
 31%|███       | 4112/13317 [42:22<1:34:10,  1.63it/s]
                                                      
{'loss': 0.2985, 'grad_norm': 0.5842728614807129, 'learning_rate': 8.730159599753687e-06, 'epoch': 0.93}

 31%|███       | 4112/13317 [42:22<1:34:10,  1.63it/s]
 31%|███       | 4113/13317 [42:23<1:34:10,  1.63it/s]
                                                      
{'loss': 0.2147, 'grad_norm': 0.48068687319755554, 'learning_rate': 8.729286706262524e-06, 'epoch': 0.93}

 31%|███       | 4113/13317 [42:23<1:34:10,  1.63it/s]
 31%|███       | 4114/13317 [42:23<1:34:09,  1.63it/s]
                                                      
{'loss': 0.2706, 'grad_norm': 0.48201653361320496, 'learning_rate': 8.728413556529876e-06, 'epoch': 0.93}

 31%|███       | 4114/13317 [42:23<1:34:09,  1.63it/s]
 31%|███       | 4115/13317 [42:24<1:34:10,  1.63it/s]
                                                      
{'loss': 0.1999, 'grad_norm': 0.47993749380111694, 'learning_rate': 8.727540150615741e-06, 'epoch': 0.93}

 31%|███       | 4115/13317 [42:24<1:34:10,  1.63it/s]
 31%|███       | 4116/13317 [42:24<1:34:06,  1.63it/s]
                                                      
{'loss': 0.1691, 'grad_norm': 0.37474769353866577, 'learning_rate': 8.72666648858013e-06, 'epoch': 0.93}

 31%|███       | 4116/13317 [42:25<1:34:06,  1.63it/s]
 31%|███       | 4117/13317 [42:25<1:34:04,  1.63it/s]
                                                      
{'loss': 0.2552, 'grad_norm': 0.4901628792285919, 'learning_rate': 8.725792570483073e-06, 'epoch': 0.93}

 31%|███       | 4117/13317 [42:25<1:34:04,  1.63it/s]
 31%|███       | 4118/13317 [42:26<1:34:02,  1.63it/s]
                                                      
{'loss': 0.2351, 'grad_norm': 0.5522165298461914, 'learning_rate': 8.724918396384617e-06, 'epoch': 0.93}

 31%|███       | 4118/13317 [42:26<1:34:02,  1.63it/s]
 31%|███       | 4119/13317 [42:26<1:34:01,  1.63it/s]
                                                      
{'loss': 0.2877, 'grad_norm': 0.5869778394699097, 'learning_rate': 8.724043966344827e-06, 'epoch': 0.93}

 31%|███       | 4119/13317 [42:26<1:34:01,  1.63it/s]
 31%|███       | 4120/13317 [42:27<1:34:00,  1.63it/s]
                                                      
{'loss': 0.2704, 'grad_norm': 0.588955283164978, 'learning_rate': 8.723169280423784e-06, 'epoch': 0.93}

 31%|███       | 4120/13317 [42:27<1:34:00,  1.63it/s]
 31%|███       | 4121/13317 [42:28<1:34:00,  1.63it/s]
                                                      
{'loss': 0.1859, 'grad_norm': 0.47072240710258484, 'learning_rate': 8.722294338681592e-06, 'epoch': 0.93}

 31%|███       | 4121/13317 [42:28<1:34:00,  1.63it/s]
 31%|███       | 4122/13317 [42:28<1:34:00,  1.63it/s]
                                                      
{'loss': 0.2483, 'grad_norm': 0.4945480525493622, 'learning_rate': 8.721419141178364e-06, 'epoch': 0.93}

 31%|███       | 4122/13317 [42:28<1:34:00,  1.63it/s]
 31%|███       | 4123/13317 [42:29<1:33:58,  1.63it/s]
                                                      
{'loss': 0.2474, 'grad_norm': 0.5249830484390259, 'learning_rate': 8.72054368797424e-06, 'epoch': 0.93}

 31%|███       | 4123/13317 [42:29<1:33:58,  1.63it/s]
 31%|███       | 4124/13317 [42:29<1:33:58,  1.63it/s]
                                                      
{'loss': 0.2454, 'grad_norm': 0.5788245797157288, 'learning_rate': 8.719667979129372e-06, 'epoch': 0.93}

 31%|███       | 4124/13317 [42:29<1:33:58,  1.63it/s]
 31%|███       | 4125/13317 [42:30<1:33:57,  1.63it/s]
                                                      
{'loss': 0.2283, 'grad_norm': 0.528562068939209, 'learning_rate': 8.718792014703928e-06, 'epoch': 0.93}

 31%|███       | 4125/13317 [42:30<1:33:57,  1.63it/s]
 31%|███       | 4126/13317 [42:31<1:33:56,  1.63it/s]
                                                      
{'loss': 0.2877, 'grad_norm': 0.5345867276191711, 'learning_rate': 8.717915794758097e-06, 'epoch': 0.93}

 31%|███       | 4126/13317 [42:31<1:33:56,  1.63it/s]
 31%|███       | 4127/13317 [42:31<1:33:56,  1.63it/s]
                                                      
{'loss': 0.2321, 'grad_norm': 0.5053929090499878, 'learning_rate': 8.717039319352085e-06, 'epoch': 0.93}

 31%|███       | 4127/13317 [42:31<1:33:56,  1.63it/s]
 31%|███       | 4128/13317 [42:32<1:33:58,  1.63it/s]
                                                      
{'loss': 0.2669, 'grad_norm': 0.5675768852233887, 'learning_rate': 8.716162588546116e-06, 'epoch': 0.93}

 31%|███       | 4128/13317 [42:32<1:33:58,  1.63it/s]
 31%|███       | 4129/13317 [42:32<1:33:54,  1.63it/s]
                                                      
{'loss': 0.2043, 'grad_norm': 0.4532223641872406, 'learning_rate': 8.71528560240043e-06, 'epoch': 0.93}

 31%|███       | 4129/13317 [42:33<1:33:54,  1.63it/s]
 31%|███       | 4130/13317 [42:33<1:33:58,  1.63it/s]
                                                      
{'loss': 0.2285, 'grad_norm': 0.4025949239730835, 'learning_rate': 8.714408360975284e-06, 'epoch': 0.93}

 31%|███       | 4130/13317 [42:33<1:33:58,  1.63it/s]
 31%|███       | 4131/13317 [42:34<1:33:55,  1.63it/s]
                                                      
{'loss': 0.2647, 'grad_norm': 0.45629510283470154, 'learning_rate': 8.713530864330955e-06, 'epoch': 0.93}

 31%|███       | 4131/13317 [42:34<1:33:55,  1.63it/s]
 31%|███       | 4132/13317 [42:34<1:33:54,  1.63it/s]
                                                      
{'loss': 0.2238, 'grad_norm': 0.4808061718940735, 'learning_rate': 8.712653112527738e-06, 'epoch': 0.93}

 31%|███       | 4132/13317 [42:34<1:33:54,  1.63it/s]
 31%|███       | 4133/13317 [42:35<1:33:55,  1.63it/s]
                                                      
{'loss': 0.2899, 'grad_norm': 0.5864943265914917, 'learning_rate': 8.711775105625939e-06, 'epoch': 0.93}

 31%|███       | 4133/13317 [42:35<1:33:55,  1.63it/s]
 31%|███       | 4134/13317 [42:35<1:33:51,  1.63it/s]
                                                      
{'loss': 0.3139, 'grad_norm': 0.6354511976242065, 'learning_rate': 8.71089684368589e-06, 'epoch': 0.93}

 31%|███       | 4134/13317 [42:36<1:33:51,  1.63it/s]
 31%|███       | 4135/13317 [42:36<1:33:50,  1.63it/s]
                                                      
{'loss': 0.2134, 'grad_norm': 0.5294080376625061, 'learning_rate': 8.710018326767937e-06, 'epoch': 0.93}

 31%|███       | 4135/13317 [42:36<1:33:50,  1.63it/s]
 31%|███       | 4136/13317 [42:37<1:33:52,  1.63it/s]
                                                      
{'loss': 0.2963, 'grad_norm': 0.5812967419624329, 'learning_rate': 8.709139554932444e-06, 'epoch': 0.93}

 31%|███       | 4136/13317 [42:37<1:33:52,  1.63it/s]
 31%|███       | 4137/13317 [42:37<1:33:50,  1.63it/s]
                                                      
{'loss': 0.2016, 'grad_norm': 0.460406094789505, 'learning_rate': 8.708260528239788e-06, 'epoch': 0.93}

 31%|███       | 4137/13317 [42:37<1:33:50,  1.63it/s]
 31%|███       | 4138/13317 [42:38<1:33:48,  1.63it/s]
                                                      
{'loss': 0.317, 'grad_norm': 0.5237157344818115, 'learning_rate': 8.707381246750371e-06, 'epoch': 0.93}

 31%|███       | 4138/13317 [42:38<1:33:48,  1.63it/s]
 31%|███       | 4139/13317 [42:39<1:33:48,  1.63it/s]
                                                      
{'loss': 0.2759, 'grad_norm': 0.5630931854248047, 'learning_rate': 8.706501710524609e-06, 'epoch': 0.93}

 31%|███       | 4139/13317 [42:39<1:33:48,  1.63it/s]
 31%|███       | 4140/13317 [42:39<1:33:47,  1.63it/s]
                                                      
{'loss': 0.2415, 'grad_norm': 0.47875508666038513, 'learning_rate': 8.705621919622933e-06, 'epoch': 0.93}

 31%|███       | 4140/13317 [42:39<1:33:47,  1.63it/s]
 31%|███       | 4141/13317 [42:40<1:33:45,  1.63it/s]
                                                      
{'loss': 0.2407, 'grad_norm': 0.49596843123435974, 'learning_rate': 8.704741874105796e-06, 'epoch': 0.93}

 31%|███       | 4141/13317 [42:40<1:33:45,  1.63it/s]
 31%|███       | 4142/13317 [42:40<1:33:47,  1.63it/s]
                                                      
{'loss': 0.2364, 'grad_norm': 0.4314301013946533, 'learning_rate': 8.703861574033666e-06, 'epoch': 0.93}

 31%|███       | 4142/13317 [42:41<1:33:47,  1.63it/s]
 31%|███       | 4143/13317 [42:41<1:33:47,  1.63it/s]
                                                      
{'loss': 0.2333, 'grad_norm': 0.5364363789558411, 'learning_rate': 8.702981019467029e-06, 'epoch': 0.93}

 31%|███       | 4143/13317 [42:41<1:33:47,  1.63it/s]
 31%|███       | 4144/13317 [42:42<1:33:45,  1.63it/s]
                                                      
{'loss': 0.2182, 'grad_norm': 0.5309389233589172, 'learning_rate': 8.702100210466387e-06, 'epoch': 0.93}

 31%|███       | 4144/13317 [42:42<1:33:45,  1.63it/s]
 31%|███       | 4145/13317 [42:42<1:33:43,  1.63it/s]
                                                      
{'loss': 0.2928, 'grad_norm': 0.5576470494270325, 'learning_rate': 8.701219147092263e-06, 'epoch': 0.93}

 31%|███       | 4145/13317 [42:42<1:33:43,  1.63it/s]
 31%|███       | 4146/13317 [42:43<1:33:48,  1.63it/s]
                                                      
{'loss': 0.2005, 'grad_norm': 0.4506024718284607, 'learning_rate': 8.700337829405194e-06, 'epoch': 0.93}

 31%|███       | 4146/13317 [42:43<1:33:48,  1.63it/s]
 31%|███       | 4147/13317 [42:43<1:33:47,  1.63it/s]
                                                      
{'loss': 0.2687, 'grad_norm': 0.5049952268600464, 'learning_rate': 8.699456257465737e-06, 'epoch': 0.93}

 31%|███       | 4147/13317 [42:44<1:33:47,  1.63it/s]
 31%|███       | 4148/13317 [42:44<1:33:45,  1.63it/s]
                                                      
{'loss': 0.2997, 'grad_norm': 0.6373891234397888, 'learning_rate': 8.698574431334463e-06, 'epoch': 0.93}

 31%|███       | 4148/13317 [42:44<1:33:45,  1.63it/s]
 31%|███       | 4149/13317 [42:45<1:33:47,  1.63it/s]
                                                      
{'loss': 0.2064, 'grad_norm': 0.48347365856170654, 'learning_rate': 8.697692351071966e-06, 'epoch': 0.93}

 31%|███       | 4149/13317 [42:45<1:33:47,  1.63it/s]
 31%|███       | 4150/13317 [42:45<1:33:46,  1.63it/s]
                                                      
{'loss': 0.2896, 'grad_norm': 0.6032630205154419, 'learning_rate': 8.696810016738852e-06, 'epoch': 0.93}

 31%|███       | 4150/13317 [42:45<1:33:46,  1.63it/s]
 31%|███       | 4151/13317 [42:46<1:33:38,  1.63it/s]
                                                      
{'loss': 0.2911, 'grad_norm': 0.590092658996582, 'learning_rate': 8.695927428395749e-06, 'epoch': 0.94}

 31%|███       | 4151/13317 [42:46<1:33:38,  1.63it/s]
 31%|███       | 4152/13317 [42:47<1:33:42,  1.63it/s]
                                                      
{'loss': 0.2555, 'grad_norm': 0.4920317828655243, 'learning_rate': 8.695044586103297e-06, 'epoch': 0.94}

 31%|███       | 4152/13317 [42:47<1:33:42,  1.63it/s]
 31%|███       | 4153/13317 [42:47<1:33:40,  1.63it/s]
                                                      
{'loss': 0.2418, 'grad_norm': 0.5696350932121277, 'learning_rate': 8.694161489922156e-06, 'epoch': 0.94}

 31%|███       | 4153/13317 [42:47<1:33:40,  1.63it/s]
 31%|███       | 4154/13317 [42:48<1:33:39,  1.63it/s]
                                                      
{'loss': 0.2754, 'grad_norm': 0.5082764029502869, 'learning_rate': 8.69327813991301e-06, 'epoch': 0.94}

 31%|███       | 4154/13317 [42:48<1:33:39,  1.63it/s]
 31%|███       | 4155/13317 [42:48<1:33:40,  1.63it/s]
                                                      
{'loss': 0.2102, 'grad_norm': 0.5356817245483398, 'learning_rate': 8.69239453613655e-06, 'epoch': 0.94}

 31%|███       | 4155/13317 [42:48<1:33:40,  1.63it/s]
 31%|███       | 4156/13317 [42:49<1:33:39,  1.63it/s]
                                                      
{'loss': 0.2738, 'grad_norm': 0.5980523824691772, 'learning_rate': 8.691510678653489e-06, 'epoch': 0.94}

 31%|███       | 4156/13317 [42:49<1:33:39,  1.63it/s]
 31%|███       | 4157/13317 [42:50<1:33:40,  1.63it/s]
                                                      
{'loss': 0.2579, 'grad_norm': 0.49483418464660645, 'learning_rate': 8.690626567524559e-06, 'epoch': 0.94}

 31%|███       | 4157/13317 [42:50<1:33:40,  1.63it/s]
 31%|███       | 4158/13317 [42:50<1:33:42,  1.63it/s]
                                                      
{'loss': 0.2092, 'grad_norm': 0.44856834411621094, 'learning_rate': 8.689742202810505e-06, 'epoch': 0.94}

 31%|███       | 4158/13317 [42:50<1:33:42,  1.63it/s]
 31%|███       | 4159/13317 [42:51<1:33:40,  1.63it/s]
                                                      
{'loss': 0.3251, 'grad_norm': 0.640198290348053, 'learning_rate': 8.688857584572094e-06, 'epoch': 0.94}

 31%|███       | 4159/13317 [42:51<1:33:40,  1.63it/s]
 31%|███       | 4160/13317 [42:51<1:33:37,  1.63it/s]
                                                      
{'loss': 0.2253, 'grad_norm': 0.5414546728134155, 'learning_rate': 8.68797271287011e-06, 'epoch': 0.94}

 31%|███       | 4160/13317 [42:52<1:33:37,  1.63it/s]
 31%|███       | 4161/13317 [42:52<1:33:35,  1.63it/s]
                                                      
{'loss': 0.2832, 'grad_norm': 0.5204766988754272, 'learning_rate': 8.687087587765352e-06, 'epoch': 0.94}

 31%|███       | 4161/13317 [42:52<1:33:35,  1.63it/s]
 31%|███▏      | 4162/13317 [42:53<1:33:36,  1.63it/s]
                                                      
{'loss': 0.2228, 'grad_norm': 0.5769000053405762, 'learning_rate': 8.686202209318637e-06, 'epoch': 0.94}

 31%|███▏      | 4162/13317 [42:53<1:33:36,  1.63it/s]
 31%|███▏      | 4163/13317 [42:53<1:33:38,  1.63it/s]
                                                      
{'loss': 0.2824, 'grad_norm': 0.5717889666557312, 'learning_rate': 8.6853165775908e-06, 'epoch': 0.94}

 31%|███▏      | 4163/13317 [42:53<1:33:38,  1.63it/s]
 31%|███▏      | 4164/13317 [42:54<1:33:36,  1.63it/s]
                                                      
{'loss': 0.2583, 'grad_norm': 0.5422737002372742, 'learning_rate': 8.684430692642692e-06, 'epoch': 0.94}

 31%|███▏      | 4164/13317 [42:54<1:33:36,  1.63it/s]
 31%|███▏      | 4165/13317 [42:54<1:33:36,  1.63it/s]
                                                      
{'loss': 0.2658, 'grad_norm': 0.5364064574241638, 'learning_rate': 8.683544554535185e-06, 'epoch': 0.94}

 31%|███▏      | 4165/13317 [42:55<1:33:36,  1.63it/s]
 31%|███▏      | 4166/13317 [42:55<1:33:34,  1.63it/s]
                                                      
{'loss': 0.2323, 'grad_norm': 0.49883005023002625, 'learning_rate': 8.682658163329164e-06, 'epoch': 0.94}

 31%|███▏      | 4166/13317 [42:55<1:33:34,  1.63it/s]
 31%|███▏      | 4167/13317 [42:56<1:33:34,  1.63it/s]
                                                      
{'loss': 0.251, 'grad_norm': 0.5211261510848999, 'learning_rate': 8.681771519085539e-06, 'epoch': 0.94}

 31%|███▏      | 4167/13317 [42:56<1:33:34,  1.63it/s]
 31%|███▏      | 4168/13317 [42:56<1:33:36,  1.63it/s]
                                                      
{'loss': 0.322, 'grad_norm': 0.5859634876251221, 'learning_rate': 8.680884621865221e-06, 'epoch': 0.94}

 31%|███▏      | 4168/13317 [42:56<1:33:36,  1.63it/s]
 31%|███▏      | 4169/13317 [42:57<1:33:33,  1.63it/s]
                                                      
{'loss': 0.2009, 'grad_norm': 0.4449346959590912, 'learning_rate': 8.67999747172916e-06, 'epoch': 0.94}

 31%|███▏      | 4169/13317 [42:57<1:33:33,  1.63it/s]
 31%|███▏      | 4170/13317 [42:58<1:33:30,  1.63it/s]
                                                      
{'loss': 0.2322, 'grad_norm': 0.744380533695221, 'learning_rate': 8.679110068738306e-06, 'epoch': 0.94}

 31%|███▏      | 4170/13317 [42:58<1:33:30,  1.63it/s]
 31%|███▏      | 4171/13317 [42:58<1:33:34,  1.63it/s]
                                                      
{'loss': 0.239, 'grad_norm': 0.48880910873413086, 'learning_rate': 8.678222412953637e-06, 'epoch': 0.94}

 31%|███▏      | 4171/13317 [42:58<1:33:34,  1.63it/s]
 31%|███▏      | 4172/13317 [42:59<1:33:33,  1.63it/s]
                                                      
{'loss': 0.2699, 'grad_norm': 0.5874391198158264, 'learning_rate': 8.677334504436142e-06, 'epoch': 0.94}

 31%|███▏      | 4172/13317 [42:59<1:33:33,  1.63it/s]
 31%|███▏      | 4173/13317 [42:59<1:33:29,  1.63it/s]
                                                      
{'loss': 0.2342, 'grad_norm': 0.5163384675979614, 'learning_rate': 8.676446343246832e-06, 'epoch': 0.94}

 31%|███▏      | 4173/13317 [43:00<1:33:29,  1.63it/s]
 31%|███▏      | 4174/13317 [43:00<1:33:29,  1.63it/s]
                                                      
{'loss': 0.2064, 'grad_norm': 0.46332740783691406, 'learning_rate': 8.675557929446728e-06, 'epoch': 0.94}

 31%|███▏      | 4174/13317 [43:00<1:33:29,  1.63it/s]
 31%|███▏      | 4175/13317 [43:01<1:33:27,  1.63it/s]
                                                      
{'loss': 0.2316, 'grad_norm': 0.4855860471725464, 'learning_rate': 8.674669263096879e-06, 'epoch': 0.94}

 31%|███▏      | 4175/13317 [43:01<1:33:27,  1.63it/s]
 31%|███▏      | 4176/13317 [43:01<1:33:24,  1.63it/s]
                                                      
{'loss': 0.2676, 'grad_norm': 0.6231251955032349, 'learning_rate': 8.673780344258342e-06, 'epoch': 0.94}

 31%|███▏      | 4176/13317 [43:01<1:33:24,  1.63it/s]
 31%|███▏      | 4177/13317 [43:02<1:33:25,  1.63it/s]
                                                      
{'loss': 0.2745, 'grad_norm': 0.5614831447601318, 'learning_rate': 8.672891172992198e-06, 'epoch': 0.94}

 31%|███▏      | 4177/13317 [43:02<1:33:25,  1.63it/s]
 31%|███▏      | 4178/13317 [43:02<1:33:25,  1.63it/s]
                                                      
{'loss': 0.1974, 'grad_norm': 0.484260231256485, 'learning_rate': 8.672001749359542e-06, 'epoch': 0.94}

 31%|███▏      | 4178/13317 [43:03<1:33:25,  1.63it/s]
 31%|███▏      | 4179/13317 [43:03<1:33:21,  1.63it/s]
                                                      
{'loss': 0.231, 'grad_norm': 0.49738427996635437, 'learning_rate': 8.671112073421484e-06, 'epoch': 0.94}

 31%|███▏      | 4179/13317 [43:03<1:33:21,  1.63it/s]
 31%|███▏      | 4180/13317 [43:04<1:33:24,  1.63it/s]
                                                      
{'loss': 0.2303, 'grad_norm': 0.5214279294013977, 'learning_rate': 8.670222145239158e-06, 'epoch': 0.94}

 31%|███▏      | 4180/13317 [43:04<1:33:24,  1.63it/s]
 31%|███▏      | 4181/13317 [43:04<1:33:22,  1.63it/s]
                                                      
{'loss': 0.1866, 'grad_norm': 0.4330577850341797, 'learning_rate': 8.669331964873709e-06, 'epoch': 0.94}

 31%|███▏      | 4181/13317 [43:04<1:33:22,  1.63it/s]
 31%|███▏      | 4182/13317 [43:05<1:33:21,  1.63it/s]
                                                      
{'loss': 0.2167, 'grad_norm': 0.4694700837135315, 'learning_rate': 8.668441532386303e-06, 'epoch': 0.94}

 31%|███▏      | 4182/13317 [43:05<1:33:21,  1.63it/s]
 31%|███▏      | 4183/13317 [43:06<1:33:20,  1.63it/s]
                                                      
{'loss': 0.23, 'grad_norm': 0.5059525966644287, 'learning_rate': 8.66755084783812e-06, 'epoch': 0.94}

 31%|███▏      | 4183/13317 [43:06<1:33:20,  1.63it/s]
 31%|███▏      | 4184/13317 [43:06<1:33:20,  1.63it/s]
                                                      
{'loss': 0.3223, 'grad_norm': 0.6607219576835632, 'learning_rate': 8.666659911290363e-06, 'epoch': 0.94}

 31%|███▏      | 4184/13317 [43:06<1:33:20,  1.63it/s]
 31%|███▏      | 4185/13317 [43:07<1:33:18,  1.63it/s]
                                                      
{'loss': 0.2586, 'grad_norm': 0.5053427219390869, 'learning_rate': 8.665768722804244e-06, 'epoch': 0.94}

 31%|███▏      | 4185/13317 [43:07<1:33:18,  1.63it/s]
 31%|███▏      | 4186/13317 [43:07<1:33:18,  1.63it/s]
                                                      
{'loss': 0.2012, 'grad_norm': 0.48763853311538696, 'learning_rate': 8.664877282441003e-06, 'epoch': 0.94}

 31%|███▏      | 4186/13317 [43:08<1:33:18,  1.63it/s]
 31%|███▏      | 4187/13317 [43:08<1:33:20,  1.63it/s]
                                                      
{'loss': 0.2929, 'grad_norm': 0.5259266495704651, 'learning_rate': 8.663985590261888e-06, 'epoch': 0.94}

 31%|███▏      | 4187/13317 [43:08<1:33:20,  1.63it/s]
 31%|███▏      | 4188/13317 [43:09<1:33:18,  1.63it/s]
                                                      
{'loss': 0.2521, 'grad_norm': 0.5589292049407959, 'learning_rate': 8.663093646328166e-06, 'epoch': 0.94}

 31%|███▏      | 4188/13317 [43:09<1:33:18,  1.63it/s]
 31%|███▏      | 4189/13317 [43:09<1:33:18,  1.63it/s]
                                                      
{'loss': 0.2076, 'grad_norm': 0.48135942220687866, 'learning_rate': 8.662201450701126e-06, 'epoch': 0.94}

 31%|███▏      | 4189/13317 [43:09<1:33:18,  1.63it/s]
 31%|███▏      | 4190/13317 [43:10<1:33:20,  1.63it/s]
                                                      
{'loss': 0.2611, 'grad_norm': 0.5443137288093567, 'learning_rate': 8.661309003442072e-06, 'epoch': 0.94}

 31%|███▏      | 4190/13317 [43:10<1:33:20,  1.63it/s]
 31%|███▏      | 4191/13317 [43:10<1:33:16,  1.63it/s]
                                                      
{'loss': 0.2288, 'grad_norm': 0.5488364100456238, 'learning_rate': 8.660416304612323e-06, 'epoch': 0.94}

 31%|███▏      | 4191/13317 [43:11<1:33:16,  1.63it/s]
 31%|███▏      | 4192/13317 [43:11<1:33:18,  1.63it/s]
                                                      
{'loss': 0.2727, 'grad_norm': 0.5474535226821899, 'learning_rate': 8.659523354273214e-06, 'epoch': 0.94}

 31%|███▏      | 4192/13317 [43:11<1:33:18,  1.63it/s]
 31%|███▏      | 4193/13317 [43:12<1:33:14,  1.63it/s]
                                                      
{'loss': 0.2682, 'grad_norm': 0.6569675803184509, 'learning_rate': 8.658630152486104e-06, 'epoch': 0.94}

 31%|███▏      | 4193/13317 [43:12<1:33:14,  1.63it/s]
 31%|███▏      | 4194/13317 [43:12<1:33:13,  1.63it/s]
                                                      
{'loss': 0.2465, 'grad_norm': 0.45298701524734497, 'learning_rate': 8.657736699312365e-06, 'epoch': 0.94}

 31%|███▏      | 4194/13317 [43:12<1:33:13,  1.63it/s]
 32%|███▏      | 4195/13317 [43:13<1:33:12,  1.63it/s]
                                                      
{'loss': 0.3865, 'grad_norm': 0.6103184223175049, 'learning_rate': 8.656842994813386e-06, 'epoch': 0.95}

 32%|███▏      | 4195/13317 [43:13<1:33:12,  1.63it/s]
 32%|███▏      | 4196/13317 [43:14<1:33:13,  1.63it/s]
                                                      
{'loss': 0.2437, 'grad_norm': 0.5519054532051086, 'learning_rate': 8.655949039050572e-06, 'epoch': 0.95}

 32%|███▏      | 4196/13317 [43:14<1:33:13,  1.63it/s]
 32%|███▏      | 4197/13317 [43:14<1:33:12,  1.63it/s]
                                                      
{'loss': 0.2615, 'grad_norm': 0.572892963886261, 'learning_rate': 8.655054832085352e-06, 'epoch': 0.95}

 32%|███▏      | 4197/13317 [43:14<1:33:12,  1.63it/s]
 32%|███▏      | 4198/13317 [43:15<1:33:15,  1.63it/s]
                                                      
{'loss': 0.2373, 'grad_norm': 0.46048590540885925, 'learning_rate': 8.654160373979163e-06, 'epoch': 0.95}

 32%|███▏      | 4198/13317 [43:15<1:33:15,  1.63it/s]
 32%|███▏      | 4199/13317 [43:15<1:33:14,  1.63it/s]
                                                      
{'loss': 0.3038, 'grad_norm': 0.6488249897956848, 'learning_rate': 8.653265664793466e-06, 'epoch': 0.95}

 32%|███▏      | 4199/13317 [43:15<1:33:14,  1.63it/s]
 32%|███▏      | 4200/13317 [43:16<1:33:12,  1.63it/s]
                                                      
{'loss': 0.2744, 'grad_norm': 0.5547277927398682, 'learning_rate': 8.652370704589736e-06, 'epoch': 0.95}

 32%|███▏      | 4200/13317 [43:16<1:33:12,  1.63it/s]
 32%|███▏      | 4201/13317 [43:17<1:33:12,  1.63it/s]
                                                      
{'loss': 0.2978, 'grad_norm': 0.562756359577179, 'learning_rate': 8.651475493429469e-06, 'epoch': 0.95}

 32%|███▏      | 4201/13317 [43:17<1:33:12,  1.63it/s]
 32%|███▏      | 4202/13317 [43:17<1:33:14,  1.63it/s]
                                                      
{'loss': 0.2471, 'grad_norm': 0.5226801037788391, 'learning_rate': 8.65058003137417e-06, 'epoch': 0.95}

 32%|███▏      | 4202/13317 [43:17<1:33:14,  1.63it/s]
 32%|███▏      | 4203/13317 [43:18<1:33:12,  1.63it/s]
                                                      
{'loss': 0.3178, 'grad_norm': 0.5759134888648987, 'learning_rate': 8.649684318485372e-06, 'epoch': 0.95}

 32%|███▏      | 4203/13317 [43:18<1:33:12,  1.63it/s]
 32%|███▏      | 4204/13317 [43:18<1:33:09,  1.63it/s]
                                                      
{'loss': 0.246, 'grad_norm': 0.5488865375518799, 'learning_rate': 8.648788354824617e-06, 'epoch': 0.95}

 32%|███▏      | 4204/13317 [43:19<1:33:09,  1.63it/s]
 32%|███▏      | 4205/13317 [43:19<1:33:10,  1.63it/s]
                                                      
{'loss': 0.2645, 'grad_norm': 0.5084179639816284, 'learning_rate': 8.647892140453467e-06, 'epoch': 0.95}

 32%|███▏      | 4205/13317 [43:19<1:33:10,  1.63it/s]
 32%|███▏      | 4206/13317 [43:20<1:33:08,  1.63it/s]
                                                      
{'loss': 0.3084, 'grad_norm': 0.5774059891700745, 'learning_rate': 8.646995675433503e-06, 'epoch': 0.95}

 32%|███▏      | 4206/13317 [43:20<1:33:08,  1.63it/s]
 32%|███▏      | 4207/13317 [43:20<1:33:11,  1.63it/s]
                                                      
{'loss': 0.2277, 'grad_norm': 0.4769032299518585, 'learning_rate': 8.646098959826322e-06, 'epoch': 0.95}

 32%|███▏      | 4207/13317 [43:20<1:33:11,  1.63it/s]
 32%|███▏      | 4208/13317 [43:21<1:33:10,  1.63it/s]
                                                      
{'loss': 0.2553, 'grad_norm': 0.5222684741020203, 'learning_rate': 8.645201993693537e-06, 'epoch': 0.95}

 32%|███▏      | 4208/13317 [43:21<1:33:10,  1.63it/s]
 32%|███▏      | 4209/13317 [43:21<1:33:09,  1.63it/s]
                                                      
{'loss': 0.3049, 'grad_norm': 0.6217694878578186, 'learning_rate': 8.64430477709678e-06, 'epoch': 0.95}

 32%|███▏      | 4209/13317 [43:22<1:33:09,  1.63it/s]
 32%|███▏      | 4210/13317 [43:22<1:33:07,  1.63it/s]
                                                      
{'loss': 0.2077, 'grad_norm': 0.4721110463142395, 'learning_rate': 8.643407310097695e-06, 'epoch': 0.95}

 32%|███▏      | 4210/13317 [43:22<1:33:07,  1.63it/s]
 32%|███▏      | 4211/13317 [43:23<1:33:05,  1.63it/s]
                                                      
{'loss': 0.1947, 'grad_norm': 0.46641796827316284, 'learning_rate': 8.642509592757953e-06, 'epoch': 0.95}

 32%|███▏      | 4211/13317 [43:23<1:33:05,  1.63it/s]
 32%|███▏      | 4212/13317 [43:23<1:33:06,  1.63it/s]
                                                      
{'loss': 0.2082, 'grad_norm': 0.49415504932403564, 'learning_rate': 8.641611625139235e-06, 'epoch': 0.95}

 32%|███▏      | 4212/13317 [43:23<1:33:06,  1.63it/s]
 32%|███▏      | 4213/13317 [43:24<1:33:05,  1.63it/s]
                                                      
{'loss': 0.2067, 'grad_norm': 0.4996008276939392, 'learning_rate': 8.640713407303239e-06, 'epoch': 0.95}

 32%|███▏      | 4213/13317 [43:24<1:33:05,  1.63it/s]
 32%|███▏      | 4214/13317 [43:25<1:33:03,  1.63it/s]
                                                      
{'loss': 0.2059, 'grad_norm': 0.4396611154079437, 'learning_rate': 8.639814939311684e-06, 'epoch': 0.95}

 32%|███▏      | 4214/13317 [43:25<1:33:03,  1.63it/s]
 32%|███▏      | 4215/13317 [43:25<1:33:03,  1.63it/s]
                                                      
{'loss': 0.2023, 'grad_norm': 0.5960217118263245, 'learning_rate': 8.638916221226303e-06, 'epoch': 0.95}

 32%|███▏      | 4215/13317 [43:25<1:33:03,  1.63it/s]
 32%|███▏      | 4216/13317 [43:26<1:33:01,  1.63it/s]
                                                      
{'loss': 0.2408, 'grad_norm': 0.4748380780220032, 'learning_rate': 8.638017253108848e-06, 'epoch': 0.95}

 32%|███▏      | 4216/13317 [43:26<1:33:01,  1.63it/s]
 32%|███▏      | 4217/13317 [43:26<1:33:01,  1.63it/s]
                                                      
{'loss': 0.22, 'grad_norm': 0.49779242277145386, 'learning_rate': 8.637118035021089e-06, 'epoch': 0.95}

 32%|███▏      | 4217/13317 [43:27<1:33:01,  1.63it/s]
 32%|███▏      | 4218/13317 [43:27<1:33:06,  1.63it/s]
                                                      
{'loss': 0.2005, 'grad_norm': 0.4698963761329651, 'learning_rate': 8.63621856702481e-06, 'epoch': 0.95}

 32%|███▏      | 4218/13317 [43:27<1:33:06,  1.63it/s]
 32%|███▏      | 4219/13317 [43:28<1:33:03,  1.63it/s]
                                                      
{'loss': 0.2266, 'grad_norm': 0.48179998993873596, 'learning_rate': 8.635318849181815e-06, 'epoch': 0.95}

 32%|███▏      | 4219/13317 [43:28<1:33:03,  1.63it/s]
 32%|███▏      | 4220/13317 [43:28<1:33:01,  1.63it/s]
                                                      
{'loss': 0.2921, 'grad_norm': 0.5564152002334595, 'learning_rate': 8.634418881553923e-06, 'epoch': 0.95}

 32%|███▏      | 4220/13317 [43:28<1:33:01,  1.63it/s]
 32%|███▏      | 4221/13317 [43:29<1:33:01,  1.63it/s]
                                                      
{'loss': 0.2351, 'grad_norm': 0.6168543100357056, 'learning_rate': 8.633518664202971e-06, 'epoch': 0.95}

 32%|███▏      | 4221/13317 [43:29<1:33:01,  1.63it/s]
 32%|███▏      | 4222/13317 [43:29<1:32:59,  1.63it/s]
                                                      
{'loss': 0.2113, 'grad_norm': 0.5157584547996521, 'learning_rate': 8.632618197190817e-06, 'epoch': 0.95}

 32%|███▏      | 4222/13317 [43:30<1:32:59,  1.63it/s]
 32%|███▏      | 4223/13317 [43:30<1:32:59,  1.63it/s]
                                                      
{'loss': 0.3349, 'grad_norm': 0.649715006351471, 'learning_rate': 8.631717480579329e-06, 'epoch': 0.95}

 32%|███▏      | 4223/13317 [43:30<1:32:59,  1.63it/s]
 32%|███▏      | 4224/13317 [43:31<1:32:59,  1.63it/s]
                                                      
{'loss': 0.2484, 'grad_norm': 0.6179946064949036, 'learning_rate': 8.630816514430396e-06, 'epoch': 0.95}

 32%|███▏      | 4224/13317 [43:31<1:32:59,  1.63it/s]
 32%|███▏      | 4225/13317 [43:31<1:32:57,  1.63it/s]
                                                      
{'loss': 0.2282, 'grad_norm': 0.4938258230686188, 'learning_rate': 8.629915298805925e-06, 'epoch': 0.95}

 32%|███▏      | 4225/13317 [43:31<1:32:57,  1.63it/s]
 32%|███▏      | 4226/13317 [43:32<1:32:54,  1.63it/s]
                                                      
{'loss': 0.202, 'grad_norm': 0.44291403889656067, 'learning_rate': 8.62901383376784e-06, 'epoch': 0.95}

 32%|███▏      | 4226/13317 [43:32<1:32:54,  1.63it/s]
 32%|███▏      | 4227/13317 [43:33<1:32:57,  1.63it/s]
                                                      
{'loss': 0.2047, 'grad_norm': 0.4768671989440918, 'learning_rate': 8.628112119378078e-06, 'epoch': 0.95}

 32%|███▏      | 4227/13317 [43:33<1:32:57,  1.63it/s]
 32%|███▏      | 4228/13317 [43:33<1:32:55,  1.63it/s]
                                                      
{'loss': 0.2141, 'grad_norm': 0.4680904746055603, 'learning_rate': 8.6272101556986e-06, 'epoch': 0.95}

 32%|███▏      | 4228/13317 [43:33<1:32:55,  1.63it/s]
 32%|███▏      | 4229/13317 [43:34<1:32:54,  1.63it/s]
                                                      
{'loss': 0.2264, 'grad_norm': 0.549738883972168, 'learning_rate': 8.626307942791379e-06, 'epoch': 0.95}

 32%|███▏      | 4229/13317 [43:34<1:32:54,  1.63it/s]
 32%|███▏      | 4230/13317 [43:34<1:32:57,  1.63it/s]
                                                      
{'loss': 0.2374, 'grad_norm': 0.4806957244873047, 'learning_rate': 8.625405480718405e-06, 'epoch': 0.95}

 32%|███▏      | 4230/13317 [43:35<1:32:57,  1.63it/s]
 32%|███▏      | 4231/13317 [43:35<1:32:53,  1.63it/s]
                                                      
{'loss': 0.2116, 'grad_norm': 0.47647741436958313, 'learning_rate': 8.62450276954169e-06, 'epoch': 0.95}

 32%|███▏      | 4231/13317 [43:35<1:32:53,  1.63it/s]
 32%|███▏      | 4232/13317 [43:36<1:32:53,  1.63it/s]
                                                      
{'loss': 0.2356, 'grad_norm': 0.4727253317832947, 'learning_rate': 8.623599809323256e-06, 'epoch': 0.95}

 32%|███▏      | 4232/13317 [43:36<1:32:53,  1.63it/s]
 32%|███▏      | 4233/13317 [43:36<1:32:50,  1.63it/s]
                                                      
{'loss': 0.2986, 'grad_norm': 0.634233295917511, 'learning_rate': 8.622696600125149e-06, 'epoch': 0.95}

 32%|███▏      | 4233/13317 [43:36<1:32:50,  1.63it/s]
 32%|███▏      | 4234/13317 [43:37<1:32:49,  1.63it/s]
                                                      
{'loss': 0.2174, 'grad_norm': 0.43639156222343445, 'learning_rate': 8.621793142009428e-06, 'epoch': 0.95}

 32%|███▏      | 4234/13317 [43:37<1:32:49,  1.63it/s]
 32%|███▏      | 4235/13317 [43:37<1:32:49,  1.63it/s]
                                                      
{'loss': 0.2549, 'grad_norm': 0.5707977414131165, 'learning_rate': 8.620889435038167e-06, 'epoch': 0.95}

 32%|███▏      | 4235/13317 [43:38<1:32:49,  1.63it/s]
 32%|███▏      | 4236/13317 [43:38<1:32:47,  1.63it/s]
                                                      
{'loss': 0.2752, 'grad_norm': 0.5175776481628418, 'learning_rate': 8.619985479273465e-06, 'epoch': 0.95}

 32%|███▏      | 4236/13317 [43:38<1:32:47,  1.63it/s]
 32%|███▏      | 4237/13317 [43:39<1:32:47,  1.63it/s]
                                                      
{'loss': 0.2244, 'grad_norm': 0.5138694643974304, 'learning_rate': 8.619081274777434e-06, 'epoch': 0.95}

 32%|███▏      | 4237/13317 [43:39<1:32:47,  1.63it/s]
 32%|███▏      | 4238/13317 [43:39<1:32:47,  1.63it/s]
                                                      
{'loss': 0.2414, 'grad_norm': 0.5383557081222534, 'learning_rate': 8.618176821612197e-06, 'epoch': 0.95}

 32%|███▏      | 4238/13317 [43:39<1:32:47,  1.63it/s]
 32%|███▏      | 4239/13317 [43:40<1:32:48,  1.63it/s]
                                                      
{'loss': 0.1607, 'grad_norm': 0.44410040974617004, 'learning_rate': 8.617272119839903e-06, 'epoch': 0.95}

 32%|███▏      | 4239/13317 [43:40<1:32:48,  1.63it/s]
 32%|███▏      | 4240/13317 [43:41<1:32:49,  1.63it/s]
                                                      
{'loss': 0.2287, 'grad_norm': 0.5039393305778503, 'learning_rate': 8.616367169522715e-06, 'epoch': 0.96}

 32%|███▏      | 4240/13317 [43:41<1:32:49,  1.63it/s]
 32%|███▏      | 4241/13317 [43:41<1:32:48,  1.63it/s]
                                                      
{'loss': 0.2063, 'grad_norm': 0.5065698027610779, 'learning_rate': 8.615461970722811e-06, 'epoch': 0.96}

 32%|███▏      | 4241/13317 [43:41<1:32:48,  1.63it/s]
 32%|███▏      | 4242/13317 [43:42<1:32:50,  1.63it/s]
                                                      
{'loss': 0.3451, 'grad_norm': 0.6150543689727783, 'learning_rate': 8.61455652350239e-06, 'epoch': 0.96}

 32%|███▏      | 4242/13317 [43:42<1:32:50,  1.63it/s]
 32%|███▏      | 4243/13317 [43:42<1:32:50,  1.63it/s]
                                                      
{'loss': 0.2563, 'grad_norm': 0.5133724212646484, 'learning_rate': 8.613650827923662e-06, 'epoch': 0.96}

 32%|███▏      | 4243/13317 [43:42<1:32:50,  1.63it/s]
 32%|███▏      | 4244/13317 [43:43<1:32:49,  1.63it/s]
                                                      
{'loss': 0.2378, 'grad_norm': 0.49474602937698364, 'learning_rate': 8.612744884048861e-06, 'epoch': 0.96}

 32%|███▏      | 4244/13317 [43:43<1:32:49,  1.63it/s]
 32%|███▏      | 4245/13317 [43:44<1:32:48,  1.63it/s]
                                                      
{'loss': 0.2416, 'grad_norm': 0.5118690133094788, 'learning_rate': 8.611838691940236e-06, 'epoch': 0.96}

 32%|███▏      | 4245/13317 [43:44<1:32:48,  1.63it/s]
 32%|███▏      | 4246/13317 [43:44<1:32:45,  1.63it/s]
                                                      
{'loss': 0.1981, 'grad_norm': 0.45892274379730225, 'learning_rate': 8.610932251660046e-06, 'epoch': 0.96}

 32%|███▏      | 4246/13317 [43:44<1:32:45,  1.63it/s]
 32%|███▏      | 4247/13317 [43:45<1:38:02,  1.54it/s]
                                                      
{'loss': 0.2317, 'grad_norm': 0.47518208622932434, 'learning_rate': 8.610025563270582e-06, 'epoch': 0.96}

 32%|███▏      | 4247/13317 [43:45<1:38:02,  1.54it/s]
 32%|███▏      | 4248/13317 [43:46<1:36:25,  1.57it/s]
                                                      
{'loss': 0.2366, 'grad_norm': 0.5033785700798035, 'learning_rate': 8.609118626834135e-06, 'epoch': 0.96}

 32%|███▏      | 4248/13317 [43:46<1:36:25,  1.57it/s]
 32%|███▏      | 4249/13317 [43:46<1:35:19,  1.59it/s]
                                                      
{'loss': 0.3454, 'grad_norm': 0.7162356972694397, 'learning_rate': 8.608211442413027e-06, 'epoch': 0.96}

 32%|███▏      | 4249/13317 [43:46<1:35:19,  1.59it/s]
 32%|███▏      | 4250/13317 [43:47<1:34:38,  1.60it/s]
                                                      
{'loss': 0.2288, 'grad_norm': 0.487576961517334, 'learning_rate': 8.607304010069586e-06, 'epoch': 0.96}

 32%|███▏      | 4250/13317 [43:47<1:34:38,  1.60it/s]
 32%|███▏      | 4251/13317 [43:47<1:34:02,  1.61it/s]
                                                      
{'loss': 0.2157, 'grad_norm': 0.4644017517566681, 'learning_rate': 8.606396329866165e-06, 'epoch': 0.96}

 32%|███▏      | 4251/13317 [43:48<1:34:02,  1.61it/s]
 32%|███▏      | 4252/13317 [43:48<1:33:45,  1.61it/s]
                                                      
{'loss': 0.2068, 'grad_norm': 0.4368003010749817, 'learning_rate': 8.60548840186513e-06, 'epoch': 0.96}

 32%|███▏      | 4252/13317 [43:48<1:33:45,  1.61it/s]
 32%|███▏      | 4253/13317 [43:49<1:33:26,  1.62it/s]
                                                      
{'loss': 0.2112, 'grad_norm': 0.5834970474243164, 'learning_rate': 8.60458022612887e-06, 'epoch': 0.96}

 32%|███▏      | 4253/13317 [43:49<1:33:26,  1.62it/s]
 32%|███▏      | 4254/13317 [43:49<1:33:09,  1.62it/s]
                                                      
{'loss': 0.2299, 'grad_norm': 0.47473061084747314, 'learning_rate': 8.60367180271978e-06, 'epoch': 0.96}

 32%|███▏      | 4254/13317 [43:49<1:33:09,  1.62it/s]
 32%|███▏      | 4255/13317 [43:50<1:32:59,  1.62it/s]
                                                      
{'loss': 0.2743, 'grad_norm': 0.5893845558166504, 'learning_rate': 8.60276313170028e-06, 'epoch': 0.96}

 32%|███▏      | 4255/13317 [43:50<1:32:59,  1.62it/s]
 32%|███▏      | 4256/13317 [43:50<1:32:49,  1.63it/s]
                                                      
{'loss': 0.2985, 'grad_norm': 0.551262617111206, 'learning_rate': 8.601854213132808e-06, 'epoch': 0.96}

 32%|███▏      | 4256/13317 [43:51<1:32:49,  1.63it/s]
 32%|███▏      | 4257/13317 [43:51<1:32:47,  1.63it/s]
                                                      
{'loss': 0.2575, 'grad_norm': 0.5677189230918884, 'learning_rate': 8.600945047079813e-06, 'epoch': 0.96}

 32%|███▏      | 4257/13317 [43:51<1:32:47,  1.63it/s]
 32%|███▏      | 4258/13317 [43:52<1:32:44,  1.63it/s]
                                                      
{'loss': 0.1862, 'grad_norm': 0.4721943736076355, 'learning_rate': 8.600035633603767e-06, 'epoch': 0.96}

 32%|███▏      | 4258/13317 [43:52<1:32:44,  1.63it/s]
 32%|███▏      | 4259/13317 [43:52<1:32:41,  1.63it/s]
                                                      
{'loss': 0.2329, 'grad_norm': 0.449490487575531, 'learning_rate': 8.599125972767155e-06, 'epoch': 0.96}

 32%|███▏      | 4259/13317 [43:52<1:32:41,  1.63it/s]
 32%|███▏      | 4260/13317 [43:53<1:32:47,  1.63it/s]
                                                      
{'loss': 0.2222, 'grad_norm': 0.4835489094257355, 'learning_rate': 8.59821606463248e-06, 'epoch': 0.96}

 32%|███▏      | 4260/13317 [43:53<1:32:47,  1.63it/s]
 32%|███▏      | 4261/13317 [43:54<1:32:43,  1.63it/s]
                                                      
{'loss': 0.2305, 'grad_norm': 0.5966092944145203, 'learning_rate': 8.597305909262264e-06, 'epoch': 0.96}

 32%|███▏      | 4261/13317 [43:54<1:32:43,  1.63it/s]
 32%|███▏      | 4262/13317 [43:54<1:32:43,  1.63it/s]
                                                      
{'loss': 0.2731, 'grad_norm': 0.5575417280197144, 'learning_rate': 8.596395506719044e-06, 'epoch': 0.96}

 32%|███▏      | 4262/13317 [43:54<1:32:43,  1.63it/s]
 32%|███▏      | 4263/13317 [43:55<1:32:41,  1.63it/s]
                                                      
{'loss': 0.2635, 'grad_norm': 0.5200009346008301, 'learning_rate': 8.595484857065372e-06, 'epoch': 0.96}

 32%|███▏      | 4263/13317 [43:55<1:32:41,  1.63it/s]
 32%|███▏      | 4264/13317 [43:55<1:32:42,  1.63it/s]
                                                      
{'loss': 0.2809, 'grad_norm': 0.579536497592926, 'learning_rate': 8.59457396036382e-06, 'epoch': 0.96}

 32%|███▏      | 4264/13317 [43:55<1:32:42,  1.63it/s]
 32%|███▏      | 4265/13317 [43:56<1:32:44,  1.63it/s]
                                                      
{'loss': 0.2875, 'grad_norm': 0.5992927551269531, 'learning_rate': 8.593662816676977e-06, 'epoch': 0.96}

 32%|███▏      | 4265/13317 [43:56<1:32:44,  1.63it/s]
 32%|███▏      | 4266/13317 [43:57<1:32:42,  1.63it/s]
                                                      
{'loss': 0.2057, 'grad_norm': 0.5394103527069092, 'learning_rate': 8.59275142606745e-06, 'epoch': 0.96}

 32%|███▏      | 4266/13317 [43:57<1:32:42,  1.63it/s]
 32%|███▏      | 4267/13317 [43:57<1:32:47,  1.63it/s]
                                                      
{'loss': 0.1972, 'grad_norm': 0.44549381732940674, 'learning_rate': 8.591839788597858e-06, 'epoch': 0.96}

 32%|███▏      | 4267/13317 [43:57<1:32:47,  1.63it/s]
 32%|███▏      | 4268/13317 [43:58<1:32:42,  1.63it/s]
                                                      
{'loss': 0.2705, 'grad_norm': 0.5746995210647583, 'learning_rate': 8.590927904330842e-06, 'epoch': 0.96}

 32%|███▏      | 4268/13317 [43:58<1:32:42,  1.63it/s]
 32%|███▏      | 4269/13317 [43:58<1:32:39,  1.63it/s]
                                                      
{'loss': 0.3449, 'grad_norm': 0.5990719795227051, 'learning_rate': 8.590015773329056e-06, 'epoch': 0.96}

 32%|███▏      | 4269/13317 [43:59<1:32:39,  1.63it/s]
 32%|███▏      | 4270/13317 [43:59<1:32:43,  1.63it/s]
                                                      
{'loss': 0.2418, 'grad_norm': 0.5077854990959167, 'learning_rate': 8.589103395655177e-06, 'epoch': 0.96}

 32%|███▏      | 4270/13317 [43:59<1:32:43,  1.63it/s]
 32%|███▏      | 4271/13317 [44:00<1:32:40,  1.63it/s]
                                                      
{'loss': 0.1556, 'grad_norm': 0.3800909221172333, 'learning_rate': 8.588190771371891e-06, 'epoch': 0.96}

 32%|███▏      | 4271/13317 [44:00<1:32:40,  1.63it/s]
 32%|███▏      | 4272/13317 [44:00<1:32:45,  1.63it/s]
                                                      
{'loss': 0.252, 'grad_norm': 0.5713374018669128, 'learning_rate': 8.587277900541908e-06, 'epoch': 0.96}

 32%|███▏      | 4272/13317 [44:00<1:32:45,  1.63it/s]
 32%|███▏      | 4273/13317 [44:01<1:32:43,  1.63it/s]
                                                      
{'loss': 0.2145, 'grad_norm': 0.4492753744125366, 'learning_rate': 8.586364783227949e-06, 'epoch': 0.96}

 32%|███▏      | 4273/13317 [44:01<1:32:43,  1.63it/s]
 32%|███▏      | 4274/13317 [44:02<1:32:38,  1.63it/s]
                                                      
{'loss': 0.1833, 'grad_norm': 0.4851554036140442, 'learning_rate': 8.585451419492757e-06, 'epoch': 0.96}

 32%|███▏      | 4274/13317 [44:02<1:32:38,  1.63it/s]
 32%|███▏      | 4275/13317 [44:02<1:32:38,  1.63it/s]
                                                      
{'loss': 0.307, 'grad_norm': 0.5392456650733948, 'learning_rate': 8.584537809399089e-06, 'epoch': 0.96}

 32%|███▏      | 4275/13317 [44:02<1:32:38,  1.63it/s]
 32%|███▏      | 4276/13317 [44:03<1:32:34,  1.63it/s]
                                                      
{'loss': 0.2191, 'grad_norm': 0.4551559090614319, 'learning_rate': 8.583623953009722e-06, 'epoch': 0.96}

 32%|███▏      | 4276/13317 [44:03<1:32:34,  1.63it/s]
 32%|███▏      | 4277/13317 [44:03<1:32:39,  1.63it/s]
                                                      
{'loss': 0.2539, 'grad_norm': 0.5221605896949768, 'learning_rate': 8.582709850387441e-06, 'epoch': 0.96}

 32%|███▏      | 4277/13317 [44:03<1:32:39,  1.63it/s]
 32%|███▏      | 4278/13317 [44:04<1:32:33,  1.63it/s]
                                                      
{'loss': 0.206, 'grad_norm': 0.4275699555873871, 'learning_rate': 8.581795501595062e-06, 'epoch': 0.96}

 32%|███▏      | 4278/13317 [44:04<1:32:33,  1.63it/s]
 32%|███▏      | 4279/13317 [44:05<1:32:35,  1.63it/s]
                                                      
{'loss': 0.2096, 'grad_norm': 0.47436878085136414, 'learning_rate': 8.580880906695407e-06, 'epoch': 0.96}

 32%|███▏      | 4279/13317 [44:05<1:32:35,  1.63it/s]
 32%|███▏      | 4280/13317 [44:05<1:32:37,  1.63it/s]
                                                      
{'loss': 0.2703, 'grad_norm': 0.5225781798362732, 'learning_rate': 8.579966065751318e-06, 'epoch': 0.96}

 32%|███▏      | 4280/13317 [44:05<1:32:37,  1.63it/s]
 32%|███▏      | 4281/13317 [44:06<1:32:29,  1.63it/s]
                                                      
{'loss': 0.299, 'grad_norm': 0.5980818867683411, 'learning_rate': 8.579050978825657e-06, 'epoch': 0.96}

 32%|███▏      | 4281/13317 [44:06<1:32:29,  1.63it/s]
 32%|███▏      | 4282/13317 [44:06<1:32:37,  1.63it/s]
                                                      
{'loss': 0.2213, 'grad_norm': 0.4718390107154846, 'learning_rate': 8.578135645981296e-06, 'epoch': 0.96}

 32%|███▏      | 4282/13317 [44:07<1:32:37,  1.63it/s]
 32%|███▏      | 4283/13317 [44:07<1:32:34,  1.63it/s]
                                                      
{'loss': 0.2829, 'grad_norm': 0.5465973019599915, 'learning_rate': 8.577220067281131e-06, 'epoch': 0.96}

 32%|███▏      | 4283/13317 [44:07<1:32:34,  1.63it/s]
 32%|███▏      | 4284/13317 [44:08<1:32:35,  1.63it/s]
                                                      
{'loss': 0.2019, 'grad_norm': 0.48684388399124146, 'learning_rate': 8.576304242788071e-06, 'epoch': 0.97}

 32%|███▏      | 4284/13317 [44:08<1:32:35,  1.63it/s]
 32%|███▏      | 4285/13317 [44:08<1:32:37,  1.63it/s]
                                                      
{'loss': 0.3121, 'grad_norm': 0.6566863656044006, 'learning_rate': 8.575388172565043e-06, 'epoch': 0.97}

 32%|███▏      | 4285/13317 [44:08<1:32:37,  1.63it/s]
 32%|███▏      | 4286/13317 [44:09<1:32:40,  1.62it/s]
                                                      
{'loss': 0.1898, 'grad_norm': 0.5015695095062256, 'learning_rate': 8.57447185667499e-06, 'epoch': 0.97}

 32%|███▏      | 4286/13317 [44:09<1:32:40,  1.62it/s]
 32%|███▏      | 4287/13317 [44:10<1:33:03,  1.62it/s]
                                                      
{'loss': 0.2463, 'grad_norm': 0.5534515380859375, 'learning_rate': 8.573555295180876e-06, 'epoch': 0.97}

 32%|███▏      | 4287/13317 [44:10<1:33:03,  1.62it/s]
 32%|███▏      | 4288/13317 [44:10<1:32:49,  1.62it/s]
                                                      
{'loss': 0.2716, 'grad_norm': 0.5073380470275879, 'learning_rate': 8.572638488145672e-06, 'epoch': 0.97}

 32%|███▏      | 4288/13317 [44:10<1:32:49,  1.62it/s]
 32%|███▏      | 4289/13317 [44:11<1:32:51,  1.62it/s]
                                                      
{'loss': 0.2475, 'grad_norm': 0.4960489869117737, 'learning_rate': 8.571721435632379e-06, 'epoch': 0.97}

 32%|███▏      | 4289/13317 [44:11<1:32:51,  1.62it/s]
 32%|███▏      | 4290/13317 [44:11<1:32:43,  1.62it/s]
                                                      
{'loss': 0.2991, 'grad_norm': 0.5126152038574219, 'learning_rate': 8.570804137704005e-06, 'epoch': 0.97}

 32%|███▏      | 4290/13317 [44:11<1:32:43,  1.62it/s]
 32%|███▏      | 4291/13317 [44:12<1:32:35,  1.62it/s]
                                                      
{'loss': 0.232, 'grad_norm': 0.5436259508132935, 'learning_rate': 8.569886594423578e-06, 'epoch': 0.97}

 32%|███▏      | 4291/13317 [44:12<1:32:35,  1.62it/s]
 32%|███▏      | 4292/13317 [44:13<1:32:36,  1.62it/s]
                                                      
{'loss': 0.2489, 'grad_norm': 0.50332111120224, 'learning_rate': 8.568968805854144e-06, 'epoch': 0.97}

 32%|███▏      | 4292/13317 [44:13<1:32:36,  1.62it/s]
 32%|███▏      | 4293/13317 [44:13<1:32:31,  1.63it/s]
                                                      
{'loss': 0.2769, 'grad_norm': 0.588525652885437, 'learning_rate': 8.568050772058763e-06, 'epoch': 0.97}

 32%|███▏      | 4293/13317 [44:13<1:32:31,  1.63it/s]
 32%|███▏      | 4294/13317 [44:14<1:32:24,  1.63it/s]
                                                      
{'loss': 0.2728, 'grad_norm': 0.5353415012359619, 'learning_rate': 8.567132493100515e-06, 'epoch': 0.97}

 32%|███▏      | 4294/13317 [44:14<1:32:24,  1.63it/s]
 32%|███▏      | 4295/13317 [44:14<1:32:24,  1.63it/s]
                                                      
{'loss': 0.2338, 'grad_norm': 0.5105951428413391, 'learning_rate': 8.566213969042497e-06, 'epoch': 0.97}

 32%|███▏      | 4295/13317 [44:15<1:32:24,  1.63it/s]
 32%|███▏      | 4296/13317 [44:15<1:32:20,  1.63it/s]
                                                      
{'loss': 0.1968, 'grad_norm': 0.47193899750709534, 'learning_rate': 8.565295199947816e-06, 'epoch': 0.97}

 32%|███▏      | 4296/13317 [44:15<1:32:20,  1.63it/s]
 32%|███▏      | 4297/13317 [44:16<1:32:15,  1.63it/s]
                                                      
{'loss': 0.2524, 'grad_norm': 0.5117133855819702, 'learning_rate': 8.564376185879608e-06, 'epoch': 0.97}

 32%|███▏      | 4297/13317 [44:16<1:32:15,  1.63it/s]
 32%|███▏      | 4298/13317 [44:16<1:32:14,  1.63it/s]
                                                      
{'loss': 0.2652, 'grad_norm': 0.5659831166267395, 'learning_rate': 8.563456926901015e-06, 'epoch': 0.97}

 32%|███▏      | 4298/13317 [44:16<1:32:14,  1.63it/s]
 32%|███▏      | 4299/13317 [44:17<1:32:15,  1.63it/s]
                                                      
{'loss': 0.1808, 'grad_norm': 0.44036245346069336, 'learning_rate': 8.562537423075201e-06, 'epoch': 0.97}

 32%|███▏      | 4299/13317 [44:17<1:32:15,  1.63it/s]
 32%|███▏      | 4300/13317 [44:18<1:32:42,  1.62it/s]
                                                      
{'loss': 0.2634, 'grad_norm': 0.5644093751907349, 'learning_rate': 8.561617674465345e-06, 'epoch': 0.97}

 32%|███▏      | 4300/13317 [44:18<1:32:42,  1.62it/s]
 32%|███▏      | 4301/13317 [44:18<1:32:33,  1.62it/s]
                                                      
{'loss': 0.2262, 'grad_norm': 0.5657991766929626, 'learning_rate': 8.560697681134644e-06, 'epoch': 0.97}

 32%|███▏      | 4301/13317 [44:18<1:32:33,  1.62it/s]
 32%|███▏      | 4302/13317 [44:19<1:32:28,  1.62it/s]
                                                      
{'loss': 0.3319, 'grad_norm': 0.572868287563324, 'learning_rate': 8.559777443146309e-06, 'epoch': 0.97}

 32%|███▏      | 4302/13317 [44:19<1:32:28,  1.62it/s]
 32%|███▏      | 4303/13317 [44:19<1:32:21,  1.63it/s]
                                                      
{'loss': 0.2253, 'grad_norm': 0.5470212697982788, 'learning_rate': 8.558856960563573e-06, 'epoch': 0.97}

 32%|███▏      | 4303/13317 [44:19<1:32:21,  1.63it/s]
 32%|███▏      | 4304/13317 [44:20<1:32:18,  1.63it/s]
                                                      
{'loss': 0.2986, 'grad_norm': 0.5997740626335144, 'learning_rate': 8.557936233449683e-06, 'epoch': 0.97}

 32%|███▏      | 4304/13317 [44:20<1:32:18,  1.63it/s]
 32%|███▏      | 4305/13317 [44:21<1:32:23,  1.63it/s]
                                                      
{'loss': 0.2531, 'grad_norm': 0.5104705691337585, 'learning_rate': 8.557015261867902e-06, 'epoch': 0.97}

 32%|███▏      | 4305/13317 [44:21<1:32:23,  1.63it/s]
 32%|███▏      | 4306/13317 [44:21<1:32:16,  1.63it/s]
                                                      
{'loss': 0.2648, 'grad_norm': 0.5823173522949219, 'learning_rate': 8.556094045881511e-06, 'epoch': 0.97}

 32%|███▏      | 4306/13317 [44:21<1:32:16,  1.63it/s]
 32%|███▏      | 4307/13317 [44:22<1:32:18,  1.63it/s]
                                                      
{'loss': 0.2518, 'grad_norm': 0.4930797219276428, 'learning_rate': 8.555172585553804e-06, 'epoch': 0.97}

 32%|███▏      | 4307/13317 [44:22<1:32:18,  1.63it/s]
 32%|███▏      | 4308/13317 [44:22<1:32:16,  1.63it/s]
                                                      
{'loss': 0.2504, 'grad_norm': 0.647269070148468, 'learning_rate': 8.554250880948101e-06, 'epoch': 0.97}

 32%|███▏      | 4308/13317 [44:23<1:32:16,  1.63it/s]
 32%|███▏      | 4309/13317 [44:23<1:32:20,  1.63it/s]
                                                      
{'loss': 0.2268, 'grad_norm': 0.5076454281806946, 'learning_rate': 8.553328932127727e-06, 'epoch': 0.97}

 32%|███▏      | 4309/13317 [44:23<1:32:20,  1.63it/s]
 32%|███▏      | 4310/13317 [44:24<1:32:24,  1.62it/s]
                                                      
{'loss': 0.2406, 'grad_norm': 0.5156317949295044, 'learning_rate': 8.552406739156032e-06, 'epoch': 0.97}

 32%|███▏      | 4310/13317 [44:24<1:32:24,  1.62it/s]
 32%|███▏      | 4311/13317 [44:24<1:32:16,  1.63it/s]
                                                      
{'loss': 0.1789, 'grad_norm': 0.4805050492286682, 'learning_rate': 8.55148430209638e-06, 'epoch': 0.97}

 32%|███▏      | 4311/13317 [44:24<1:32:16,  1.63it/s]
 32%|███▏      | 4312/13317 [44:25<1:32:19,  1.63it/s]
                                                      
{'loss': 0.1934, 'grad_norm': 0.45779159665107727, 'learning_rate': 8.550561621012155e-06, 'epoch': 0.97}

 32%|███▏      | 4312/13317 [44:25<1:32:19,  1.63it/s]
 32%|███▏      | 4313/13317 [44:25<1:32:14,  1.63it/s]
                                                      
{'loss': 0.2609, 'grad_norm': 0.4690445065498352, 'learning_rate': 8.549638695966751e-06, 'epoch': 0.97}

 32%|███▏      | 4313/13317 [44:26<1:32:14,  1.63it/s]
 32%|███▏      | 4314/13317 [44:26<1:32:19,  1.63it/s]
                                                      
{'loss': 0.2673, 'grad_norm': 0.5437567234039307, 'learning_rate': 8.548715527023584e-06, 'epoch': 0.97}

 32%|███▏      | 4314/13317 [44:26<1:32:19,  1.63it/s]
 32%|███▏      | 4315/13317 [44:27<1:32:20,  1.62it/s]
                                                      
{'loss': 0.3179, 'grad_norm': 0.5555591583251953, 'learning_rate': 8.547792114246087e-06, 'epoch': 0.97}

 32%|███▏      | 4315/13317 [44:27<1:32:20,  1.62it/s]
 32%|███▏      | 4316/13317 [44:27<1:32:13,  1.63it/s]
                                                      
{'loss': 0.2522, 'grad_norm': 0.4789219796657562, 'learning_rate': 8.546868457697707e-06, 'epoch': 0.97}

 32%|███▏      | 4316/13317 [44:27<1:32:13,  1.63it/s]
 32%|███▏      | 4317/13317 [44:28<1:32:16,  1.63it/s]
                                                      
{'loss': 0.1888, 'grad_norm': 0.5238465666770935, 'learning_rate': 8.54594455744191e-06, 'epoch': 0.97}

 32%|███▏      | 4317/13317 [44:28<1:32:16,  1.63it/s]
 32%|███▏      | 4318/13317 [44:29<1:32:12,  1.63it/s]
                                                      
{'loss': 0.3287, 'grad_norm': 0.6156890392303467, 'learning_rate': 8.545020413542176e-06, 'epoch': 0.97}

 32%|███▏      | 4318/13317 [44:29<1:32:12,  1.63it/s]
 32%|███▏      | 4319/13317 [44:29<1:32:14,  1.63it/s]
                                                      
{'loss': 0.1627, 'grad_norm': 0.3938937485218048, 'learning_rate': 8.544096026062004e-06, 'epoch': 0.97}

 32%|███▏      | 4319/13317 [44:29<1:32:14,  1.63it/s]
 32%|███▏      | 4320/13317 [44:30<1:32:13,  1.63it/s]
                                                      
{'loss': 0.2357, 'grad_norm': 0.516272246837616, 'learning_rate': 8.543171395064908e-06, 'epoch': 0.97}

 32%|███▏      | 4320/13317 [44:30<1:32:13,  1.63it/s]
 32%|███▏      | 4321/13317 [44:30<1:32:09,  1.63it/s]
                                                      
{'loss': 0.3114, 'grad_norm': 0.7235705256462097, 'learning_rate': 8.542246520614424e-06, 'epoch': 0.97}

 32%|███▏      | 4321/13317 [44:31<1:32:09,  1.63it/s]
 32%|███▏      | 4322/13317 [44:31<1:32:14,  1.63it/s]
                                                      
{'loss': 0.3092, 'grad_norm': 0.5346482992172241, 'learning_rate': 8.541321402774097e-06, 'epoch': 0.97}

 32%|███▏      | 4322/13317 [44:31<1:32:14,  1.63it/s]
 32%|███▏      | 4323/13317 [44:32<1:32:08,  1.63it/s]
                                                      
{'loss': 0.2676, 'grad_norm': 0.5089733004570007, 'learning_rate': 8.540396041607495e-06, 'epoch': 0.97}

 32%|███▏      | 4323/13317 [44:32<1:32:08,  1.63it/s]
 32%|███▏      | 4324/13317 [44:32<1:32:10,  1.63it/s]
                                                      
{'loss': 0.2592, 'grad_norm': 0.477703720331192, 'learning_rate': 8.539470437178197e-06, 'epoch': 0.97}

 32%|███▏      | 4324/13317 [44:32<1:32:10,  1.63it/s]
 32%|███▏      | 4325/13317 [44:33<1:32:09,  1.63it/s]
                                                      
{'loss': 0.1972, 'grad_norm': 0.44921639561653137, 'learning_rate': 8.538544589549802e-06, 'epoch': 0.97}

 32%|███▏      | 4325/13317 [44:33<1:32:09,  1.63it/s]
 32%|███▏      | 4326/13317 [44:33<1:32:05,  1.63it/s]
                                                      
{'loss': 0.1978, 'grad_norm': 0.47286179661750793, 'learning_rate': 8.53761849878593e-06, 'epoch': 0.97}

 32%|███▏      | 4326/13317 [44:34<1:32:05,  1.63it/s]
 32%|███▏      | 4327/13317 [44:34<1:32:02,  1.63it/s]
                                                      
{'loss': 0.2455, 'grad_norm': 0.47877049446105957, 'learning_rate': 8.536692164950209e-06, 'epoch': 0.97}

 32%|███▏      | 4327/13317 [44:34<1:32:02,  1.63it/s]
 32%|███▏      | 4328/13317 [44:35<1:32:01,  1.63it/s]
                                                      
{'loss': 0.1951, 'grad_norm': 0.4443974792957306, 'learning_rate': 8.535765588106288e-06, 'epoch': 0.97}

 32%|███▏      | 4328/13317 [44:35<1:32:01,  1.63it/s]
 33%|███▎      | 4329/13317 [44:35<1:32:07,  1.63it/s]
                                                      
{'loss': 0.2656, 'grad_norm': 0.5733115077018738, 'learning_rate': 8.534838768317836e-06, 'epoch': 0.98}

 33%|███▎      | 4329/13317 [44:35<1:32:07,  1.63it/s]
 33%|███▎      | 4330/13317 [44:36<1:32:06,  1.63it/s]
                                                      
{'loss': 0.1889, 'grad_norm': 0.448809415102005, 'learning_rate': 8.533911705648532e-06, 'epoch': 0.98}

 33%|███▎      | 4330/13317 [44:36<1:32:06,  1.63it/s]
 33%|███▎      | 4331/13317 [44:37<1:31:59,  1.63it/s]
                                                      
{'loss': 0.2196, 'grad_norm': 0.4646798074245453, 'learning_rate': 8.532984400162077e-06, 'epoch': 0.98}

 33%|███▎      | 4331/13317 [44:37<1:31:59,  1.63it/s]
 33%|███▎      | 4332/13317 [44:37<1:32:02,  1.63it/s]
                                                      
{'loss': 0.2707, 'grad_norm': 0.7270167469978333, 'learning_rate': 8.532056851922185e-06, 'epoch': 0.98}

 33%|███▎      | 4332/13317 [44:37<1:32:02,  1.63it/s]
 33%|███▎      | 4333/13317 [44:38<1:32:01,  1.63it/s]
                                                      
{'loss': 0.2674, 'grad_norm': 0.556828498840332, 'learning_rate': 8.531129060992589e-06, 'epoch': 0.98}

 33%|███▎      | 4333/13317 [44:38<1:32:01,  1.63it/s]
 33%|███▎      | 4334/13317 [44:38<1:31:59,  1.63it/s]
                                                      
{'loss': 0.2461, 'grad_norm': 0.5066354274749756, 'learning_rate': 8.53020102743704e-06, 'epoch': 0.98}

 33%|███▎      | 4334/13317 [44:39<1:31:59,  1.63it/s]
 33%|███▎      | 4335/13317 [44:39<1:32:03,  1.63it/s]
                                                      
{'loss': 0.2402, 'grad_norm': 0.5706592798233032, 'learning_rate': 8.5292727513193e-06, 'epoch': 0.98}

 33%|███▎      | 4335/13317 [44:39<1:32:03,  1.63it/s]
 33%|███▎      | 4336/13317 [44:40<1:31:59,  1.63it/s]
                                                      
{'loss': 0.2418, 'grad_norm': 0.5484632253646851, 'learning_rate': 8.528344232703154e-06, 'epoch': 0.98}

 33%|███▎      | 4336/13317 [44:40<1:31:59,  1.63it/s]
 33%|███▎      | 4337/13317 [44:40<1:32:04,  1.63it/s]
                                                      
{'loss': 0.2506, 'grad_norm': 0.47587013244628906, 'learning_rate': 8.5274154716524e-06, 'epoch': 0.98}

 33%|███▎      | 4337/13317 [44:40<1:32:04,  1.63it/s]
 33%|███▎      | 4338/13317 [44:41<1:32:02,  1.63it/s]
                                                      
{'loss': 0.2661, 'grad_norm': 0.5147255659103394, 'learning_rate': 8.526486468230857e-06, 'epoch': 0.98}

 33%|███▎      | 4338/13317 [44:41<1:32:02,  1.63it/s]
 33%|███▎      | 4339/13317 [44:41<1:31:59,  1.63it/s]
                                                      
{'loss': 0.2406, 'grad_norm': 0.5685962438583374, 'learning_rate': 8.525557222502353e-06, 'epoch': 0.98}

 33%|███▎      | 4339/13317 [44:42<1:31:59,  1.63it/s]
 33%|███▎      | 4340/13317 [44:42<1:32:02,  1.63it/s]
                                                      
{'loss': 0.2751, 'grad_norm': 0.5929262638092041, 'learning_rate': 8.524627734530738e-06, 'epoch': 0.98}

 33%|███▎      | 4340/13317 [44:42<1:32:02,  1.63it/s]
 33%|███▎      | 4341/13317 [44:43<1:31:56,  1.63it/s]
                                                      
{'loss': 0.1837, 'grad_norm': 0.46953511238098145, 'learning_rate': 8.523698004379878e-06, 'epoch': 0.98}

 33%|███▎      | 4341/13317 [44:43<1:31:56,  1.63it/s]
 33%|███▎      | 4342/13317 [44:43<1:32:20,  1.62it/s]
                                                      
{'loss': 0.2535, 'grad_norm': 0.5575631260871887, 'learning_rate': 8.522768032113655e-06, 'epoch': 0.98}

 33%|███▎      | 4342/13317 [44:43<1:32:20,  1.62it/s]
 33%|███▎      | 4343/13317 [44:44<1:32:11,  1.62it/s]
                                                      
{'loss': 0.3449, 'grad_norm': 0.5708798766136169, 'learning_rate': 8.52183781779597e-06, 'epoch': 0.98}

 33%|███▎      | 4343/13317 [44:44<1:32:11,  1.62it/s]
 33%|███▎      | 4344/13317 [44:45<1:32:11,  1.62it/s]
                                                      
{'loss': 0.2351, 'grad_norm': 0.5087423324584961, 'learning_rate': 8.520907361490735e-06, 'epoch': 0.98}

 33%|███▎      | 4344/13317 [44:45<1:32:11,  1.62it/s]
 33%|███▎      | 4345/13317 [44:45<1:32:10,  1.62it/s]
                                                      
{'loss': 0.1855, 'grad_norm': 0.40109652280807495, 'learning_rate': 8.519976663261888e-06, 'epoch': 0.98}

 33%|███▎      | 4345/13317 [44:45<1:32:10,  1.62it/s]
 33%|███▎      | 4346/13317 [44:46<1:32:01,  1.62it/s]
                                                      
{'loss': 0.2621, 'grad_norm': 0.49335893988609314, 'learning_rate': 8.519045723173372e-06, 'epoch': 0.98}

 33%|███▎      | 4346/13317 [44:46<1:32:01,  1.62it/s]
 33%|███▎      | 4347/13317 [44:46<1:32:00,  1.62it/s]
                                                      
{'loss': 0.3036, 'grad_norm': 0.49865052103996277, 'learning_rate': 8.518114541289155e-06, 'epoch': 0.98}

 33%|███▎      | 4347/13317 [44:47<1:32:00,  1.62it/s]
 33%|███▎      | 4348/13317 [44:47<1:31:54,  1.63it/s]
                                                      
{'loss': 0.4359, 'grad_norm': 0.6547760367393494, 'learning_rate': 8.517183117673218e-06, 'epoch': 0.98}

 33%|███▎      | 4348/13317 [44:47<1:31:54,  1.63it/s]
 33%|███▎      | 4349/13317 [44:48<1:31:59,  1.62it/s]
                                                      
{'loss': 0.2184, 'grad_norm': 0.4653278589248657, 'learning_rate': 8.51625145238956e-06, 'epoch': 0.98}

 33%|███▎      | 4349/13317 [44:48<1:31:59,  1.62it/s]
 33%|███▎      | 4350/13317 [44:48<1:32:00,  1.62it/s]
                                                      
{'loss': 0.2962, 'grad_norm': 0.5524643063545227, 'learning_rate': 8.515319545502198e-06, 'epoch': 0.98}

 33%|███▎      | 4350/13317 [44:48<1:32:00,  1.62it/s]
 33%|███▎      | 4351/13317 [44:49<1:31:54,  1.63it/s]
                                                      
{'loss': 0.2567, 'grad_norm': 0.5736774206161499, 'learning_rate': 8.514387397075162e-06, 'epoch': 0.98}

 33%|███▎      | 4351/13317 [44:49<1:31:54,  1.63it/s]
 33%|███▎      | 4352/13317 [44:49<1:31:55,  1.63it/s]
                                                      
{'loss': 0.189, 'grad_norm': 0.44915875792503357, 'learning_rate': 8.5134550071725e-06, 'epoch': 0.98}

 33%|███▎      | 4352/13317 [44:50<1:31:55,  1.63it/s]
 33%|███▎      | 4353/13317 [44:50<1:31:47,  1.63it/s]
                                                      
{'loss': 0.2092, 'grad_norm': 0.4515904188156128, 'learning_rate': 8.512522375858279e-06, 'epoch': 0.98}

 33%|███▎      | 4353/13317 [44:50<1:31:47,  1.63it/s]
 33%|███▎      | 4354/13317 [44:51<1:31:49,  1.63it/s]
                                                      
{'loss': 0.2547, 'grad_norm': 0.4720892012119293, 'learning_rate': 8.511589503196578e-06, 'epoch': 0.98}

 33%|███▎      | 4354/13317 [44:51<1:31:49,  1.63it/s]
 33%|███▎      | 4355/13317 [44:51<1:31:51,  1.63it/s]
                                                      
{'loss': 0.2064, 'grad_norm': 0.4505533277988434, 'learning_rate': 8.510656389251498e-06, 'epoch': 0.98}

 33%|███▎      | 4355/13317 [44:51<1:31:51,  1.63it/s]
 33%|███▎      | 4356/13317 [44:52<1:31:42,  1.63it/s]
                                                      
{'loss': 0.2149, 'grad_norm': 0.48221269249916077, 'learning_rate': 8.509723034087152e-06, 'epoch': 0.98}

 33%|███▎      | 4356/13317 [44:52<1:31:42,  1.63it/s]
 33%|███▎      | 4357/13317 [44:53<1:31:49,  1.63it/s]
                                                      
{'loss': 0.2626, 'grad_norm': 0.5232760906219482, 'learning_rate': 8.508789437767673e-06, 'epoch': 0.98}

 33%|███▎      | 4357/13317 [44:53<1:31:49,  1.63it/s]
 33%|███▎      | 4358/13317 [44:53<1:31:45,  1.63it/s]
                                                      
{'loss': 0.226, 'grad_norm': 0.44577839970588684, 'learning_rate': 8.507855600357206e-06, 'epoch': 0.98}

 33%|███▎      | 4358/13317 [44:53<1:31:45,  1.63it/s]
 33%|███▎      | 4359/13317 [44:54<1:31:42,  1.63it/s]
                                                      
{'loss': 0.2605, 'grad_norm': 0.5955910682678223, 'learning_rate': 8.50692152191992e-06, 'epoch': 0.98}

 33%|███▎      | 4359/13317 [44:54<1:31:42,  1.63it/s]
 33%|███▎      | 4360/13317 [44:54<1:32:08,  1.62it/s]
                                                      
{'loss': 0.2216, 'grad_norm': 0.532555103302002, 'learning_rate': 8.505987202519991e-06, 'epoch': 0.98}

 33%|███▎      | 4360/13317 [44:55<1:32:08,  1.62it/s]
 33%|███▎      | 4361/13317 [44:55<1:31:59,  1.62it/s]
                                                      
{'loss': 0.2219, 'grad_norm': 0.5751959085464478, 'learning_rate': 8.505052642221624e-06, 'epoch': 0.98}

 33%|███▎      | 4361/13317 [44:55<1:31:59,  1.62it/s]
 33%|███▎      | 4362/13317 [44:56<1:31:59,  1.62it/s]
                                                      
{'loss': 0.271, 'grad_norm': 0.5050458908081055, 'learning_rate': 8.504117841089024e-06, 'epoch': 0.98}

 33%|███▎      | 4362/13317 [44:56<1:31:59,  1.62it/s]
 33%|███▎      | 4363/13317 [44:56<1:31:50,  1.62it/s]
                                                      
{'loss': 0.2209, 'grad_norm': 0.586198627948761, 'learning_rate': 8.503182799186427e-06, 'epoch': 0.98}

 33%|███▎      | 4363/13317 [44:56<1:31:50,  1.62it/s]
 33%|███▎      | 4364/13317 [44:57<1:31:48,  1.63it/s]
                                                      
{'loss': 0.2557, 'grad_norm': 0.4356827139854431, 'learning_rate': 8.50224751657808e-06, 'epoch': 0.98}

 33%|███▎      | 4364/13317 [44:57<1:31:48,  1.63it/s]
 33%|███▎      | 4365/13317 [44:57<1:31:51,  1.62it/s]
                                                      
{'loss': 0.2053, 'grad_norm': 0.4997684955596924, 'learning_rate': 8.501311993328246e-06, 'epoch': 0.98}

 33%|███▎      | 4365/13317 [44:58<1:31:51,  1.62it/s]
 33%|███▎      | 4366/13317 [44:58<1:31:45,  1.63it/s]
                                                      
{'loss': 0.2488, 'grad_norm': 0.4660671651363373, 'learning_rate': 8.500376229501204e-06, 'epoch': 0.98}

 33%|███▎      | 4366/13317 [44:58<1:31:45,  1.63it/s]
 33%|███▎      | 4367/13317 [44:59<1:32:08,  1.62it/s]
                                                      
{'loss': 0.2633, 'grad_norm': 0.5221158266067505, 'learning_rate': 8.499440225161256e-06, 'epoch': 0.98}

 33%|███▎      | 4367/13317 [44:59<1:32:08,  1.62it/s]
 33%|███▎      | 4368/13317 [44:59<1:31:56,  1.62it/s]
                                                      
{'loss': 0.2298, 'grad_norm': 0.45707955956459045, 'learning_rate': 8.498503980372708e-06, 'epoch': 0.98}

 33%|███▎      | 4368/13317 [44:59<1:31:56,  1.62it/s]
 33%|███▎      | 4369/13317 [45:00<1:31:52,  1.62it/s]
                                                      
{'loss': 0.3097, 'grad_norm': 0.5868667364120483, 'learning_rate': 8.497567495199896e-06, 'epoch': 0.98}

 33%|███▎      | 4369/13317 [45:00<1:31:52,  1.62it/s]
 33%|███▎      | 4370/13317 [45:01<1:31:50,  1.62it/s]
                                                      
{'loss': 0.173, 'grad_norm': 0.4154578745365143, 'learning_rate': 8.496630769707163e-06, 'epoch': 0.98}

 33%|███▎      | 4370/13317 [45:01<1:31:50,  1.62it/s]
 33%|███▎      | 4371/13317 [45:01<1:31:49,  1.62it/s]
                                                      
{'loss': 0.2775, 'grad_norm': 0.5255468487739563, 'learning_rate': 8.495693803958874e-06, 'epoch': 0.98}

 33%|███▎      | 4371/13317 [45:01<1:31:49,  1.62it/s]
 33%|███▎      | 4372/13317 [45:02<1:31:50,  1.62it/s]
                                                      
{'loss': 0.3039, 'grad_norm': 0.5940401554107666, 'learning_rate': 8.494756598019408e-06, 'epoch': 0.98}

 33%|███▎      | 4372/13317 [45:02<1:31:50,  1.62it/s]
 33%|███▎      | 4373/13317 [45:02<1:31:42,  1.63it/s]
                                                      
{'loss': 0.2783, 'grad_norm': 0.5802193880081177, 'learning_rate': 8.49381915195316e-06, 'epoch': 0.99}

 33%|███▎      | 4373/13317 [45:03<1:31:42,  1.63it/s]
 33%|███▎      | 4374/13317 [45:03<1:31:44,  1.62it/s]
                                                      
{'loss': 0.2317, 'grad_norm': 0.413742333650589, 'learning_rate': 8.492881465824542e-06, 'epoch': 0.99}

 33%|███▎      | 4374/13317 [45:03<1:31:44,  1.62it/s]
 33%|███▎      | 4375/13317 [45:04<1:31:47,  1.62it/s]
                                                      
{'loss': 0.1861, 'grad_norm': 0.4345613121986389, 'learning_rate': 8.491943539697986e-06, 'epoch': 0.99}

 33%|███▎      | 4375/13317 [45:04<1:31:47,  1.62it/s]
 33%|███▎      | 4376/13317 [45:04<1:31:41,  1.63it/s]
                                                      
{'loss': 0.2733, 'grad_norm': 0.5155895352363586, 'learning_rate': 8.491005373637935e-06, 'epoch': 0.99}

 33%|███▎      | 4376/13317 [45:04<1:31:41,  1.63it/s]
 33%|███▎      | 4377/13317 [45:05<1:31:46,  1.62it/s]
                                                      
{'loss': 0.3418, 'grad_norm': 0.6559327840805054, 'learning_rate': 8.490066967708853e-06, 'epoch': 0.99}

 33%|███▎      | 4377/13317 [45:05<1:31:46,  1.62it/s]
 33%|███▎      | 4378/13317 [45:05<1:31:37,  1.63it/s]
                                                      
{'loss': 0.3094, 'grad_norm': 0.6467146873474121, 'learning_rate': 8.489128321975214e-06, 'epoch': 0.99}

 33%|███▎      | 4378/13317 [45:06<1:31:37,  1.63it/s]
 33%|███▎      | 4379/13317 [45:06<1:31:38,  1.63it/s]
                                                      
{'loss': 0.2386, 'grad_norm': 0.4541405141353607, 'learning_rate': 8.488189436501517e-06, 'epoch': 0.99}

 33%|███▎      | 4379/13317 [45:06<1:31:38,  1.63it/s]
 33%|███▎      | 4380/13317 [45:07<1:31:40,  1.62it/s]
                                                      
{'loss': 0.3281, 'grad_norm': 0.6289421916007996, 'learning_rate': 8.487250311352273e-06, 'epoch': 0.99}

 33%|███▎      | 4380/13317 [45:07<1:31:40,  1.62it/s]
 33%|███▎      | 4381/13317 [45:07<1:31:32,  1.63it/s]
                                                      
{'loss': 0.2639, 'grad_norm': 0.6615769267082214, 'learning_rate': 8.486310946592008e-06, 'epoch': 0.99}

 33%|███▎      | 4381/13317 [45:07<1:31:32,  1.63it/s]
 33%|███▎      | 4382/13317 [45:08<1:31:30,  1.63it/s]
                                                      
{'loss': 0.2866, 'grad_norm': 0.5554279685020447, 'learning_rate': 8.48537134228527e-06, 'epoch': 0.99}

 33%|███▎      | 4382/13317 [45:08<1:31:30,  1.63it/s]
 33%|███▎      | 4383/13317 [45:09<1:31:25,  1.63it/s]
                                                      
{'loss': 0.1845, 'grad_norm': 0.49707937240600586, 'learning_rate': 8.484431498496612e-06, 'epoch': 0.99}

 33%|███▎      | 4383/13317 [45:09<1:31:25,  1.63it/s]
 33%|███▎      | 4384/13317 [45:09<1:31:30,  1.63it/s]
                                                      
{'loss': 0.213, 'grad_norm': 0.5354635119438171, 'learning_rate': 8.48349141529062e-06, 'epoch': 0.99}

 33%|███▎      | 4384/13317 [45:09<1:31:30,  1.63it/s]
 33%|███▎      | 4385/13317 [45:10<1:31:33,  1.63it/s]
                                                      
{'loss': 0.2787, 'grad_norm': 0.5649435520172119, 'learning_rate': 8.482551092731883e-06, 'epoch': 0.99}

 33%|███▎      | 4385/13317 [45:10<1:31:33,  1.63it/s]
 33%|███▎      | 4386/13317 [45:10<1:31:27,  1.63it/s]
                                                      
{'loss': 0.2268, 'grad_norm': 0.4997624456882477, 'learning_rate': 8.481610530885012e-06, 'epoch': 0.99}

 33%|███▎      | 4386/13317 [45:11<1:31:27,  1.63it/s]
 33%|███▎      | 4387/13317 [45:11<1:31:28,  1.63it/s]
                                                      
{'loss': 0.2676, 'grad_norm': 0.6278727650642395, 'learning_rate': 8.480669729814635e-06, 'epoch': 0.99}

 33%|███▎      | 4387/13317 [45:11<1:31:28,  1.63it/s]
 33%|███▎      | 4388/13317 [45:12<1:31:23,  1.63it/s]
                                                      
{'loss': 0.1998, 'grad_norm': 0.47970810532569885, 'learning_rate': 8.479728689585392e-06, 'epoch': 0.99}

 33%|███▎      | 4388/13317 [45:12<1:31:23,  1.63it/s]
 33%|███▎      | 4389/13317 [45:12<1:31:27,  1.63it/s]
                                                      
{'loss': 0.1838, 'grad_norm': 0.4084934890270233, 'learning_rate': 8.478787410261943e-06, 'epoch': 0.99}

 33%|███▎      | 4389/13317 [45:12<1:31:27,  1.63it/s]
 33%|███▎      | 4390/13317 [45:13<1:31:29,  1.63it/s]
                                                      
{'loss': 0.2415, 'grad_norm': 0.6010171175003052, 'learning_rate': 8.477845891908966e-06, 'epoch': 0.99}

 33%|███▎      | 4390/13317 [45:13<1:31:29,  1.63it/s]
 33%|███▎      | 4391/13317 [45:13<1:31:24,  1.63it/s]
                                                      
{'loss': 0.2599, 'grad_norm': 0.5429404377937317, 'learning_rate': 8.476904134591153e-06, 'epoch': 0.99}

 33%|███▎      | 4391/13317 [45:14<1:31:24,  1.63it/s]
 33%|███▎      | 4392/13317 [45:14<1:31:30,  1.63it/s]
                                                      
{'loss': 0.2956, 'grad_norm': 0.5464534163475037, 'learning_rate': 8.475962138373212e-06, 'epoch': 0.99}

 33%|███▎      | 4392/13317 [45:14<1:31:30,  1.63it/s]
 33%|███▎      | 4393/13317 [45:15<1:31:25,  1.63it/s]
                                                      
{'loss': 0.2864, 'grad_norm': 0.5753281116485596, 'learning_rate': 8.475019903319868e-06, 'epoch': 0.99}

 33%|███▎      | 4393/13317 [45:15<1:31:25,  1.63it/s]
 33%|███▎      | 4394/13317 [45:15<1:31:28,  1.63it/s]
                                                      
{'loss': 0.1916, 'grad_norm': 0.49122488498687744, 'learning_rate': 8.474077429495862e-06, 'epoch': 0.99}

 33%|███▎      | 4394/13317 [45:15<1:31:28,  1.63it/s]
 33%|███▎      | 4395/13317 [45:16<1:31:25,  1.63it/s]
                                                      
{'loss': 0.1958, 'grad_norm': 0.42159515619277954, 'learning_rate': 8.473134716965953e-06, 'epoch': 0.99}

 33%|███▎      | 4395/13317 [45:16<1:31:25,  1.63it/s]
 33%|███▎      | 4396/13317 [45:17<1:31:18,  1.63it/s]
                                                      
{'loss': 0.4053, 'grad_norm': 0.5134534239768982, 'learning_rate': 8.472191765794912e-06, 'epoch': 0.99}

 33%|███▎      | 4396/13317 [45:17<1:31:18,  1.63it/s]
 33%|███▎      | 4397/13317 [45:17<1:31:31,  1.62it/s]
                                                      
{'loss': 0.2691, 'grad_norm': 0.5288718938827515, 'learning_rate': 8.471248576047537e-06, 'epoch': 0.99}

 33%|███▎      | 4397/13317 [45:17<1:31:31,  1.62it/s]
 33%|███▎      | 4398/13317 [45:18<1:31:21,  1.63it/s]
                                                      
{'loss': 0.2933, 'grad_norm': 0.5697013735771179, 'learning_rate': 8.470305147788628e-06, 'epoch': 0.99}

 33%|███▎      | 4398/13317 [45:18<1:31:21,  1.63it/s]
 33%|███▎      | 4399/13317 [45:18<1:31:26,  1.63it/s]
                                                      
{'loss': 0.2554, 'grad_norm': 0.5341165065765381, 'learning_rate': 8.469361481083012e-06, 'epoch': 0.99}

 33%|███▎      | 4399/13317 [45:19<1:31:26,  1.63it/s]
 33%|███▎      | 4400/13317 [45:19<1:31:27,  1.62it/s]
                                                      
{'loss': 0.2767, 'grad_norm': 0.6753548979759216, 'learning_rate': 8.468417575995526e-06, 'epoch': 0.99}

 33%|███▎      | 4400/13317 [45:19<1:31:27,  1.62it/s]
 33%|███▎      | 4401/13317 [45:20<1:31:20,  1.63it/s]
                                                      
{'loss': 0.2723, 'grad_norm': 0.6337680816650391, 'learning_rate': 8.467473432591032e-06, 'epoch': 0.99}

 33%|███▎      | 4401/13317 [45:20<1:31:20,  1.63it/s]
 33%|███▎      | 4402/13317 [45:20<1:31:21,  1.63it/s]
                                                      
{'loss': 0.3384, 'grad_norm': 0.5843144655227661, 'learning_rate': 8.466529050934397e-06, 'epoch': 0.99}

 33%|███▎      | 4402/13317 [45:20<1:31:21,  1.63it/s]
 33%|███▎      | 4403/13317 [45:21<1:31:16,  1.63it/s]
                                                      
{'loss': 0.2425, 'grad_norm': 0.4587763249874115, 'learning_rate': 8.46558443109051e-06, 'epoch': 0.99}

 33%|███▎      | 4403/13317 [45:21<1:31:16,  1.63it/s]
 33%|███▎      | 4404/13317 [45:21<1:31:17,  1.63it/s]
                                                      
{'loss': 0.2995, 'grad_norm': 0.5704914331436157, 'learning_rate': 8.464639573124283e-06, 'epoch': 0.99}

 33%|███▎      | 4404/13317 [45:22<1:31:17,  1.63it/s]
 33%|███▎      | 4405/13317 [45:22<1:31:20,  1.63it/s]
                                                      
{'loss': 0.2584, 'grad_norm': 0.6005959510803223, 'learning_rate': 8.463694477100629e-06, 'epoch': 0.99}

 33%|███▎      | 4405/13317 [45:22<1:31:20,  1.63it/s]
 33%|███▎      | 4406/13317 [45:23<1:31:15,  1.63it/s]
                                                      
{'loss': 0.2333, 'grad_norm': 0.49299588799476624, 'learning_rate': 8.462749143084492e-06, 'epoch': 0.99}

 33%|███▎      | 4406/13317 [45:23<1:31:15,  1.63it/s]
 33%|███▎      | 4407/13317 [45:23<1:31:18,  1.63it/s]
                                                      
{'loss': 0.2601, 'grad_norm': 0.5738091468811035, 'learning_rate': 8.461803571140826e-06, 'epoch': 0.99}

 33%|███▎      | 4407/13317 [45:23<1:31:18,  1.63it/s]
 33%|███▎      | 4408/13317 [45:24<1:31:11,  1.63it/s]
                                                      
{'loss': 0.2532, 'grad_norm': 0.5253760814666748, 'learning_rate': 8.460857761334599e-06, 'epoch': 0.99}

 33%|███▎      | 4408/13317 [45:24<1:31:11,  1.63it/s]
 33%|███▎      | 4409/13317 [45:25<1:31:13,  1.63it/s]
                                                      
{'loss': 0.1931, 'grad_norm': 0.42991605401039124, 'learning_rate': 8.4599117137308e-06, 'epoch': 0.99}

 33%|███▎      | 4409/13317 [45:25<1:31:13,  1.63it/s]
 33%|███▎      | 4410/13317 [45:25<1:31:15,  1.63it/s]
                                                      
{'loss': 0.2179, 'grad_norm': 0.5644302368164062, 'learning_rate': 8.45896542839443e-06, 'epoch': 0.99}

 33%|███▎      | 4410/13317 [45:25<1:31:15,  1.63it/s]
 33%|███▎      | 4411/13317 [45:26<1:31:09,  1.63it/s]
                                                      
{'loss': 0.2698, 'grad_norm': 0.4537931978702545, 'learning_rate': 8.458018905390512e-06, 'epoch': 0.99}

 33%|███▎      | 4411/13317 [45:26<1:31:09,  1.63it/s]
 33%|███▎      | 4412/13317 [45:26<1:31:17,  1.63it/s]
                                                      
{'loss': 0.3498, 'grad_norm': 0.5297544598579407, 'learning_rate': 8.457072144784083e-06, 'epoch': 0.99}

 33%|███▎      | 4412/13317 [45:27<1:31:17,  1.63it/s]
 33%|███▎      | 4413/13317 [45:27<1:31:14,  1.63it/s]
                                                      
{'loss': 0.2026, 'grad_norm': 0.4526389539241791, 'learning_rate': 8.456125146640192e-06, 'epoch': 0.99}

 33%|███▎      | 4413/13317 [45:27<1:31:14,  1.63it/s]
 33%|███▎      | 4414/13317 [45:28<1:31:18,  1.63it/s]
                                                      
{'loss': 0.2126, 'grad_norm': 0.46670857071876526, 'learning_rate': 8.45517791102391e-06, 'epoch': 0.99}

 33%|███▎      | 4414/13317 [45:28<1:31:18,  1.63it/s]
 33%|███▎      | 4415/13317 [45:28<1:31:21,  1.62it/s]
                                                      
{'loss': 0.2569, 'grad_norm': 0.4828846752643585, 'learning_rate': 8.454230438000322e-06, 'epoch': 0.99}

 33%|███▎      | 4415/13317 [45:28<1:31:21,  1.62it/s]
 33%|███▎      | 4416/13317 [45:29<1:31:14,  1.63it/s]
                                                      
{'loss': 0.1945, 'grad_norm': 0.4311355948448181, 'learning_rate': 8.453282727634527e-06, 'epoch': 0.99}

 33%|███▎      | 4416/13317 [45:29<1:31:14,  1.63it/s]
 33%|███▎      | 4417/13317 [45:29<1:31:11,  1.63it/s]
                                                      
{'loss': 0.2657, 'grad_norm': 0.5255663990974426, 'learning_rate': 8.452334779991646e-06, 'epoch': 1.0}

 33%|███▎      | 4417/13317 [45:30<1:31:11,  1.63it/s]
 33%|███▎      | 4418/13317 [45:30<1:31:07,  1.63it/s]
                                                      
{'loss': 0.2781, 'grad_norm': 0.5098595023155212, 'learning_rate': 8.451386595136813e-06, 'epoch': 1.0}

 33%|███▎      | 4418/13317 [45:30<1:31:07,  1.63it/s]
 33%|███▎      | 4419/13317 [45:31<1:31:08,  1.63it/s]
                                                      
{'loss': 0.2131, 'grad_norm': 0.5076853632926941, 'learning_rate': 8.450438173135176e-06, 'epoch': 1.0}

 33%|███▎      | 4419/13317 [45:31<1:31:08,  1.63it/s]
 33%|███▎      | 4420/13317 [45:31<1:31:10,  1.63it/s]
                                                      
{'loss': 0.2531, 'grad_norm': 0.5900160074234009, 'learning_rate': 8.449489514051901e-06, 'epoch': 1.0}

 33%|███▎      | 4420/13317 [45:31<1:31:10,  1.63it/s]
 33%|███▎      | 4421/13317 [45:32<1:31:03,  1.63it/s]
                                                      
{'loss': 0.2592, 'grad_norm': 0.5818672776222229, 'learning_rate': 8.448540617952175e-06, 'epoch': 1.0}

 33%|███▎      | 4421/13317 [45:32<1:31:03,  1.63it/s]
 33%|███▎      | 4422/13317 [45:33<1:31:07,  1.63it/s]
                                                      
{'loss': 0.2233, 'grad_norm': 0.4956727921962738, 'learning_rate': 8.447591484901192e-06, 'epoch': 1.0}

 33%|███▎      | 4422/13317 [45:33<1:31:07,  1.63it/s]
 33%|███▎      | 4423/13317 [45:33<1:31:02,  1.63it/s]
                                                      
{'loss': 0.2673, 'grad_norm': 0.6314260363578796, 'learning_rate': 8.446642114964173e-06, 'epoch': 1.0}

 33%|███▎      | 4423/13317 [45:33<1:31:02,  1.63it/s]
 33%|███▎      | 4424/13317 [45:34<1:30:57,  1.63it/s]
                                                      
{'loss': 0.2627, 'grad_norm': 0.5923470854759216, 'learning_rate': 8.445692508206347e-06, 'epoch': 1.0}

 33%|███▎      | 4424/13317 [45:34<1:30:57,  1.63it/s]
 33%|███▎      | 4425/13317 [45:34<1:31:02,  1.63it/s]
                                                      
{'loss': 0.2515, 'grad_norm': 0.5489805340766907, 'learning_rate': 8.44474266469296e-06, 'epoch': 1.0}

 33%|███▎      | 4425/13317 [45:35<1:31:02,  1.63it/s]
 33%|███▎      | 4426/13317 [45:35<1:30:59,  1.63it/s]
                                                      
{'loss': 0.2912, 'grad_norm': 0.6037473678588867, 'learning_rate': 8.443792584489282e-06, 'epoch': 1.0}

 33%|███▎      | 4426/13317 [45:35<1:30:59,  1.63it/s]
 33%|███▎      | 4427/13317 [45:36<1:31:03,  1.63it/s]
                                                      
{'loss': 0.1975, 'grad_norm': 0.4577751159667969, 'learning_rate': 8.442842267660587e-06, 'epoch': 1.0}

 33%|███▎      | 4427/13317 [45:36<1:31:03,  1.63it/s]
 33%|███▎      | 4428/13317 [45:36<1:30:58,  1.63it/s]
                                                      
{'loss': 0.2441, 'grad_norm': 0.6260323524475098, 'learning_rate': 8.441891714272174e-06, 'epoch': 1.0}

 33%|███▎      | 4428/13317 [45:36<1:30:58,  1.63it/s]
 33%|███▎      | 4429/13317 [45:37<1:30:56,  1.63it/s]
                                                      
{'loss': 0.2967, 'grad_norm': 0.6219196915626526, 'learning_rate': 8.440940924389358e-06, 'epoch': 1.0}

 33%|███▎      | 4429/13317 [45:37<1:30:56,  1.63it/s]
 33%|███▎      | 4430/13317 [45:37<1:30:59,  1.63it/s]
                                                      
{'loss': 0.3093, 'grad_norm': 0.672958254814148, 'learning_rate': 8.439989898077469e-06, 'epoch': 1.0}

 33%|███▎      | 4430/13317 [45:38<1:30:59,  1.63it/s]
 33%|███▎      | 4431/13317 [45:38<1:30:58,  1.63it/s]
                                                      
{'loss': 0.206, 'grad_norm': 0.497312992811203, 'learning_rate': 8.439038635401849e-06, 'epoch': 1.0}

 33%|███▎      | 4431/13317 [45:38<1:30:58,  1.63it/s]
 33%|███▎      | 4432/13317 [45:39<1:31:04,  1.63it/s]
                                                      
{'loss': 0.2072, 'grad_norm': 0.4347589612007141, 'learning_rate': 8.438087136427862e-06, 'epoch': 1.0}

 33%|███▎      | 4432/13317 [45:39<1:31:04,  1.63it/s]
 33%|███▎      | 4433/13317 [45:39<1:30:58,  1.63it/s]
                                                      
{'loss': 0.2452, 'grad_norm': 0.5452925562858582, 'learning_rate': 8.437135401220886e-06, 'epoch': 1.0}

 33%|███▎      | 4433/13317 [45:39<1:30:58,  1.63it/s]
 33%|███▎      | 4434/13317 [45:40<1:30:53,  1.63it/s]
                                                      
{'loss': 0.2868, 'grad_norm': 0.4859726130962372, 'learning_rate': 8.436183429846314e-06, 'epoch': 1.0}

 33%|███▎      | 4434/13317 [45:40<1:30:53,  1.63it/s]
 33%|███▎      | 4435/13317 [45:41<1:30:52,  1.63it/s]
                                                      
{'loss': 0.2389, 'grad_norm': 0.5175515413284302, 'learning_rate': 8.435231222369559e-06, 'epoch': 1.0}

 33%|███▎      | 4435/13317 [45:41<1:30:52,  1.63it/s]
 33%|███▎      | 4436/13317 [45:41<1:30:50,  1.63it/s]
                                                      
{'loss': 0.2194, 'grad_norm': 0.4877791702747345, 'learning_rate': 8.434278778856044e-06, 'epoch': 1.0}

 33%|███▎      | 4436/13317 [45:41<1:30:50,  1.63it/s]
 33%|███▎      | 4437/13317 [45:42<1:30:49,  1.63it/s]
                                                      
{'loss': 0.273, 'grad_norm': 0.5960177779197693, 'learning_rate': 8.433326099371216e-06, 'epoch': 1.0}

 33%|███▎      | 4437/13317 [45:42<1:30:49,  1.63it/s]
 33%|███▎      | 4438/13317 [45:42<1:30:54,  1.63it/s]
                                                      
{'loss': 0.1548, 'grad_norm': 0.38030505180358887, 'learning_rate': 8.432373183980532e-06, 'epoch': 1.0}

 33%|███▎      | 4438/13317 [45:42<1:30:54,  1.63it/s]
 33%|███▎      | 4439/13317 [45:43<1:30:43,  1.63it/s]
                                                      
{'loss': 0.249, 'grad_norm': 0.512580931186676, 'learning_rate': 8.431420032749468e-06, 'epoch': 1.0}

 33%|███▎      | 4439/13317 [45:43<1:30:43,  1.63it/s]
 33%|███▎      | 4440/13317 [45:44<1:33:20,  1.58it/s]
                                                      
{'loss': 0.2361, 'grad_norm': 1.1820507049560547, 'learning_rate': 8.430466645743515e-06, 'epoch': 1.0}

 33%|███▎      | 4440/13317 [45:44<1:33:20,  1.58it/s]
 33%|███▎      | 4441/13317 [45:44<1:32:29,  1.60it/s]
                                                      
{'loss': 0.1851, 'grad_norm': 0.3966095745563507, 'learning_rate': 8.429513023028181e-06, 'epoch': 1.0}

 33%|███▎      | 4441/13317 [45:44<1:32:29,  1.60it/s]
 33%|███▎      | 4442/13317 [45:45<1:32:05,  1.61it/s]
                                                      
{'loss': 0.1752, 'grad_norm': 0.4312831461429596, 'learning_rate': 8.42855916466899e-06, 'epoch': 1.0}

 33%|███▎      | 4442/13317 [45:45<1:32:05,  1.61it/s]
 33%|███▎      | 4443/13317 [45:45<1:31:41,  1.61it/s]
                                                      
{'loss': 0.2695, 'grad_norm': 0.5274630784988403, 'learning_rate': 8.427605070731482e-06, 'epoch': 1.0}

 33%|███▎      | 4443/13317 [45:46<1:31:41,  1.61it/s]
 33%|███▎      | 4444/13317 [45:46<1:31:24,  1.62it/s]
                                                      
{'loss': 0.2042, 'grad_norm': 0.4918327033519745, 'learning_rate': 8.426650741281214e-06, 'epoch': 1.0}

 33%|███▎      | 4444/13317 [45:46<1:31:24,  1.62it/s]
 33%|███▎      | 4445/13317 [45:47<1:31:38,  1.61it/s]
                                                      
{'loss': 0.2086, 'grad_norm': 0.44127357006073, 'learning_rate': 8.425696176383758e-06, 'epoch': 1.0}

 33%|███▎      | 4445/13317 [45:47<1:31:38,  1.61it/s]
 33%|███▎      | 4446/13317 [45:47<1:31:20,  1.62it/s]
                                                      
{'loss': 0.2133, 'grad_norm': 0.47360628843307495, 'learning_rate': 8.424741376104704e-06, 'epoch': 1.0}

 33%|███▎      | 4446/13317 [45:47<1:31:20,  1.62it/s]
 33%|███▎      | 4447/13317 [45:48<1:31:14,  1.62it/s]
                                                      
{'loss': 0.1593, 'grad_norm': 0.3815549314022064, 'learning_rate': 8.423786340509653e-06, 'epoch': 1.0}

 33%|███▎      | 4447/13317 [45:48<1:31:14,  1.62it/s]
 33%|███▎      | 4448/13317 [45:49<1:31:03,  1.62it/s]
                                                      
{'loss': 0.2155, 'grad_norm': 0.5186254382133484, 'learning_rate': 8.42283106966423e-06, 'epoch': 1.0}

 33%|███▎      | 4448/13317 [45:49<1:31:03,  1.62it/s]
 33%|███▎      | 4449/13317 [45:49<1:31:05,  1.62it/s]
                                                      
{'loss': 0.2762, 'grad_norm': 0.5622355341911316, 'learning_rate': 8.421875563634071e-06, 'epoch': 1.0}

 33%|███▎      | 4449/13317 [45:49<1:31:05,  1.62it/s]
 33%|███▎      | 4450/13317 [45:50<1:31:02,  1.62it/s]
                                                      
{'loss': 0.1927, 'grad_norm': 0.4784732162952423, 'learning_rate': 8.42091982248483e-06, 'epoch': 1.0}

 33%|███▎      | 4450/13317 [45:50<1:31:02,  1.62it/s]
 33%|███▎      | 4451/13317 [45:50<1:30:56,  1.62it/s]
                                                      
{'loss': 0.2256, 'grad_norm': 0.4607066512107849, 'learning_rate': 8.419963846282176e-06, 'epoch': 1.0}

 33%|███▎      | 4451/13317 [45:51<1:30:56,  1.62it/s]
 33%|███▎      | 4452/13317 [45:51<1:31:18,  1.62it/s]
                                                      
{'loss': 0.1701, 'grad_norm': 0.44585156440734863, 'learning_rate': 8.419007635091793e-06, 'epoch': 1.0}

 33%|███▎      | 4452/13317 [45:51<1:31:18,  1.62it/s]
 33%|███▎      | 4453/13317 [45:52<1:31:12,  1.62it/s]
                                                      
{'loss': 0.1855, 'grad_norm': 0.42458096146583557, 'learning_rate': 8.418051188979386e-06, 'epoch': 1.0}

 33%|███▎      | 4453/13317 [45:52<1:31:12,  1.62it/s]
 33%|███▎      | 4454/13317 [45:52<1:31:35,  1.61it/s]
                                                      
{'loss': 0.2042, 'grad_norm': 0.4487585425376892, 'learning_rate': 8.41709450801067e-06, 'epoch': 1.0}

 33%|███▎      | 4454/13317 [45:52<1:31:35,  1.61it/s]
 33%|███▎      | 4455/13317 [45:53<1:31:57,  1.61it/s]
                                                      
{'loss': 0.1812, 'grad_norm': 0.4959597885608673, 'learning_rate': 8.416137592251383e-06, 'epoch': 1.0}

 33%|███▎      | 4455/13317 [45:53<1:31:57,  1.61it/s]
 33%|███▎      | 4456/13317 [45:54<1:31:41,  1.61it/s]
                                                      
{'loss': 0.1698, 'grad_norm': 0.42960619926452637, 'learning_rate': 8.41518044176727e-06, 'epoch': 1.0}

 33%|███▎      | 4456/13317 [45:54<1:31:41,  1.61it/s]
 33%|███▎      | 4457/13317 [45:54<1:31:37,  1.61it/s]
                                                      
{'loss': 0.2409, 'grad_norm': 0.5381656885147095, 'learning_rate': 8.414223056624103e-06, 'epoch': 1.0}

 33%|███▎      | 4457/13317 [45:54<1:31:37,  1.61it/s]
 33%|███▎      | 4458/13317 [45:55<1:31:22,  1.62it/s]
                                                      
{'loss': 0.2181, 'grad_norm': 0.5198270678520203, 'learning_rate': 8.41326543688766e-06, 'epoch': 1.0}

 33%|███▎      | 4458/13317 [45:55<1:31:22,  1.62it/s]
 33%|███▎      | 4459/13317 [45:55<1:31:19,  1.62it/s]
                                                      
{'loss': 0.2383, 'grad_norm': 0.5229595303535461, 'learning_rate': 8.41230758262374e-06, 'epoch': 1.0}

 33%|███▎      | 4459/13317 [45:56<1:31:19,  1.62it/s]
 33%|███▎      | 4460/13317 [45:56<1:31:13,  1.62it/s]
                                                      
{'loss': 0.1831, 'grad_norm': 0.517943799495697, 'learning_rate': 8.411349493898162e-06, 'epoch': 1.0}

 33%|███▎      | 4460/13317 [45:56<1:31:13,  1.62it/s]
 33%|███▎      | 4461/13317 [45:57<1:31:03,  1.62it/s]
                                                      
{'loss': 0.1898, 'grad_norm': 0.4183853566646576, 'learning_rate': 8.41039117077675e-06, 'epoch': 1.0}

 33%|███▎      | 4461/13317 [45:57<1:31:03,  1.62it/s]
 34%|███▎      | 4462/13317 [45:57<1:31:25,  1.61it/s]
                                                      
{'loss': 0.2072, 'grad_norm': 0.5169556736946106, 'learning_rate': 8.409432613325358e-06, 'epoch': 1.01}

 34%|███▎      | 4462/13317 [45:57<1:31:25,  1.61it/s]
 34%|███▎      | 4463/13317 [45:58<1:31:12,  1.62it/s]
                                                      
{'loss': 0.239, 'grad_norm': 0.5544926524162292, 'learning_rate': 8.408473821609845e-06, 'epoch': 1.01}

 34%|███▎      | 4463/13317 [45:58<1:31:12,  1.62it/s]
 34%|███▎      | 4464/13317 [45:58<1:31:07,  1.62it/s]
                                                      
{'loss': 0.2052, 'grad_norm': 0.5240196585655212, 'learning_rate': 8.40751479569609e-06, 'epoch': 1.01}

 34%|███▎      | 4464/13317 [45:59<1:31:07,  1.62it/s]
 34%|███▎      | 4465/13317 [45:59<1:31:13,  1.62it/s]
                                                      
{'loss': 0.2029, 'grad_norm': 0.5033271312713623, 'learning_rate': 8.40655553564999e-06, 'epoch': 1.01}

 34%|███▎      | 4465/13317 [45:59<1:31:13,  1.62it/s]
 34%|███▎      | 4466/13317 [46:00<1:31:03,  1.62it/s]
                                                      
{'loss': 0.2234, 'grad_norm': 0.537635862827301, 'learning_rate': 8.405596041537454e-06, 'epoch': 1.01}

 34%|███▎      | 4466/13317 [46:00<1:31:03,  1.62it/s]
 34%|███▎      | 4467/13317 [46:00<1:31:03,  1.62it/s]
                                                      
{'loss': 0.2261, 'grad_norm': 0.5603316426277161, 'learning_rate': 8.404636313424414e-06, 'epoch': 1.01}

 34%|███▎      | 4467/13317 [46:00<1:31:03,  1.62it/s]
 34%|███▎      | 4468/13317 [46:01<1:30:55,  1.62it/s]
                                                      
{'loss': 0.1849, 'grad_norm': 0.516369640827179, 'learning_rate': 8.403676351376809e-06, 'epoch': 1.01}

 34%|███▎      | 4468/13317 [46:01<1:30:55,  1.62it/s]
 34%|███▎      | 4469/13317 [46:02<1:30:47,  1.62it/s]
                                                      
{'loss': 0.1857, 'grad_norm': 0.5707314610481262, 'learning_rate': 8.402716155460599e-06, 'epoch': 1.01}

 34%|███▎      | 4469/13317 [46:02<1:30:47,  1.62it/s]
 34%|███▎      | 4470/13317 [46:02<1:30:52,  1.62it/s]
                                                      
{'loss': 0.2376, 'grad_norm': 0.6455710530281067, 'learning_rate': 8.401755725741761e-06, 'epoch': 1.01}

 34%|███▎      | 4470/13317 [46:02<1:30:52,  1.62it/s]
 34%|███▎      | 4471/13317 [46:03<1:30:45,  1.62it/s]
                                                      
{'loss': 0.2064, 'grad_norm': 0.5182927250862122, 'learning_rate': 8.400795062286287e-06, 'epoch': 1.01}

 34%|███▎      | 4471/13317 [46:03<1:30:45,  1.62it/s]
 34%|███▎      | 4472/13317 [46:03<1:30:48,  1.62it/s]
                                                      
{'loss': 0.1592, 'grad_norm': 0.4844706654548645, 'learning_rate': 8.399834165160183e-06, 'epoch': 1.01}

 34%|███▎      | 4472/13317 [46:04<1:30:48,  1.62it/s]
 34%|███▎      | 4473/13317 [46:04<1:30:42,  1.63it/s]
                                                      
{'loss': 0.1835, 'grad_norm': 0.5058581829071045, 'learning_rate': 8.398873034429475e-06, 'epoch': 1.01}

 34%|███▎      | 4473/13317 [46:04<1:30:42,  1.63it/s]
 34%|███▎      | 4474/13317 [46:05<1:30:42,  1.62it/s]
                                                      
{'loss': 0.1897, 'grad_norm': 0.4977978765964508, 'learning_rate': 8.397911670160201e-06, 'epoch': 1.01}

 34%|███▎      | 4474/13317 [46:05<1:30:42,  1.62it/s]
 34%|███▎      | 4475/13317 [46:05<1:30:46,  1.62it/s]
                                                      
{'loss': 0.168, 'grad_norm': 0.45657438039779663, 'learning_rate': 8.396950072418418e-06, 'epoch': 1.01}

 34%|███▎      | 4475/13317 [46:05<1:30:46,  1.62it/s]
 34%|███▎      | 4476/13317 [46:06<1:30:39,  1.63it/s]
                                                      
{'loss': 0.2209, 'grad_norm': 0.5796810984611511, 'learning_rate': 8.395988241270197e-06, 'epoch': 1.01}

 34%|███▎      | 4476/13317 [46:06<1:30:39,  1.63it/s]
 34%|███▎      | 4477/13317 [46:06<1:30:41,  1.62it/s]
                                                      
{'loss': 0.2, 'grad_norm': 0.5720324516296387, 'learning_rate': 8.395026176781627e-06, 'epoch': 1.01}

 34%|███▎      | 4477/13317 [46:07<1:30:41,  1.62it/s]
 34%|███▎      | 4478/13317 [46:07<1:30:36,  1.63it/s]
                                                      
{'loss': 0.184, 'grad_norm': 0.44398564100265503, 'learning_rate': 8.39406387901881e-06, 'epoch': 1.01}

 34%|███▎      | 4478/13317 [46:07<1:30:36,  1.63it/s]
 34%|███▎      | 4479/13317 [46:08<1:30:34,  1.63it/s]
                                                      
{'loss': 0.1928, 'grad_norm': 0.4985230565071106, 'learning_rate': 8.39310134804787e-06, 'epoch': 1.01}

 34%|███▎      | 4479/13317 [46:08<1:30:34,  1.63it/s]
 34%|███▎      | 4480/13317 [46:08<1:30:36,  1.63it/s]
                                                      
{'loss': 0.2006, 'grad_norm': 0.5804016590118408, 'learning_rate': 8.392138583934938e-06, 'epoch': 1.01}

 34%|███▎      | 4480/13317 [46:08<1:30:36,  1.63it/s]
 34%|███▎      | 4481/13317 [46:09<1:30:31,  1.63it/s]
                                                      
{'loss': 0.221, 'grad_norm': 0.5250278115272522, 'learning_rate': 8.39117558674617e-06, 'epoch': 1.01}

 34%|███▎      | 4481/13317 [46:09<1:30:31,  1.63it/s]
 34%|███▎      | 4482/13317 [46:10<1:30:33,  1.63it/s]
                                                      
{'loss': 0.1878, 'grad_norm': 0.49435922503471375, 'learning_rate': 8.390212356547734e-06, 'epoch': 1.01}

 34%|███▎      | 4482/13317 [46:10<1:30:33,  1.63it/s]
 34%|███▎      | 4483/13317 [46:10<1:30:27,  1.63it/s]
                                                      
{'loss': 0.2114, 'grad_norm': 0.5285784602165222, 'learning_rate': 8.389248893405811e-06, 'epoch': 1.01}

 34%|███▎      | 4483/13317 [46:10<1:30:27,  1.63it/s]
 34%|███▎      | 4484/13317 [46:11<1:30:29,  1.63it/s]
                                                      
{'loss': 0.184, 'grad_norm': 0.4435827434062958, 'learning_rate': 8.388285197386605e-06, 'epoch': 1.01}

 34%|███▎      | 4484/13317 [46:11<1:30:29,  1.63it/s]
 34%|███▎      | 4485/13317 [46:11<1:30:26,  1.63it/s]
                                                      
{'loss': 0.1719, 'grad_norm': 0.42449620366096497, 'learning_rate': 8.38732126855633e-06, 'epoch': 1.01}

 34%|███▎      | 4485/13317 [46:12<1:30:26,  1.63it/s]
 34%|███▎      | 4486/13317 [46:12<1:30:23,  1.63it/s]
                                                      
{'loss': 0.2636, 'grad_norm': 0.6152002811431885, 'learning_rate': 8.386357106981217e-06, 'epoch': 1.01}

 34%|███▎      | 4486/13317 [46:12<1:30:23,  1.63it/s]
 34%|███▎      | 4487/13317 [46:13<1:30:23,  1.63it/s]
                                                      
{'loss': 0.1729, 'grad_norm': 0.4588373899459839, 'learning_rate': 8.385392712727516e-06, 'epoch': 1.01}

 34%|███▎      | 4487/13317 [46:13<1:30:23,  1.63it/s]
 34%|███▎      | 4488/13317 [46:13<1:30:21,  1.63it/s]
                                                      
{'loss': 0.1923, 'grad_norm': 0.4565897583961487, 'learning_rate': 8.38442808586149e-06, 'epoch': 1.01}

 34%|███▎      | 4488/13317 [46:13<1:30:21,  1.63it/s]
 34%|███▎      | 4489/13317 [46:14<1:30:19,  1.63it/s]
                                                      
{'loss': 0.2073, 'grad_norm': 0.636408269405365, 'learning_rate': 8.38346322644942e-06, 'epoch': 1.01}

 34%|███▎      | 4489/13317 [46:14<1:30:19,  1.63it/s]
 34%|███▎      | 4490/13317 [46:14<1:30:22,  1.63it/s]
                                                      
{'loss': 0.1797, 'grad_norm': 0.4460062086582184, 'learning_rate': 8.382498134557602e-06, 'epoch': 1.01}

 34%|███▎      | 4490/13317 [46:15<1:30:22,  1.63it/s]
 34%|███▎      | 4491/13317 [46:15<1:30:20,  1.63it/s]
                                                      
{'loss': 0.2811, 'grad_norm': 0.5922544598579407, 'learning_rate': 8.38153281025235e-06, 'epoch': 1.01}

 34%|███▎      | 4491/13317 [46:15<1:30:20,  1.63it/s]
 34%|███▎      | 4492/13317 [46:16<1:30:25,  1.63it/s]
                                                      
{'loss': 0.1681, 'grad_norm': 0.4660332500934601, 'learning_rate': 8.380567253599984e-06, 'epoch': 1.01}

 34%|███▎      | 4492/13317 [46:16<1:30:25,  1.63it/s]
 34%|███▎      | 4493/13317 [46:16<1:30:21,  1.63it/s]
                                                      
{'loss': 0.2202, 'grad_norm': 0.4934726655483246, 'learning_rate': 8.379601464666858e-06, 'epoch': 1.01}

 34%|███▎      | 4493/13317 [46:16<1:30:21,  1.63it/s]
 34%|███▎      | 4494/13317 [46:17<1:30:19,  1.63it/s]
                                                      
{'loss': 0.1985, 'grad_norm': 0.5692464709281921, 'learning_rate': 8.378635443519327e-06, 'epoch': 1.01}

 34%|███▎      | 4494/13317 [46:17<1:30:19,  1.63it/s]
 34%|███▍      | 4495/13317 [46:18<1:30:24,  1.63it/s]
                                                      
{'loss': 0.2335, 'grad_norm': 0.5669459104537964, 'learning_rate': 8.377669190223765e-06, 'epoch': 1.01}

 34%|███▍      | 4495/13317 [46:18<1:30:24,  1.63it/s]
 34%|███▍      | 4496/13317 [46:18<1:30:22,  1.63it/s]
                                                      
{'loss': 0.1749, 'grad_norm': 0.43509405851364136, 'learning_rate': 8.376702704846569e-06, 'epoch': 1.01}

 34%|███▍      | 4496/13317 [46:18<1:30:22,  1.63it/s]
 34%|███▍      | 4497/13317 [46:19<1:30:27,  1.62it/s]
                                                      
{'loss': 0.1707, 'grad_norm': 0.46954724192619324, 'learning_rate': 8.375735987454145e-06, 'epoch': 1.01}

 34%|███▍      | 4497/13317 [46:19<1:30:27,  1.62it/s]
 34%|███▍      | 4498/13317 [46:19<1:30:23,  1.63it/s]
                                                      
{'loss': 0.2664, 'grad_norm': 0.5801684856414795, 'learning_rate': 8.374769038112913e-06, 'epoch': 1.01}

 34%|███▍      | 4498/13317 [46:20<1:30:23,  1.63it/s]
 34%|███▍      | 4499/13317 [46:20<1:30:22,  1.63it/s]
                                                      
{'loss': 0.1973, 'grad_norm': 0.5420024394989014, 'learning_rate': 8.373801856889316e-06, 'epoch': 1.01}

 34%|███▍      | 4499/13317 [46:20<1:30:22,  1.63it/s]
 34%|███▍      | 4500/13317 [46:21<1:30:25,  1.63it/s]
                                                      
{'loss': 0.2194, 'grad_norm': 0.5644185543060303, 'learning_rate': 8.37283444384981e-06, 'epoch': 1.01}

 34%|███▍      | 4500/13317 [46:21<1:30:25,  1.63it/s]

  0%|          | 0/13 [00:00<?, ?it/s]

 15%|█▌        | 2/13 [00:00<00:00, 13.33it/s]

 31%|███       | 4/13 [00:00<00:01,  8.39it/s]

 38%|███▊      | 5/13 [00:00<00:01,  7.79it/s]

 46%|████▌     | 6/13 [00:00<00:00,  7.42it/s]

 54%|█████▍    | 7/13 [00:00<00:00,  7.18it/s]

 62%|██████▏   | 8/13 [00:01<00:00,  7.01it/s]

 69%|██████▉   | 9/13 [00:01<00:00,  6.90it/s]

 77%|███████▋  | 10/13 [00:01<00:00,  6.83it/s]

 85%|████████▍ | 11/13 [00:01<00:00,  6.78it/s]

 92%|█████████▏| 12/13 [00:01<00:00,  6.74it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]
                                                      

                                               
{'eval_loss': 0.6302960515022278, 'eval_runtime': 1.9544, 'eval_samples_per_second': 102.335, 'eval_steps_per_second': 6.652, 'epoch': 1.01}

 34%|███▍      | 4500/13317 [46:23<1:30:25,  1.63it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]

                                               
 34%|███▍      | 4501/13317 [46:23<2:56:44,  1.20s/it]
                                                      
{'loss': 0.1881, 'grad_norm': 0.4796017110347748, 'learning_rate': 8.371866799060866e-06, 'epoch': 1.01}

 34%|███▍      | 4501/13317 [46:23<2:56:44,  1.20s/it]
 34%|███▍      | 4502/13317 [46:24<2:30:45,  1.03s/it]
                                                      
{'loss': 0.1674, 'grad_norm': 0.492373526096344, 'learning_rate': 8.37089892258897e-06, 'epoch': 1.01}

 34%|███▍      | 4502/13317 [46:24<2:30:45,  1.03s/it]
 34%|███▍      | 4503/13317 [46:24<2:12:35,  1.11it/s]
                                                      
{'loss': 0.1722, 'grad_norm': 0.4495049715042114, 'learning_rate': 8.369930814500626e-06, 'epoch': 1.01}

 34%|███▍      | 4503/13317 [46:25<2:12:35,  1.11it/s]
 34%|███▍      | 4504/13317 [46:25<1:59:58,  1.22it/s]
                                                      
{'loss': 0.2418, 'grad_norm': 0.6097770929336548, 'learning_rate': 8.368962474862354e-06, 'epoch': 1.01}

 34%|███▍      | 4504/13317 [46:25<1:59:58,  1.22it/s]
 34%|███▍      | 4505/13317 [46:26<1:51:05,  1.32it/s]
                                                      
{'loss': 0.1489, 'grad_norm': 0.48086071014404297, 'learning_rate': 8.36799390374069e-06, 'epoch': 1.01}

 34%|███▍      | 4505/13317 [46:26<1:51:05,  1.32it/s]
 34%|███▍      | 4506/13317 [46:26<1:44:48,  1.40it/s]
                                                      
{'loss': 0.1776, 'grad_norm': 0.48339104652404785, 'learning_rate': 8.367025101202182e-06, 'epoch': 1.02}

 34%|███▍      | 4506/13317 [46:26<1:44:48,  1.40it/s]
 34%|███▍      | 4507/13317 [46:27<1:40:34,  1.46it/s]
                                                      
{'loss': 0.2187, 'grad_norm': 0.5219004154205322, 'learning_rate': 8.366056067313399e-06, 'epoch': 1.02}

 34%|███▍      | 4507/13317 [46:27<1:40:34,  1.46it/s]
 34%|███▍      | 4508/13317 [46:27<1:37:24,  1.51it/s]
                                                      
{'loss': 0.2342, 'grad_norm': 0.5743196606636047, 'learning_rate': 8.365086802140927e-06, 'epoch': 1.02}

 34%|███▍      | 4508/13317 [46:28<1:37:24,  1.51it/s]
 34%|███▍      | 4509/13317 [46:28<1:35:20,  1.54it/s]
                                                      
{'loss': 0.1759, 'grad_norm': 0.5399138927459717, 'learning_rate': 8.364117305751359e-06, 'epoch': 1.02}

 34%|███▍      | 4509/13317 [46:28<1:35:20,  1.54it/s]
 34%|███▍      | 4510/13317 [46:29<1:33:46,  1.57it/s]
                                                      
{'loss': 0.1884, 'grad_norm': 0.4752030670642853, 'learning_rate': 8.363147578211312e-06, 'epoch': 1.02}

 34%|███▍      | 4510/13317 [46:29<1:33:46,  1.57it/s]
 34%|███▍      | 4511/13317 [46:29<1:32:39,  1.58it/s]
                                                      
{'loss': 0.225, 'grad_norm': 0.5357242226600647, 'learning_rate': 8.362177619587418e-06, 'epoch': 1.02}

 34%|███▍      | 4511/13317 [46:29<1:32:39,  1.58it/s]
 34%|███▍      | 4512/13317 [46:30<1:31:57,  1.60it/s]
                                                      
{'loss': 0.156, 'grad_norm': 0.544958770275116, 'learning_rate': 8.361207429946319e-06, 'epoch': 1.02}

 34%|███▍      | 4512/13317 [46:30<1:31:57,  1.60it/s]
 34%|███▍      | 4513/13317 [46:31<1:31:21,  1.61it/s]
                                                      
{'loss': 0.2119, 'grad_norm': 0.5469934344291687, 'learning_rate': 8.360237009354684e-06, 'epoch': 1.02}

 34%|███▍      | 4513/13317 [46:31<1:31:21,  1.61it/s]
 34%|███▍      | 4514/13317 [46:31<1:31:05,  1.61it/s]
                                                      
{'loss': 0.2757, 'grad_norm': 0.6796353459358215, 'learning_rate': 8.359266357879188e-06, 'epoch': 1.02}

 34%|███▍      | 4514/13317 [46:31<1:31:05,  1.61it/s]
 34%|███▍      | 4515/13317 [46:32<1:30:44,  1.62it/s]
                                                      
{'loss': 0.2317, 'grad_norm': 0.6092532873153687, 'learning_rate': 8.358295475586522e-06, 'epoch': 1.02}

 34%|███▍      | 4515/13317 [46:32<1:30:44,  1.62it/s]
 34%|███▍      | 4516/13317 [46:32<1:30:31,  1.62it/s]
                                                      
{'loss': 0.2404, 'grad_norm': 0.5564289689064026, 'learning_rate': 8.3573243625434e-06, 'epoch': 1.02}

 34%|███▍      | 4516/13317 [46:33<1:30:31,  1.62it/s]
 34%|███▍      | 4517/13317 [46:33<1:30:27,  1.62it/s]
                                                      
{'loss': 0.2157, 'grad_norm': 0.4953174293041229, 'learning_rate': 8.356353018816547e-06, 'epoch': 1.02}

 34%|███▍      | 4517/13317 [46:33<1:30:27,  1.62it/s]
 34%|███▍      | 4518/13317 [46:34<1:30:20,  1.62it/s]
                                                      
{'loss': 0.2058, 'grad_norm': 0.5418611764907837, 'learning_rate': 8.355381444472703e-06, 'epoch': 1.02}

 34%|███▍      | 4518/13317 [46:34<1:30:20,  1.62it/s]
 34%|███▍      | 4519/13317 [46:34<1:30:22,  1.62it/s]
                                                      
{'loss': 0.2229, 'grad_norm': 0.5410998463630676, 'learning_rate': 8.354409639578628e-06, 'epoch': 1.02}

 34%|███▍      | 4519/13317 [46:34<1:30:22,  1.62it/s]
 34%|███▍      | 4520/13317 [46:35<1:30:20,  1.62it/s]
                                                      
{'loss': 0.2012, 'grad_norm': 0.5199941992759705, 'learning_rate': 8.353437604201092e-06, 'epoch': 1.02}

 34%|███▍      | 4520/13317 [46:35<1:30:20,  1.62it/s]
 34%|███▍      | 4521/13317 [46:35<1:30:10,  1.63it/s]
                                                      
{'loss': 0.2068, 'grad_norm': 0.5042938590049744, 'learning_rate': 8.352465338406887e-06, 'epoch': 1.02}

 34%|███▍      | 4521/13317 [46:36<1:30:10,  1.63it/s]
 34%|███▍      | 4522/13317 [46:36<1:30:15,  1.62it/s]
                                                      
{'loss': 0.1727, 'grad_norm': 0.3850330412387848, 'learning_rate': 8.351492842262814e-06, 'epoch': 1.02}

 34%|███▍      | 4522/13317 [46:36<1:30:15,  1.62it/s]
 34%|███▍      | 4523/13317 [46:37<1:30:09,  1.63it/s]
                                                      
{'loss': 0.1963, 'grad_norm': 0.43144217133522034, 'learning_rate': 8.350520115835701e-06, 'epoch': 1.02}

 34%|███▍      | 4523/13317 [46:37<1:30:09,  1.63it/s]
 34%|███▍      | 4524/13317 [46:37<1:30:03,  1.63it/s]
                                                      
{'loss': 0.1962, 'grad_norm': 0.5043724179267883, 'learning_rate': 8.349547159192377e-06, 'epoch': 1.02}

 34%|███▍      | 4524/13317 [46:37<1:30:03,  1.63it/s]
 34%|███▍      | 4525/13317 [46:38<1:30:06,  1.63it/s]
                                                      
{'loss': 0.1755, 'grad_norm': 0.46422821283340454, 'learning_rate': 8.348573972399698e-06, 'epoch': 1.02}

 34%|███▍      | 4525/13317 [46:38<1:30:06,  1.63it/s]
 34%|███▍      | 4526/13317 [46:39<1:30:03,  1.63it/s]
                                                      
{'loss': 0.1678, 'grad_norm': 0.444173127412796, 'learning_rate': 8.347600555524534e-06, 'epoch': 1.02}

 34%|███▍      | 4526/13317 [46:39<1:30:03,  1.63it/s]
 34%|███▍      | 4527/13317 [46:39<1:30:01,  1.63it/s]
                                                      
{'loss': 0.1835, 'grad_norm': 0.47497686743736267, 'learning_rate': 8.346626908633764e-06, 'epoch': 1.02}

 34%|███▍      | 4527/13317 [46:39<1:30:01,  1.63it/s]
 34%|███▍      | 4528/13317 [46:40<1:29:59,  1.63it/s]
                                                      
{'loss': 0.1887, 'grad_norm': 0.4784148335456848, 'learning_rate': 8.345653031794292e-06, 'epoch': 1.02}

 34%|███▍      | 4528/13317 [46:40<1:29:59,  1.63it/s]
 34%|███▍      | 4529/13317 [46:40<1:29:58,  1.63it/s]
                                                      
{'loss': 0.2324, 'grad_norm': 0.5166813731193542, 'learning_rate': 8.344678925073033e-06, 'epoch': 1.02}

 34%|███▍      | 4529/13317 [46:41<1:29:58,  1.63it/s]
 34%|███▍      | 4530/13317 [46:41<1:30:00,  1.63it/s]
                                                      
{'loss': 0.1641, 'grad_norm': 0.42045480012893677, 'learning_rate': 8.343704588536917e-06, 'epoch': 1.02}

 34%|███▍      | 4530/13317 [46:41<1:30:00,  1.63it/s]
 34%|███▍      | 4531/13317 [46:42<1:29:55,  1.63it/s]
                                                      
{'loss': 0.1764, 'grad_norm': 0.4525672197341919, 'learning_rate': 8.342730022252891e-06, 'epoch': 1.02}

 34%|███▍      | 4531/13317 [46:42<1:29:55,  1.63it/s]
 34%|███▍      | 4532/13317 [46:42<1:30:29,  1.62it/s]
                                                      
{'loss': 0.2764, 'grad_norm': 0.48279446363449097, 'learning_rate': 8.34175522628792e-06, 'epoch': 1.02}

 34%|███▍      | 4532/13317 [46:42<1:30:29,  1.62it/s]
 34%|███▍      | 4533/13317 [46:43<1:30:15,  1.62it/s]
                                                      
{'loss': 0.1868, 'grad_norm': 0.5658074021339417, 'learning_rate': 8.340780200708982e-06, 'epoch': 1.02}

 34%|███▍      | 4533/13317 [46:43<1:30:15,  1.62it/s]
 34%|███▍      | 4534/13317 [46:43<1:30:38,  1.61it/s]
                                                      
{'loss': 0.2024, 'grad_norm': 0.4563232362270355, 'learning_rate': 8.339804945583073e-06, 'epoch': 1.02}

 34%|███▍      | 4534/13317 [46:44<1:30:38,  1.61it/s]
 34%|███▍      | 4535/13317 [46:44<1:30:27,  1.62it/s]
                                                      
{'loss': 0.2158, 'grad_norm': 0.5927747488021851, 'learning_rate': 8.3388294609772e-06, 'epoch': 1.02}

 34%|███▍      | 4535/13317 [46:44<1:30:27,  1.62it/s]
 34%|███▍      | 4536/13317 [46:45<1:30:14,  1.62it/s]
                                                      
{'loss': 0.2108, 'grad_norm': 0.5201278924942017, 'learning_rate': 8.337853746958391e-06, 'epoch': 1.02}

 34%|███▍      | 4536/13317 [46:45<1:30:14,  1.62it/s]
 34%|███▍      | 4537/13317 [46:45<1:30:12,  1.62it/s]
                                                      
{'loss': 0.2212, 'grad_norm': 0.5584505200386047, 'learning_rate': 8.336877803593688e-06, 'epoch': 1.02}

 34%|███▍      | 4537/13317 [46:45<1:30:12,  1.62it/s]
 34%|███▍      | 4538/13317 [46:46<1:30:06,  1.62it/s]
                                                      
{'loss': 0.2297, 'grad_norm': 0.562676191329956, 'learning_rate': 8.33590163095015e-06, 'epoch': 1.02}

 34%|███▍      | 4538/13317 [46:46<1:30:06,  1.62it/s]
 34%|███▍      | 4539/13317 [46:47<1:30:03,  1.62it/s]
                                                      
{'loss': 0.2461, 'grad_norm': 0.5986785292625427, 'learning_rate': 8.334925229094845e-06, 'epoch': 1.02}

 34%|███▍      | 4539/13317 [46:47<1:30:03,  1.62it/s]
 34%|███▍      | 4540/13317 [46:47<1:30:00,  1.63it/s]
                                                      
{'loss': 0.166, 'grad_norm': 0.40332815051078796, 'learning_rate': 8.33394859809487e-06, 'epoch': 1.02}

 34%|███▍      | 4540/13317 [46:47<1:30:00,  1.63it/s]
 34%|███▍      | 4541/13317 [46:48<1:29:56,  1.63it/s]
                                                      
{'loss': 0.1629, 'grad_norm': 0.45365625619888306, 'learning_rate': 8.332971738017324e-06, 'epoch': 1.02}

 34%|███▍      | 4541/13317 [46:48<1:29:56,  1.63it/s]
 34%|███▍      | 4542/13317 [46:48<1:30:01,  1.62it/s]
                                                      
{'loss': 0.2354, 'grad_norm': 0.6110478043556213, 'learning_rate': 8.33199464892933e-06, 'epoch': 1.02}

 34%|███▍      | 4542/13317 [46:49<1:30:01,  1.62it/s]
 34%|███▍      | 4543/13317 [46:49<1:29:55,  1.63it/s]
                                                      
{'loss': 0.1767, 'grad_norm': 0.48807644844055176, 'learning_rate': 8.331017330898024e-06, 'epoch': 1.02}

 34%|███▍      | 4543/13317 [46:49<1:29:55,  1.63it/s]
 34%|███▍      | 4544/13317 [46:50<1:29:50,  1.63it/s]
                                                      
{'loss': 0.1782, 'grad_norm': 0.4972807466983795, 'learning_rate': 8.330039783990558e-06, 'epoch': 1.02}

 34%|███▍      | 4544/13317 [46:50<1:29:50,  1.63it/s]
 34%|███▍      | 4545/13317 [46:50<1:29:54,  1.63it/s]
                                                      
{'loss': 0.2231, 'grad_norm': 0.5850635766983032, 'learning_rate': 8.3290620082741e-06, 'epoch': 1.02}

 34%|███▍      | 4545/13317 [46:50<1:29:54,  1.63it/s]
 34%|███▍      | 4546/13317 [46:51<1:29:49,  1.63it/s]
                                                      
{'loss': 0.1745, 'grad_norm': 0.575425386428833, 'learning_rate': 8.328084003815833e-06, 'epoch': 1.02}

 34%|███▍      | 4546/13317 [46:51<1:29:49,  1.63it/s]
 34%|███▍      | 4547/13317 [46:51<1:29:46,  1.63it/s]
                                                      
{'loss': 0.2521, 'grad_norm': 0.5977311134338379, 'learning_rate': 8.327105770682955e-06, 'epoch': 1.02}

 34%|███▍      | 4547/13317 [46:52<1:29:46,  1.63it/s]
 34%|███▍      | 4548/13317 [46:52<1:29:49,  1.63it/s]
                                                      
{'loss': 0.2392, 'grad_norm': 0.5513691306114197, 'learning_rate': 8.326127308942686e-06, 'epoch': 1.02}

 34%|███▍      | 4548/13317 [46:52<1:29:49,  1.63it/s]
 34%|███▍      | 4549/13317 [46:53<1:29:50,  1.63it/s]
                                                      
{'loss': 0.1752, 'grad_norm': 0.4410680830478668, 'learning_rate': 8.32514861866225e-06, 'epoch': 1.02}

 34%|███▍      | 4549/13317 [46:53<1:29:50,  1.63it/s]
 34%|███▍      | 4550/13317 [46:53<1:29:52,  1.63it/s]
                                                      
{'loss': 0.1801, 'grad_norm': 0.4542158842086792, 'learning_rate': 8.324169699908899e-06, 'epoch': 1.03}

 34%|███▍      | 4550/13317 [46:53<1:29:52,  1.63it/s]
 34%|███▍      | 4551/13317 [46:54<1:29:47,  1.63it/s]
                                                      
{'loss': 0.1496, 'grad_norm': 0.45227572321891785, 'learning_rate': 8.32319055274989e-06, 'epoch': 1.03}

 34%|███▍      | 4551/13317 [46:54<1:29:47,  1.63it/s]
 34%|███▍      | 4552/13317 [46:55<1:29:44,  1.63it/s]
                                                      
{'loss': 0.1837, 'grad_norm': 0.4821741580963135, 'learning_rate': 8.322211177252506e-06, 'epoch': 1.03}

 34%|███▍      | 4552/13317 [46:55<1:29:44,  1.63it/s]
 34%|███▍      | 4553/13317 [46:55<1:29:40,  1.63it/s]
                                                      
{'loss': 0.1969, 'grad_norm': 0.4930833876132965, 'learning_rate': 8.321231573484039e-06, 'epoch': 1.03}

 34%|███▍      | 4553/13317 [46:55<1:29:40,  1.63it/s]
 34%|███▍      | 4554/13317 [46:56<1:29:41,  1.63it/s]
                                                      
{'loss': 0.2282, 'grad_norm': 0.6562013626098633, 'learning_rate': 8.320251741511795e-06, 'epoch': 1.03}

 34%|███▍      | 4554/13317 [46:56<1:29:41,  1.63it/s]
 34%|███▍      | 4555/13317 [46:56<1:29:37,  1.63it/s]
                                                      
{'loss': 0.2415, 'grad_norm': 0.5297108888626099, 'learning_rate': 8.3192716814031e-06, 'epoch': 1.03}

 34%|███▍      | 4555/13317 [46:57<1:29:37,  1.63it/s]
 34%|███▍      | 4556/13317 [46:57<1:29:39,  1.63it/s]
                                                      
{'loss': 0.1951, 'grad_norm': 0.5499701499938965, 'learning_rate': 8.318291393225298e-06, 'epoch': 1.03}

 34%|███▍      | 4556/13317 [46:57<1:29:39,  1.63it/s]
 34%|███▍      | 4557/13317 [46:58<1:29:35,  1.63it/s]
                                                      
{'loss': 0.2733, 'grad_norm': 0.619601845741272, 'learning_rate': 8.317310877045742e-06, 'epoch': 1.03}

 34%|███▍      | 4557/13317 [46:58<1:29:35,  1.63it/s]
 34%|███▍      | 4558/13317 [46:58<1:29:33,  1.63it/s]
                                                      
{'loss': 0.2006, 'grad_norm': 0.44229066371917725, 'learning_rate': 8.316330132931805e-06, 'epoch': 1.03}

 34%|███▍      | 4558/13317 [46:58<1:29:33,  1.63it/s]
 34%|███▍      | 4559/13317 [46:59<1:29:33,  1.63it/s]
                                                      
{'loss': 0.1542, 'grad_norm': 0.4634668827056885, 'learning_rate': 8.315349160950872e-06, 'epoch': 1.03}

 34%|███▍      | 4559/13317 [46:59<1:29:33,  1.63it/s]
 34%|███▍      | 4560/13317 [46:59<1:29:32,  1.63it/s]
                                                      
{'loss': 0.232, 'grad_norm': 0.5464982986450195, 'learning_rate': 8.314367961170348e-06, 'epoch': 1.03}

 34%|███▍      | 4560/13317 [47:00<1:29:32,  1.63it/s]
 34%|███▍      | 4561/13317 [47:00<1:29:33,  1.63it/s]
                                                      
{'loss': 0.1833, 'grad_norm': 0.4657825529575348, 'learning_rate': 8.313386533657654e-06, 'epoch': 1.03}

 34%|███▍      | 4561/13317 [47:00<1:29:33,  1.63it/s]
 34%|███▍      | 4562/13317 [47:01<1:29:32,  1.63it/s]
                                                      
{'loss': 0.1934, 'grad_norm': 0.4535703659057617, 'learning_rate': 8.312404878480222e-06, 'epoch': 1.03}

 34%|███▍      | 4562/13317 [47:01<1:29:32,  1.63it/s]
 34%|███▍      | 4563/13317 [47:01<1:29:32,  1.63it/s]
                                                      
{'loss': 0.1743, 'grad_norm': 0.4897025227546692, 'learning_rate': 8.311422995705503e-06, 'epoch': 1.03}

 34%|███▍      | 4563/13317 [47:01<1:29:32,  1.63it/s]
 34%|███▍      | 4564/13317 [47:02<1:29:29,  1.63it/s]
                                                      
{'loss': 0.1665, 'grad_norm': 0.4890996217727661, 'learning_rate': 8.310440885400962e-06, 'epoch': 1.03}

 34%|███▍      | 4564/13317 [47:02<1:29:29,  1.63it/s]
 34%|███▍      | 4565/13317 [47:03<1:29:27,  1.63it/s]
                                                      
{'loss': 0.1711, 'grad_norm': 0.4607105255126953, 'learning_rate': 8.309458547634079e-06, 'epoch': 1.03}

 34%|███▍      | 4565/13317 [47:03<1:29:27,  1.63it/s]
 34%|███▍      | 4566/13317 [47:03<1:29:28,  1.63it/s]
                                                      
{'loss': 0.2088, 'grad_norm': 0.45436879992485046, 'learning_rate': 8.308475982472353e-06, 'epoch': 1.03}

 34%|███▍      | 4566/13317 [47:03<1:29:28,  1.63it/s]
 34%|███▍      | 4567/13317 [47:04<1:29:28,  1.63it/s]
                                                      
{'loss': 0.258, 'grad_norm': 0.5889075994491577, 'learning_rate': 8.307493189983297e-06, 'epoch': 1.03}

 34%|███▍      | 4567/13317 [47:04<1:29:28,  1.63it/s]
 34%|███▍      | 4568/13317 [47:04<1:29:28,  1.63it/s]
                                                      
{'loss': 0.2312, 'grad_norm': 0.5806073546409607, 'learning_rate': 8.30651017023444e-06, 'epoch': 1.03}

 34%|███▍      | 4568/13317 [47:05<1:29:28,  1.63it/s]
 34%|███▍      | 4569/13317 [47:05<1:29:27,  1.63it/s]
                                                      
{'loss': 0.1867, 'grad_norm': 0.46545442938804626, 'learning_rate': 8.305526923293322e-06, 'epoch': 1.03}

 34%|███▍      | 4569/13317 [47:05<1:29:27,  1.63it/s]
 34%|███▍      | 4570/13317 [47:06<1:29:25,  1.63it/s]
                                                      
{'loss': 0.2086, 'grad_norm': 0.506873369216919, 'learning_rate': 8.304543449227507e-06, 'epoch': 1.03}

 34%|███▍      | 4570/13317 [47:06<1:29:25,  1.63it/s]
 34%|███▍      | 4571/13317 [47:06<1:29:28,  1.63it/s]
                                                      
{'loss': 0.2126, 'grad_norm': 0.5545065402984619, 'learning_rate': 8.303559748104567e-06, 'epoch': 1.03}

 34%|███▍      | 4571/13317 [47:06<1:29:28,  1.63it/s]
 34%|███▍      | 4572/13317 [47:07<1:29:29,  1.63it/s]
                                                      
{'loss': 0.1516, 'grad_norm': 0.3924790024757385, 'learning_rate': 8.302575819992094e-06, 'epoch': 1.03}

 34%|███▍      | 4572/13317 [47:07<1:29:29,  1.63it/s]
 34%|███▍      | 4573/13317 [47:07<1:29:28,  1.63it/s]
                                                      
{'loss': 0.1848, 'grad_norm': 0.5345160961151123, 'learning_rate': 8.301591664957693e-06, 'epoch': 1.03}

 34%|███▍      | 4573/13317 [47:08<1:29:28,  1.63it/s]
 34%|███▍      | 4574/13317 [47:08<1:29:25,  1.63it/s]
                                                      
{'loss': 0.1415, 'grad_norm': 0.35839924216270447, 'learning_rate': 8.300607283068986e-06, 'epoch': 1.03}

 34%|███▍      | 4574/13317 [47:08<1:29:25,  1.63it/s]
 34%|███▍      | 4575/13317 [47:09<1:29:33,  1.63it/s]
                                                      
{'loss': 0.1892, 'grad_norm': 0.4585776925086975, 'learning_rate': 8.299622674393615e-06, 'epoch': 1.03}

 34%|███▍      | 4575/13317 [47:09<1:29:33,  1.63it/s]
 34%|███▍      | 4576/13317 [47:09<1:29:29,  1.63it/s]
                                                      
{'loss': 0.2068, 'grad_norm': 0.5727800130844116, 'learning_rate': 8.298637838999226e-06, 'epoch': 1.03}

 34%|███▍      | 4576/13317 [47:09<1:29:29,  1.63it/s]
 34%|███▍      | 4577/13317 [47:10<1:29:28,  1.63it/s]
                                                      
{'loss': 0.1803, 'grad_norm': 0.5228682160377502, 'learning_rate': 8.29765277695349e-06, 'epoch': 1.03}

 34%|███▍      | 4577/13317 [47:10<1:29:28,  1.63it/s]
 34%|███▍      | 4578/13317 [47:11<1:29:28,  1.63it/s]
                                                      
{'loss': 0.1982, 'grad_norm': 0.4925968647003174, 'learning_rate': 8.296667488324095e-06, 'epoch': 1.03}

 34%|███▍      | 4578/13317 [47:11<1:29:28,  1.63it/s]
 34%|███▍      | 4579/13317 [47:11<1:29:25,  1.63it/s]
                                                      
{'loss': 0.2025, 'grad_norm': 0.5361250042915344, 'learning_rate': 8.295681973178737e-06, 'epoch': 1.03}

 34%|███▍      | 4579/13317 [47:11<1:29:25,  1.63it/s]
 34%|███▍      | 4580/13317 [47:12<1:29:21,  1.63it/s]
                                                      
{'loss': 0.1318, 'grad_norm': 0.4226740896701813, 'learning_rate': 8.294696231585133e-06, 'epoch': 1.03}

 34%|███▍      | 4580/13317 [47:12<1:29:21,  1.63it/s]
 34%|███▍      | 4581/13317 [47:12<1:29:24,  1.63it/s]
                                                      
{'loss': 0.1618, 'grad_norm': 0.47883591055870056, 'learning_rate': 8.293710263611013e-06, 'epoch': 1.03}

 34%|███▍      | 4581/13317 [47:12<1:29:24,  1.63it/s]
 34%|███▍      | 4582/13317 [47:13<1:29:22,  1.63it/s]
                                                      
{'loss': 0.2273, 'grad_norm': 0.5787410736083984, 'learning_rate': 8.29272406932412e-06, 'epoch': 1.03}

 34%|███▍      | 4582/13317 [47:13<1:29:22,  1.63it/s]
 34%|███▍      | 4583/13317 [47:14<1:29:26,  1.63it/s]
                                                      
{'loss': 0.1872, 'grad_norm': 0.48206397891044617, 'learning_rate': 8.291737648792223e-06, 'epoch': 1.03}

 34%|███▍      | 4583/13317 [47:14<1:29:26,  1.63it/s]
 34%|███▍      | 4584/13317 [47:14<1:29:27,  1.63it/s]
                                                      
{'loss': 0.1778, 'grad_norm': 0.4940735399723053, 'learning_rate': 8.290751002083096e-06, 'epoch': 1.03}

 34%|███▍      | 4584/13317 [47:14<1:29:27,  1.63it/s]
 34%|███▍      | 4585/13317 [47:15<1:29:22,  1.63it/s]
                                                      
{'loss': 0.1213, 'grad_norm': 0.355602502822876, 'learning_rate': 8.289764129264532e-06, 'epoch': 1.03}

 34%|███▍      | 4585/13317 [47:15<1:29:22,  1.63it/s]
 34%|███▍      | 4586/13317 [47:15<1:29:23,  1.63it/s]
                                                      
{'loss': 0.2884, 'grad_norm': 0.7002478241920471, 'learning_rate': 8.288777030404337e-06, 'epoch': 1.03}

 34%|███▍      | 4586/13317 [47:16<1:29:23,  1.63it/s]
 34%|███▍      | 4587/13317 [47:16<1:29:20,  1.63it/s]
                                                      
{'loss': 0.2202, 'grad_norm': 0.5829792618751526, 'learning_rate': 8.287789705570341e-06, 'epoch': 1.03}

 34%|███▍      | 4587/13317 [47:16<1:29:20,  1.63it/s]
 34%|███▍      | 4588/13317 [47:17<1:29:16,  1.63it/s]
                                                      
{'loss': 0.197, 'grad_norm': 0.5619451403617859, 'learning_rate': 8.28680215483038e-06, 'epoch': 1.03}

 34%|███▍      | 4588/13317 [47:17<1:29:16,  1.63it/s]
 34%|███▍      | 4589/13317 [47:17<1:29:13,  1.63it/s]
                                                      
{'loss': 0.1944, 'grad_norm': 0.5113399624824524, 'learning_rate': 8.28581437825231e-06, 'epoch': 1.03}

 34%|███▍      | 4589/13317 [47:17<1:29:13,  1.63it/s]
 34%|███▍      | 4590/13317 [47:18<1:29:14,  1.63it/s]
                                                      
{'loss': 0.2303, 'grad_norm': 0.561598539352417, 'learning_rate': 8.284826375904e-06, 'epoch': 1.03}

 34%|███▍      | 4590/13317 [47:18<1:29:14,  1.63it/s]
 34%|███▍      | 4591/13317 [47:18<1:29:14,  1.63it/s]
                                                      
{'loss': 0.2096, 'grad_norm': 0.5168554186820984, 'learning_rate': 8.283838147853339e-06, 'epoch': 1.03}

 34%|███▍      | 4591/13317 [47:19<1:29:14,  1.63it/s]
 34%|███▍      | 4592/13317 [47:19<1:29:12,  1.63it/s]
                                                      
{'loss': 0.1909, 'grad_norm': 0.5586295127868652, 'learning_rate': 8.282849694168227e-06, 'epoch': 1.03}

 34%|███▍      | 4592/13317 [47:19<1:29:12,  1.63it/s]
 34%|███▍      | 4593/13317 [47:20<1:29:15,  1.63it/s]
                                                      
{'loss': 0.1403, 'grad_norm': 0.4347788393497467, 'learning_rate': 8.28186101491658e-06, 'epoch': 1.03}

 34%|███▍      | 4593/13317 [47:20<1:29:15,  1.63it/s]
 34%|███▍      | 4594/13317 [47:20<1:29:14,  1.63it/s]
                                                      
{'loss': 0.2586, 'grad_norm': 0.6886800527572632, 'learning_rate': 8.280872110166334e-06, 'epoch': 1.03}

 34%|███▍      | 4594/13317 [47:20<1:29:14,  1.63it/s]
 35%|███▍      | 4595/13317 [47:21<1:29:12,  1.63it/s]
                                                      
{'loss': 0.1883, 'grad_norm': 0.5008813142776489, 'learning_rate': 8.279882979985433e-06, 'epoch': 1.04}

 35%|███▍      | 4595/13317 [47:21<1:29:12,  1.63it/s]
 35%|███▍      | 4596/13317 [47:22<1:29:12,  1.63it/s]
                                                      
{'loss': 0.2691, 'grad_norm': 0.48183465003967285, 'learning_rate': 8.278893624441849e-06, 'epoch': 1.04}

 35%|███▍      | 4596/13317 [47:22<1:29:12,  1.63it/s]
 35%|███▍      | 4597/13317 [47:22<1:29:12,  1.63it/s]
                                                      
{'loss': 0.1913, 'grad_norm': 0.4873618483543396, 'learning_rate': 8.27790404360355e-06, 'epoch': 1.04}

 35%|███▍      | 4597/13317 [47:22<1:29:12,  1.63it/s]
 35%|███▍      | 4598/13317 [47:23<1:29:09,  1.63it/s]
                                                      
{'loss': 0.2277, 'grad_norm': 0.5748814940452576, 'learning_rate': 8.27691423753854e-06, 'epoch': 1.04}

 35%|███▍      | 4598/13317 [47:23<1:29:09,  1.63it/s]
 35%|███▍      | 4599/13317 [47:23<1:29:09,  1.63it/s]
                                                      
{'loss': 0.2358, 'grad_norm': 0.5504359006881714, 'learning_rate': 8.275924206314824e-06, 'epoch': 1.04}

 35%|███▍      | 4599/13317 [47:24<1:29:09,  1.63it/s]
 35%|███▍      | 4600/13317 [47:24<1:29:13,  1.63it/s]
                                                      
{'loss': 0.2238, 'grad_norm': 0.5306659936904907, 'learning_rate': 8.274933950000428e-06, 'epoch': 1.04}

 35%|███▍      | 4600/13317 [47:24<1:29:13,  1.63it/s]
 35%|███▍      | 4601/13317 [47:25<1:29:11,  1.63it/s]
                                                      
{'loss': 0.1699, 'grad_norm': 0.5245964527130127, 'learning_rate': 8.273943468663394e-06, 'epoch': 1.04}

 35%|███▍      | 4601/13317 [47:25<1:29:11,  1.63it/s]
 35%|███▍      | 4602/13317 [47:25<1:29:10,  1.63it/s]
                                                      
{'loss': 0.204, 'grad_norm': 0.486086905002594, 'learning_rate': 8.272952762371781e-06, 'epoch': 1.04}

 35%|███▍      | 4602/13317 [47:25<1:29:10,  1.63it/s]
 35%|███▍      | 4603/13317 [47:26<1:29:08,  1.63it/s]
                                                      
{'loss': 0.2467, 'grad_norm': 0.514229953289032, 'learning_rate': 8.271961831193656e-06, 'epoch': 1.04}

 35%|███▍      | 4603/13317 [47:26<1:29:08,  1.63it/s]
 35%|███▍      | 4604/13317 [47:26<1:29:06,  1.63it/s]
                                                      
{'loss': 0.1806, 'grad_norm': 0.521964967250824, 'learning_rate': 8.270970675197112e-06, 'epoch': 1.04}

 35%|███▍      | 4604/13317 [47:27<1:29:06,  1.63it/s]
 35%|███▍      | 4605/13317 [47:27<1:29:07,  1.63it/s]
                                                      
{'loss': 0.2111, 'grad_norm': 0.4948841631412506, 'learning_rate': 8.269979294450245e-06, 'epoch': 1.04}

 35%|███▍      | 4605/13317 [47:27<1:29:07,  1.63it/s]
 35%|███▍      | 4606/13317 [47:28<1:29:05,  1.63it/s]
                                                      
{'loss': 0.1406, 'grad_norm': 0.5080676078796387, 'learning_rate': 8.268987689021181e-06, 'epoch': 1.04}

 35%|███▍      | 4606/13317 [47:28<1:29:05,  1.63it/s]
 35%|███▍      | 4607/13317 [47:28<1:29:01,  1.63it/s]
                                                      
{'loss': 0.1936, 'grad_norm': 0.5151017308235168, 'learning_rate': 8.267995858978049e-06, 'epoch': 1.04}

 35%|███▍      | 4607/13317 [47:28<1:29:01,  1.63it/s]
 35%|███▍      | 4608/13317 [47:29<1:29:02,  1.63it/s]
                                                      
{'loss': 0.1854, 'grad_norm': 0.46766161918640137, 'learning_rate': 8.267003804389e-06, 'epoch': 1.04}

 35%|███▍      | 4608/13317 [47:29<1:29:02,  1.63it/s]
 35%|███▍      | 4609/13317 [47:30<1:29:01,  1.63it/s]
                                                      
{'loss': 0.1843, 'grad_norm': 0.5064752101898193, 'learning_rate': 8.266011525322197e-06, 'epoch': 1.04}

 35%|███▍      | 4609/13317 [47:30<1:29:01,  1.63it/s]
 35%|███▍      | 4610/13317 [47:30<1:29:03,  1.63it/s]
                                                      
{'loss': 0.2462, 'grad_norm': 0.5789831876754761, 'learning_rate': 8.265019021845822e-06, 'epoch': 1.04}

 35%|███▍      | 4610/13317 [47:30<1:29:03,  1.63it/s]
 35%|███▍      | 4611/13317 [47:31<1:29:00,  1.63it/s]
                                                      
{'loss': 0.1963, 'grad_norm': 0.4683550298213959, 'learning_rate': 8.264026294028067e-06, 'epoch': 1.04}

 35%|███▍      | 4611/13317 [47:31<1:29:00,  1.63it/s]
 35%|███▍      | 4612/13317 [47:31<1:29:01,  1.63it/s]
                                                      
{'loss': 0.336, 'grad_norm': 0.5895355939865112, 'learning_rate': 8.263033341937149e-06, 'epoch': 1.04}

 35%|███▍      | 4612/13317 [47:32<1:29:01,  1.63it/s]
 35%|███▍      | 4613/13317 [47:32<1:29:00,  1.63it/s]
                                                      
{'loss': 0.2122, 'grad_norm': 0.5199176669120789, 'learning_rate': 8.262040165641288e-06, 'epoch': 1.04}

 35%|███▍      | 4613/13317 [47:32<1:29:00,  1.63it/s]
 35%|███▍      | 4614/13317 [47:33<1:29:00,  1.63it/s]
                                                      
{'loss': 0.2479, 'grad_norm': 0.571369469165802, 'learning_rate': 8.261046765208731e-06, 'epoch': 1.04}

 35%|███▍      | 4614/13317 [47:33<1:29:00,  1.63it/s]
 35%|███▍      | 4615/13317 [47:33<1:28:58,  1.63it/s]
                                                      
{'loss': 0.3354, 'grad_norm': 1.047892689704895, 'learning_rate': 8.260053140707731e-06, 'epoch': 1.04}

 35%|███▍      | 4615/13317 [47:33<1:28:58,  1.63it/s]
 35%|███▍      | 4616/13317 [47:34<1:28:57,  1.63it/s]
                                                      
{'loss': 0.2239, 'grad_norm': 0.5014448165893555, 'learning_rate': 8.259059292206563e-06, 'epoch': 1.04}

 35%|███▍      | 4616/13317 [47:34<1:28:57,  1.63it/s]
 35%|███▍      | 4617/13317 [47:34<1:28:55,  1.63it/s]
                                                      
{'loss': 0.204, 'grad_norm': 0.5629199147224426, 'learning_rate': 8.258065219773514e-06, 'epoch': 1.04}

 35%|███▍      | 4617/13317 [47:35<1:28:55,  1.63it/s]
 35%|███▍      | 4618/13317 [47:35<1:28:54,  1.63it/s]
                                                      
{'loss': 0.1811, 'grad_norm': 0.48508378863334656, 'learning_rate': 8.257070923476889e-06, 'epoch': 1.04}

 35%|███▍      | 4618/13317 [47:35<1:28:54,  1.63it/s]
 35%|███▍      | 4619/13317 [47:36<1:28:52,  1.63it/s]
                                                      
{'loss': 0.235, 'grad_norm': 0.5642158389091492, 'learning_rate': 8.256076403385002e-06, 'epoch': 1.04}

 35%|███▍      | 4619/13317 [47:36<1:28:52,  1.63it/s]
 35%|███▍      | 4620/13317 [47:36<1:28:53,  1.63it/s]
                                                      
{'loss': 0.2623, 'grad_norm': 0.5917423963546753, 'learning_rate': 8.255081659566194e-06, 'epoch': 1.04}

 35%|███▍      | 4620/13317 [47:36<1:28:53,  1.63it/s]
 35%|███▍      | 4621/13317 [47:37<1:28:53,  1.63it/s]
                                                      
{'loss': 0.1803, 'grad_norm': 0.4501723051071167, 'learning_rate': 8.254086692088808e-06, 'epoch': 1.04}

 35%|███▍      | 4621/13317 [47:37<1:28:53,  1.63it/s]
 35%|███▍      | 4622/13317 [47:38<1:28:52,  1.63it/s]
                                                      
{'loss': 0.171, 'grad_norm': 0.5041755437850952, 'learning_rate': 8.25309150102121e-06, 'epoch': 1.04}

 35%|███▍      | 4622/13317 [47:38<1:28:52,  1.63it/s]
 35%|███▍      | 4623/13317 [47:38<1:28:53,  1.63it/s]
                                                      
{'loss': 0.1909, 'grad_norm': 0.5972818732261658, 'learning_rate': 8.252096086431785e-06, 'epoch': 1.04}

 35%|███▍      | 4623/13317 [47:38<1:28:53,  1.63it/s]
 35%|███▍      | 4624/13317 [47:39<1:28:54,  1.63it/s]
                                                      
{'loss': 0.1773, 'grad_norm': 0.4490830898284912, 'learning_rate': 8.251100448388925e-06, 'epoch': 1.04}

 35%|███▍      | 4624/13317 [47:39<1:28:54,  1.63it/s]
 35%|███▍      | 4625/13317 [47:39<1:28:56,  1.63it/s]
                                                      
{'loss': 0.1908, 'grad_norm': 0.5361659526824951, 'learning_rate': 8.250104586961039e-06, 'epoch': 1.04}

 35%|███▍      | 4625/13317 [47:39<1:28:56,  1.63it/s]
 35%|███▍      | 4626/13317 [47:40<1:28:56,  1.63it/s]
                                                      
{'loss': 0.1911, 'grad_norm': 0.5177487730979919, 'learning_rate': 8.249108502216557e-06, 'epoch': 1.04}

 35%|███▍      | 4626/13317 [47:40<1:28:56,  1.63it/s]
 35%|███▍      | 4627/13317 [47:41<1:28:57,  1.63it/s]
                                                      
{'loss': 0.1945, 'grad_norm': 0.5339378714561462, 'learning_rate': 8.248112194223916e-06, 'epoch': 1.04}

 35%|███▍      | 4627/13317 [47:41<1:28:57,  1.63it/s]
 35%|███▍      | 4628/13317 [47:41<1:28:57,  1.63it/s]
                                                      
{'loss': 0.1813, 'grad_norm': 0.5151357650756836, 'learning_rate': 8.247115663051577e-06, 'epoch': 1.04}

 35%|███▍      | 4628/13317 [47:41<1:28:57,  1.63it/s]
 35%|███▍      | 4629/13317 [47:42<1:28:49,  1.63it/s]
                                                      
{'loss': 0.1812, 'grad_norm': 0.5031183362007141, 'learning_rate': 8.246118908768012e-06, 'epoch': 1.04}

 35%|███▍      | 4629/13317 [47:42<1:28:49,  1.63it/s]
 35%|███▍      | 4630/13317 [47:42<1:28:49,  1.63it/s]
                                                      
{'loss': 0.1994, 'grad_norm': 0.5455935001373291, 'learning_rate': 8.245121931441706e-06, 'epoch': 1.04}

 35%|███▍      | 4630/13317 [47:43<1:28:49,  1.63it/s]
 35%|███▍      | 4631/13317 [47:43<1:28:49,  1.63it/s]
                                                      
{'loss': 0.1943, 'grad_norm': 0.4408615231513977, 'learning_rate': 8.244124731141166e-06, 'epoch': 1.04}

 35%|███▍      | 4631/13317 [47:43<1:28:49,  1.63it/s]
 35%|███▍      | 4632/13317 [47:44<1:28:49,  1.63it/s]
                                                      
{'loss': 0.175, 'grad_norm': 0.4625474214553833, 'learning_rate': 8.243127307934904e-06, 'epoch': 1.04}

 35%|███▍      | 4632/13317 [47:44<1:28:49,  1.63it/s]
 35%|███▍      | 4633/13317 [47:44<1:28:47,  1.63it/s]
                                                      
{'loss': 0.2245, 'grad_norm': 0.5042829513549805, 'learning_rate': 8.24212966189146e-06, 'epoch': 1.04}

 35%|███▍      | 4633/13317 [47:44<1:28:47,  1.63it/s]
 35%|███▍      | 4634/13317 [47:45<1:28:49,  1.63it/s]
                                                      
{'loss': 0.215, 'grad_norm': 0.4786224663257599, 'learning_rate': 8.241131793079377e-06, 'epoch': 1.04}

 35%|███▍      | 4634/13317 [47:45<1:28:49,  1.63it/s]
 35%|███▍      | 4635/13317 [47:45<1:28:46,  1.63it/s]
                                                      
{'loss': 0.2091, 'grad_norm': 0.508030354976654, 'learning_rate': 8.240133701567225e-06, 'epoch': 1.04}

 35%|███▍      | 4635/13317 [47:46<1:28:46,  1.63it/s]
 35%|███▍      | 4636/13317 [47:46<1:28:44,  1.63it/s]
                                                      
{'loss': 0.1872, 'grad_norm': 0.49772909283638, 'learning_rate': 8.239135387423581e-06, 'epoch': 1.04}

 35%|███▍      | 4636/13317 [47:46<1:28:44,  1.63it/s]
 35%|███▍      | 4637/13317 [47:47<1:28:43,  1.63it/s]
                                                      
{'loss': 0.182, 'grad_norm': 0.48721328377723694, 'learning_rate': 8.238136850717038e-06, 'epoch': 1.04}

 35%|███▍      | 4637/13317 [47:47<1:28:43,  1.63it/s]
 35%|███▍      | 4638/13317 [47:47<1:28:42,  1.63it/s]
                                                      
{'loss': 0.1943, 'grad_norm': 0.5246625542640686, 'learning_rate': 8.237138091516206e-06, 'epoch': 1.04}

 35%|███▍      | 4638/13317 [47:47<1:28:42,  1.63it/s]
 35%|███▍      | 4639/13317 [47:48<1:28:41,  1.63it/s]
                                                      
{'loss': 0.181, 'grad_norm': 0.4820536673069, 'learning_rate': 8.236139109889713e-06, 'epoch': 1.05}

 35%|███▍      | 4639/13317 [47:48<1:28:41,  1.63it/s]
 35%|███▍      | 4640/13317 [47:49<1:28:40,  1.63it/s]
                                                      
{'loss': 0.2286, 'grad_norm': 0.6638768911361694, 'learning_rate': 8.2351399059062e-06, 'epoch': 1.05}

 35%|███▍      | 4640/13317 [47:49<1:28:40,  1.63it/s]
 35%|███▍      | 4641/13317 [47:49<1:28:42,  1.63it/s]
                                                      
{'loss': 0.2035, 'grad_norm': 0.4760710299015045, 'learning_rate': 8.23414047963432e-06, 'epoch': 1.05}

 35%|███▍      | 4641/13317 [47:49<1:28:42,  1.63it/s]
 35%|███▍      | 4642/13317 [47:50<1:28:50,  1.63it/s]
                                                      
{'loss': 0.2052, 'grad_norm': 0.4950070083141327, 'learning_rate': 8.233140831142745e-06, 'epoch': 1.05}

 35%|███▍      | 4642/13317 [47:50<1:28:50,  1.63it/s]
 35%|███▍      | 4643/13317 [47:50<1:28:47,  1.63it/s]
                                                      
{'loss': 0.1955, 'grad_norm': 0.49545490741729736, 'learning_rate': 8.232140960500163e-06, 'epoch': 1.05}

 35%|███▍      | 4643/13317 [47:51<1:28:47,  1.63it/s]
 35%|███▍      | 4644/13317 [47:51<1:28:45,  1.63it/s]
                                                      
{'loss': 0.2409, 'grad_norm': 0.5668522715568542, 'learning_rate': 8.231140867775274e-06, 'epoch': 1.05}

 35%|███▍      | 4644/13317 [47:51<1:28:45,  1.63it/s]
 35%|███▍      | 4645/13317 [47:52<1:28:44,  1.63it/s]
                                                      
{'loss': 0.1813, 'grad_norm': 0.4947565793991089, 'learning_rate': 8.230140553036794e-06, 'epoch': 1.05}

 35%|███▍      | 4645/13317 [47:52<1:28:44,  1.63it/s]
 35%|███▍      | 4646/13317 [47:52<1:28:44,  1.63it/s]
                                                      
{'loss': 0.2105, 'grad_norm': 0.567600667476654, 'learning_rate': 8.229140016353458e-06, 'epoch': 1.05}

 35%|███▍      | 4646/13317 [47:52<1:28:44,  1.63it/s]
 35%|███▍      | 4647/13317 [47:53<1:28:45,  1.63it/s]
                                                      
{'loss': 0.1805, 'grad_norm': 0.4397186040878296, 'learning_rate': 8.228139257794012e-06, 'epoch': 1.05}

 35%|███▍      | 4647/13317 [47:53<1:28:45,  1.63it/s]
 35%|███▍      | 4648/13317 [47:53<1:28:40,  1.63it/s]
                                                      
{'loss': 0.2216, 'grad_norm': 0.5108648538589478, 'learning_rate': 8.22713827742722e-06, 'epoch': 1.05}

 35%|███▍      | 4648/13317 [47:54<1:28:40,  1.63it/s]
 35%|███▍      | 4649/13317 [47:54<1:28:38,  1.63it/s]
                                                      
{'loss': 0.2228, 'grad_norm': 0.4978594481945038, 'learning_rate': 8.226137075321858e-06, 'epoch': 1.05}

 35%|███▍      | 4649/13317 [47:54<1:28:38,  1.63it/s]
 35%|███▍      | 4650/13317 [47:55<1:28:37,  1.63it/s]
                                                      
{'loss': 0.255, 'grad_norm': 0.5196581482887268, 'learning_rate': 8.22513565154672e-06, 'epoch': 1.05}

 35%|███▍      | 4650/13317 [47:55<1:28:37,  1.63it/s]
 35%|███▍      | 4651/13317 [47:55<1:28:36,  1.63it/s]
                                                      
{'loss': 0.2065, 'grad_norm': 0.5346856117248535, 'learning_rate': 8.224134006170614e-06, 'epoch': 1.05}

 35%|███▍      | 4651/13317 [47:55<1:28:36,  1.63it/s]
 35%|███▍      | 4652/13317 [47:56<1:28:35,  1.63it/s]
                                                      
{'loss': 0.2392, 'grad_norm': 0.5456160306930542, 'learning_rate': 8.223132139262364e-06, 'epoch': 1.05}

 35%|███▍      | 4652/13317 [47:56<1:28:35,  1.63it/s]
 35%|███▍      | 4653/13317 [47:57<1:28:34,  1.63it/s]
                                                      
{'loss': 0.2386, 'grad_norm': 0.5694148540496826, 'learning_rate': 8.222130050890811e-06, 'epoch': 1.05}

 35%|███▍      | 4653/13317 [47:57<1:28:34,  1.63it/s]
 35%|███▍      | 4654/13317 [47:57<1:28:32,  1.63it/s]
                                                      
{'loss': 0.2091, 'grad_norm': 0.5654889941215515, 'learning_rate': 8.221127741124805e-06, 'epoch': 1.05}

 35%|███▍      | 4654/13317 [47:57<1:28:32,  1.63it/s]
 35%|███▍      | 4655/13317 [47:58<1:28:35,  1.63it/s]
                                                      
{'loss': 0.1901, 'grad_norm': 0.5122562050819397, 'learning_rate': 8.220125210033218e-06, 'epoch': 1.05}

 35%|███▍      | 4655/13317 [47:58<1:28:35,  1.63it/s]
 35%|███▍      | 4656/13317 [47:58<1:28:30,  1.63it/s]
                                                      
{'loss': 0.1541, 'grad_norm': 0.4810231029987335, 'learning_rate': 8.219122457684933e-06, 'epoch': 1.05}

 35%|███▍      | 4656/13317 [47:59<1:28:30,  1.63it/s]
 35%|███▍      | 4657/13317 [47:59<1:28:32,  1.63it/s]
                                                      
{'loss': 0.2077, 'grad_norm': 0.5164848566055298, 'learning_rate': 8.218119484148853e-06, 'epoch': 1.05}

 35%|███▍      | 4657/13317 [47:59<1:28:32,  1.63it/s]
 35%|███▍      | 4658/13317 [48:00<1:28:31,  1.63it/s]
                                                      
{'loss': 0.2101, 'grad_norm': 0.5746199488639832, 'learning_rate': 8.217116289493888e-06, 'epoch': 1.05}

 35%|███▍      | 4658/13317 [48:00<1:28:31,  1.63it/s]
 35%|███▍      | 4659/13317 [48:00<1:28:29,  1.63it/s]
                                                      
{'loss': 0.1986, 'grad_norm': 0.54184889793396, 'learning_rate': 8.21611287378897e-06, 'epoch': 1.05}

 35%|███▍      | 4659/13317 [48:00<1:28:29,  1.63it/s]
 35%|███▍      | 4660/13317 [48:01<1:28:27,  1.63it/s]
                                                      
{'loss': 0.217, 'grad_norm': 0.49323955178260803, 'learning_rate': 8.215109237103047e-06, 'epoch': 1.05}

 35%|███▍      | 4660/13317 [48:01<1:28:27,  1.63it/s]
 35%|███▌      | 4661/13317 [48:01<1:28:28,  1.63it/s]
                                                      
{'loss': 0.2364, 'grad_norm': 0.5332724452018738, 'learning_rate': 8.214105379505076e-06, 'epoch': 1.05}

 35%|███▌      | 4661/13317 [48:02<1:28:28,  1.63it/s]
 35%|███▌      | 4662/13317 [48:02<1:28:27,  1.63it/s]
                                                      
{'loss': 0.2294, 'grad_norm': 0.5095627903938293, 'learning_rate': 8.213101301064035e-06, 'epoch': 1.05}

 35%|███▌      | 4662/13317 [48:02<1:28:27,  1.63it/s]
 35%|███▌      | 4663/13317 [48:03<1:28:26,  1.63it/s]
                                                      
{'loss': 0.2035, 'grad_norm': 0.4919247329235077, 'learning_rate': 8.212097001848913e-06, 'epoch': 1.05}

 35%|███▌      | 4663/13317 [48:03<1:28:26,  1.63it/s]
 35%|███▌      | 4664/13317 [48:03<1:28:26,  1.63it/s]
                                                      
{'loss': 0.2126, 'grad_norm': 0.465984046459198, 'learning_rate': 8.211092481928716e-06, 'epoch': 1.05}

 35%|███▌      | 4664/13317 [48:03<1:28:26,  1.63it/s]
 35%|███▌      | 4665/13317 [48:04<1:28:25,  1.63it/s]
                                                      
{'loss': 0.2644, 'grad_norm': 0.6229833364486694, 'learning_rate': 8.210087741372467e-06, 'epoch': 1.05}

 35%|███▌      | 4665/13317 [48:04<1:28:25,  1.63it/s]
 35%|███▌      | 4666/13317 [48:05<1:28:23,  1.63it/s]
                                                      
{'loss': 0.2131, 'grad_norm': 0.5395835041999817, 'learning_rate': 8.2090827802492e-06, 'epoch': 1.05}

 35%|███▌      | 4666/13317 [48:05<1:28:23,  1.63it/s]
 35%|███▌      | 4667/13317 [48:05<1:28:29,  1.63it/s]
                                                      
{'loss': 0.2085, 'grad_norm': 0.5109013319015503, 'learning_rate': 8.20807759862797e-06, 'epoch': 1.05}

 35%|███▌      | 4667/13317 [48:05<1:28:29,  1.63it/s]
 35%|███▌      | 4668/13317 [48:06<1:28:29,  1.63it/s]
                                                      
{'loss': 0.2639, 'grad_norm': 0.5555973649024963, 'learning_rate': 8.20707219657784e-06, 'epoch': 1.05}

 35%|███▌      | 4668/13317 [48:06<1:28:29,  1.63it/s]
 35%|███▌      | 4669/13317 [48:06<1:28:28,  1.63it/s]
                                                      
{'loss': 0.1641, 'grad_norm': 0.43984532356262207, 'learning_rate': 8.206066574167893e-06, 'epoch': 1.05}

 35%|███▌      | 4669/13317 [48:06<1:28:28,  1.63it/s]
 35%|███▌      | 4670/13317 [48:07<1:28:28,  1.63it/s]
                                                      
{'loss': 0.2411, 'grad_norm': 0.5245784521102905, 'learning_rate': 8.205060731467225e-06, 'epoch': 1.05}

 35%|███▌      | 4670/13317 [48:07<1:28:28,  1.63it/s]
 35%|███▌      | 4671/13317 [48:08<1:28:24,  1.63it/s]
                                                      
{'loss': 0.1483, 'grad_norm': 0.426492303609848, 'learning_rate': 8.20405466854495e-06, 'epoch': 1.05}

 35%|███▌      | 4671/13317 [48:08<1:28:24,  1.63it/s]
 35%|███▌      | 4672/13317 [48:08<1:28:24,  1.63it/s]
                                                      
{'loss': 0.1817, 'grad_norm': 0.4494453966617584, 'learning_rate': 8.203048385470195e-06, 'epoch': 1.05}

 35%|███▌      | 4672/13317 [48:08<1:28:24,  1.63it/s]
 35%|███▌      | 4673/13317 [48:09<1:28:23,  1.63it/s]
                                                      
{'loss': 0.1521, 'grad_norm': 0.45992299914360046, 'learning_rate': 8.2020418823121e-06, 'epoch': 1.05}

 35%|███▌      | 4673/13317 [48:09<1:28:23,  1.63it/s]
 35%|███▌      | 4674/13317 [48:09<1:28:22,  1.63it/s]
                                                      
{'loss': 0.1531, 'grad_norm': 0.40997132658958435, 'learning_rate': 8.201035159139826e-06, 'epoch': 1.05}

 35%|███▌      | 4674/13317 [48:10<1:28:22,  1.63it/s]
 35%|███▌      | 4675/13317 [48:10<1:28:23,  1.63it/s]
                                                      
{'loss': 0.2161, 'grad_norm': 0.535888135433197, 'learning_rate': 8.200028216022542e-06, 'epoch': 1.05}

 35%|███▌      | 4675/13317 [48:10<1:28:23,  1.63it/s]
 35%|███▌      | 4676/13317 [48:11<1:28:22,  1.63it/s]
                                                      
{'loss': 0.2541, 'grad_norm': 0.5660127997398376, 'learning_rate': 8.199021053029438e-06, 'epoch': 1.05}

 35%|███▌      | 4676/13317 [48:11<1:28:22,  1.63it/s]
 35%|███▌      | 4677/13317 [48:11<1:28:22,  1.63it/s]
                                                      
{'loss': 0.1809, 'grad_norm': 0.543117105960846, 'learning_rate': 8.198013670229713e-06, 'epoch': 1.05}

 35%|███▌      | 4677/13317 [48:11<1:28:22,  1.63it/s]
 35%|███▌      | 4678/13317 [48:12<1:28:17,  1.63it/s]
                                                      
{'loss': 0.2192, 'grad_norm': 0.5215288400650024, 'learning_rate': 8.19700606769259e-06, 'epoch': 1.05}

 35%|███▌      | 4678/13317 [48:12<1:28:17,  1.63it/s]
 35%|███▌      | 4679/13317 [48:12<1:28:17,  1.63it/s]
                                                      
{'loss': 0.2017, 'grad_norm': 0.5518963932991028, 'learning_rate': 8.1959982454873e-06, 'epoch': 1.05}

 35%|███▌      | 4679/13317 [48:13<1:28:17,  1.63it/s]
 35%|███▌      | 4680/13317 [48:13<1:28:20,  1.63it/s]
                                                      
{'loss': 0.2088, 'grad_norm': 0.4629516303539276, 'learning_rate': 8.19499020368309e-06, 'epoch': 1.05}

 35%|███▌      | 4680/13317 [48:13<1:28:20,  1.63it/s]
 35%|███▌      | 4681/13317 [48:14<1:28:21,  1.63it/s]
                                                      
{'loss': 0.2329, 'grad_norm': 0.5213044881820679, 'learning_rate': 8.193981942349224e-06, 'epoch': 1.05}

 35%|███▌      | 4681/13317 [48:14<1:28:21,  1.63it/s]
 35%|███▌      | 4682/13317 [48:14<1:28:21,  1.63it/s]
                                                      
{'loss': 0.2016, 'grad_norm': 0.5137191414833069, 'learning_rate': 8.19297346155498e-06, 'epoch': 1.05}

 35%|███▌      | 4682/13317 [48:14<1:28:21,  1.63it/s]
 35%|███▌      | 4683/13317 [48:15<1:28:21,  1.63it/s]
                                                      
{'loss': 0.2495, 'grad_norm': 0.6772386431694031, 'learning_rate': 8.191964761369651e-06, 'epoch': 1.05}

 35%|███▌      | 4683/13317 [48:15<1:28:21,  1.63it/s]
 35%|███▌      | 4684/13317 [48:16<1:28:18,  1.63it/s]
                                                      
{'loss': 0.1864, 'grad_norm': 0.49565836787223816, 'learning_rate': 8.190955841862548e-06, 'epoch': 1.06}

 35%|███▌      | 4684/13317 [48:16<1:28:18,  1.63it/s]
 35%|███▌      | 4685/13317 [48:16<1:28:24,  1.63it/s]
                                                      
{'loss': 0.1623, 'grad_norm': 0.5212209820747375, 'learning_rate': 8.189946703102991e-06, 'epoch': 1.06}

 35%|███▌      | 4685/13317 [48:16<1:28:24,  1.63it/s]
 35%|███▌      | 4686/13317 [48:17<1:28:22,  1.63it/s]
                                                      
{'loss': 0.1845, 'grad_norm': 0.4927549660205841, 'learning_rate': 8.188937345160319e-06, 'epoch': 1.06}

 35%|███▌      | 4686/13317 [48:17<1:28:22,  1.63it/s]
 35%|███▌      | 4687/13317 [48:17<1:28:20,  1.63it/s]
                                                      
{'loss': 0.265, 'grad_norm': 0.5191597938537598, 'learning_rate': 8.187927768103887e-06, 'epoch': 1.06}

 35%|███▌      | 4687/13317 [48:18<1:28:20,  1.63it/s]
 35%|███▌      | 4688/13317 [48:18<1:28:19,  1.63it/s]
                                                      
{'loss': 0.2099, 'grad_norm': 0.5430375337600708, 'learning_rate': 8.186917972003063e-06, 'epoch': 1.06}

 35%|███▌      | 4688/13317 [48:18<1:28:19,  1.63it/s]
 35%|███▌      | 4689/13317 [48:19<1:28:17,  1.63it/s]
                                                      
{'loss': 0.166, 'grad_norm': 0.49786388874053955, 'learning_rate': 8.185907956927232e-06, 'epoch': 1.06}

 35%|███▌      | 4689/13317 [48:19<1:28:17,  1.63it/s]
 35%|███▌      | 4690/13317 [48:19<1:28:24,  1.63it/s]
                                                      
{'loss': 0.1678, 'grad_norm': 0.4029639959335327, 'learning_rate': 8.18489772294579e-06, 'epoch': 1.06}

 35%|███▌      | 4690/13317 [48:19<1:28:24,  1.63it/s]
 35%|███▌      | 4691/13317 [48:20<1:28:21,  1.63it/s]
                                                      
{'loss': 0.1366, 'grad_norm': 0.4113495349884033, 'learning_rate': 8.183887270128152e-06, 'epoch': 1.06}

 35%|███▌      | 4691/13317 [48:20<1:28:21,  1.63it/s]
 35%|███▌      | 4692/13317 [48:20<1:28:17,  1.63it/s]
                                                      
{'loss': 0.1351, 'grad_norm': 0.4131748378276825, 'learning_rate': 8.182876598543747e-06, 'epoch': 1.06}

 35%|███▌      | 4692/13317 [48:21<1:28:17,  1.63it/s]
 35%|███▌      | 4693/13317 [48:21<1:28:13,  1.63it/s]
                                                      
{'loss': 0.1731, 'grad_norm': 0.5191983580589294, 'learning_rate': 8.18186570826202e-06, 'epoch': 1.06}

 35%|███▌      | 4693/13317 [48:21<1:28:13,  1.63it/s]
 35%|███▌      | 4694/13317 [48:22<1:28:11,  1.63it/s]
                                                      
{'loss': 0.1966, 'grad_norm': 0.5059981942176819, 'learning_rate': 8.180854599352428e-06, 'epoch': 1.06}

 35%|███▌      | 4694/13317 [48:22<1:28:11,  1.63it/s]
 35%|███▌      | 4695/13317 [48:22<1:28:14,  1.63it/s]
                                                      
{'loss': 0.2781, 'grad_norm': 0.6666429042816162, 'learning_rate': 8.179843271884444e-06, 'epoch': 1.06}

 35%|███▌      | 4695/13317 [48:22<1:28:14,  1.63it/s]
 35%|███▌      | 4696/13317 [48:23<1:28:11,  1.63it/s]
                                                      
{'loss': 0.1888, 'grad_norm': 0.6113094091415405, 'learning_rate': 8.178831725927561e-06, 'epoch': 1.06}

 35%|███▌      | 4696/13317 [48:23<1:28:11,  1.63it/s]
 35%|███▌      | 4697/13317 [48:24<1:28:09,  1.63it/s]
                                                      
{'loss': 0.1448, 'grad_norm': 0.3982335031032562, 'learning_rate': 8.177819961551278e-06, 'epoch': 1.06}

 35%|███▌      | 4697/13317 [48:24<1:28:09,  1.63it/s]
 35%|███▌      | 4698/13317 [48:24<1:28:08,  1.63it/s]
                                                      
{'loss': 0.1678, 'grad_norm': 0.47618407011032104, 'learning_rate': 8.17680797882512e-06, 'epoch': 1.06}

 35%|███▌      | 4698/13317 [48:24<1:28:08,  1.63it/s]
 35%|███▌      | 4699/13317 [48:25<1:28:07,  1.63it/s]
                                                      
{'loss': 0.1869, 'grad_norm': 0.48365116119384766, 'learning_rate': 8.175795777818614e-06, 'epoch': 1.06}

 35%|███▌      | 4699/13317 [48:25<1:28:07,  1.63it/s]
 35%|███▌      | 4700/13317 [48:25<1:28:16,  1.63it/s]
                                                      
{'loss': 0.1878, 'grad_norm': 0.48578208684921265, 'learning_rate': 8.174783358601312e-06, 'epoch': 1.06}

 35%|███▌      | 4700/13317 [48:26<1:28:16,  1.63it/s]
 35%|███▌      | 4701/13317 [48:26<1:28:13,  1.63it/s]
                                                      
{'loss': 0.256, 'grad_norm': 0.5655255913734436, 'learning_rate': 8.17377072124278e-06, 'epoch': 1.06}

 35%|███▌      | 4701/13317 [48:26<1:28:13,  1.63it/s]
 35%|███▌      | 4702/13317 [48:27<1:28:10,  1.63it/s]
                                                      
{'loss': 0.2241, 'grad_norm': 0.5796880125999451, 'learning_rate': 8.172757865812596e-06, 'epoch': 1.06}

 35%|███▌      | 4702/13317 [48:27<1:28:10,  1.63it/s]
 35%|███▌      | 4703/13317 [48:27<1:28:08,  1.63it/s]
                                                      
{'loss': 0.2281, 'grad_norm': 0.49908381700515747, 'learning_rate': 8.171744792380349e-06, 'epoch': 1.06}

 35%|███▌      | 4703/13317 [48:27<1:28:08,  1.63it/s]
 35%|███▌      | 4704/13317 [48:28<1:28:06,  1.63it/s]
                                                      
{'loss': 0.171, 'grad_norm': 0.4399251341819763, 'learning_rate': 8.170731501015656e-06, 'epoch': 1.06}

 35%|███▌      | 4704/13317 [48:28<1:28:06,  1.63it/s]
 35%|███▌      | 4705/13317 [48:28<1:28:08,  1.63it/s]
                                                      
{'loss': 0.1748, 'grad_norm': 0.4969809949398041, 'learning_rate': 8.169717991788134e-06, 'epoch': 1.06}

 35%|███▌      | 4705/13317 [48:29<1:28:08,  1.63it/s]
 35%|███▌      | 4706/13317 [48:29<1:28:05,  1.63it/s]
                                                      
{'loss': 0.2342, 'grad_norm': 0.6106952428817749, 'learning_rate': 8.168704264767425e-06, 'epoch': 1.06}

 35%|███▌      | 4706/13317 [48:29<1:28:05,  1.63it/s]
 35%|███▌      | 4707/13317 [48:30<1:28:05,  1.63it/s]
                                                      
{'loss': 0.1545, 'grad_norm': 0.5664235949516296, 'learning_rate': 8.167690320023183e-06, 'epoch': 1.06}

 35%|███▌      | 4707/13317 [48:30<1:28:05,  1.63it/s]
 35%|███▌      | 4708/13317 [48:30<1:28:04,  1.63it/s]
                                                      
{'loss': 0.1919, 'grad_norm': 0.5283043384552002, 'learning_rate': 8.166676157625075e-06, 'epoch': 1.06}

 35%|███▌      | 4708/13317 [48:30<1:28:04,  1.63it/s]
 35%|███▌      | 4709/13317 [48:31<1:28:04,  1.63it/s]
                                                      
{'loss': 0.1578, 'grad_norm': 0.4766448140144348, 'learning_rate': 8.165661777642787e-06, 'epoch': 1.06}

 35%|███▌      | 4709/13317 [48:31<1:28:04,  1.63it/s]
 35%|███▌      | 4710/13317 [48:32<1:28:11,  1.63it/s]
                                                      
{'loss': 0.1839, 'grad_norm': 0.46902740001678467, 'learning_rate': 8.164647180146015e-06, 'epoch': 1.06}

 35%|███▌      | 4710/13317 [48:32<1:28:11,  1.63it/s]
 35%|███▌      | 4711/13317 [48:32<1:28:08,  1.63it/s]
                                                      
{'loss': 0.2123, 'grad_norm': 0.5262748003005981, 'learning_rate': 8.163632365204474e-06, 'epoch': 1.06}

 35%|███▌      | 4711/13317 [48:32<1:28:08,  1.63it/s]
 35%|███▌      | 4712/13317 [48:33<1:28:09,  1.63it/s]
                                                      
{'loss': 0.1838, 'grad_norm': 0.46300938725471497, 'learning_rate': 8.162617332887894e-06, 'epoch': 1.06}

 35%|███▌      | 4712/13317 [48:33<1:28:09,  1.63it/s]
 35%|███▌      | 4713/13317 [48:33<1:28:09,  1.63it/s]
                                                      
{'loss': 0.1694, 'grad_norm': 0.45138099789619446, 'learning_rate': 8.161602083266014e-06, 'epoch': 1.06}

 35%|███▌      | 4713/13317 [48:33<1:28:09,  1.63it/s]
 35%|███▌      | 4714/13317 [48:34<1:28:02,  1.63it/s]
                                                      
{'loss': 0.2444, 'grad_norm': 0.5852292776107788, 'learning_rate': 8.160586616408598e-06, 'epoch': 1.06}

 35%|███▌      | 4714/13317 [48:34<1:28:02,  1.63it/s]
 35%|███▌      | 4715/13317 [48:35<1:28:04,  1.63it/s]
                                                      
{'loss': 0.2467, 'grad_norm': 0.5017590522766113, 'learning_rate': 8.159570932385415e-06, 'epoch': 1.06}

 35%|███▌      | 4715/13317 [48:35<1:28:04,  1.63it/s]
 35%|███▌      | 4716/13317 [48:35<1:28:01,  1.63it/s]
                                                      
{'loss': 0.1977, 'grad_norm': 0.4976670742034912, 'learning_rate': 8.158555031266255e-06, 'epoch': 1.06}

 35%|███▌      | 4716/13317 [48:35<1:28:01,  1.63it/s]
 35%|███▌      | 4717/13317 [48:36<1:27:58,  1.63it/s]
                                                      
{'loss': 0.1892, 'grad_norm': 0.5060605406761169, 'learning_rate': 8.157538913120921e-06, 'epoch': 1.06}

 35%|███▌      | 4717/13317 [48:36<1:27:58,  1.63it/s]
 35%|███▌      | 4718/13317 [48:36<1:27:56,  1.63it/s]
                                                      
{'loss': 0.1961, 'grad_norm': 0.5037752389907837, 'learning_rate': 8.156522578019233e-06, 'epoch': 1.06}

 35%|███▌      | 4718/13317 [48:37<1:27:56,  1.63it/s]
 35%|███▌      | 4719/13317 [48:37<1:27:55,  1.63it/s]
                                                      
{'loss': 0.2034, 'grad_norm': 0.5404770374298096, 'learning_rate': 8.15550602603102e-06, 'epoch': 1.06}

 35%|███▌      | 4719/13317 [48:37<1:27:55,  1.63it/s]
 35%|███▌      | 4720/13317 [48:38<1:28:19,  1.62it/s]
                                                      
{'loss': 0.2071, 'grad_norm': 0.5035728216171265, 'learning_rate': 8.154489257226133e-06, 'epoch': 1.06}

 35%|███▌      | 4720/13317 [48:38<1:28:19,  1.62it/s]
 35%|███▌      | 4721/13317 [48:38<1:28:11,  1.62it/s]
                                                      
{'loss': 0.1803, 'grad_norm': 0.5133404731750488, 'learning_rate': 8.153472271674435e-06, 'epoch': 1.06}

 35%|███▌      | 4721/13317 [48:38<1:28:11,  1.62it/s]
 35%|███▌      | 4722/13317 [48:39<1:28:06,  1.63it/s]
                                                      
{'loss': 0.1906, 'grad_norm': 0.5459001064300537, 'learning_rate': 8.1524550694458e-06, 'epoch': 1.06}

 35%|███▌      | 4722/13317 [48:39<1:28:06,  1.63it/s]
 35%|███▌      | 4723/13317 [48:40<1:28:02,  1.63it/s]
                                                      
{'loss': 0.243, 'grad_norm': 0.548539400100708, 'learning_rate': 8.151437650610124e-06, 'epoch': 1.06}

 35%|███▌      | 4723/13317 [48:40<1:28:02,  1.63it/s]
 35%|███▌      | 4724/13317 [48:40<1:27:58,  1.63it/s]
                                                      
{'loss': 0.2482, 'grad_norm': 0.6090409159660339, 'learning_rate': 8.150420015237312e-06, 'epoch': 1.06}

 35%|███▌      | 4724/13317 [48:40<1:27:58,  1.63it/s]
 35%|███▌      | 4725/13317 [48:41<1:27:59,  1.63it/s]
                                                      
{'loss': 0.2233, 'grad_norm': 0.5515599846839905, 'learning_rate': 8.149402163397288e-06, 'epoch': 1.06}

 35%|███▌      | 4725/13317 [48:41<1:27:59,  1.63it/s]
 35%|███▌      | 4726/13317 [48:41<1:27:54,  1.63it/s]
                                                      
{'loss': 0.1831, 'grad_norm': 0.5358877182006836, 'learning_rate': 8.148384095159992e-06, 'epoch': 1.06}

 35%|███▌      | 4726/13317 [48:41<1:27:54,  1.63it/s]
 35%|███▌      | 4727/13317 [48:42<1:27:56,  1.63it/s]
                                                      
{'loss': 0.1529, 'grad_norm': 0.5135536193847656, 'learning_rate': 8.14736581059537e-06, 'epoch': 1.06}

 35%|███▌      | 4727/13317 [48:42<1:27:56,  1.63it/s]
 36%|███▌      | 4728/13317 [48:43<1:27:54,  1.63it/s]
                                                      
{'loss': 0.1792, 'grad_norm': 0.4118278920650482, 'learning_rate': 8.146347309773393e-06, 'epoch': 1.07}

 36%|███▌      | 4728/13317 [48:43<1:27:54,  1.63it/s]
 36%|███▌      | 4729/13317 [48:43<1:27:50,  1.63it/s]
                                                      
{'loss': 0.2015, 'grad_norm': 0.5785236358642578, 'learning_rate': 8.145328592764044e-06, 'epoch': 1.07}

 36%|███▌      | 4729/13317 [48:43<1:27:50,  1.63it/s]
 36%|███▌      | 4730/13317 [48:44<1:27:53,  1.63it/s]
                                                      
{'loss': 0.1948, 'grad_norm': 0.5357170701026917, 'learning_rate': 8.144309659637313e-06, 'epoch': 1.07}

 36%|███▌      | 4730/13317 [48:44<1:27:53,  1.63it/s]
 36%|███▌      | 4731/13317 [48:44<1:27:52,  1.63it/s]
                                                      
{'loss': 0.2571, 'grad_norm': 0.6317289471626282, 'learning_rate': 8.14329051046322e-06, 'epoch': 1.07}

 36%|███▌      | 4731/13317 [48:45<1:27:52,  1.63it/s]
 36%|███▌      | 4732/13317 [48:45<1:27:50,  1.63it/s]
                                                      
{'loss': 0.2712, 'grad_norm': 0.554287850856781, 'learning_rate': 8.142271145311784e-06, 'epoch': 1.07}

 36%|███▌      | 4732/13317 [48:45<1:27:50,  1.63it/s]
 36%|███▌      | 4733/13317 [48:46<1:27:48,  1.63it/s]
                                                      
{'loss': 0.2392, 'grad_norm': 0.5487728714942932, 'learning_rate': 8.14125156425305e-06, 'epoch': 1.07}

 36%|███▌      | 4733/13317 [48:46<1:27:48,  1.63it/s]
 36%|███▌      | 4734/13317 [48:46<1:27:46,  1.63it/s]
                                                      
{'loss': 0.2149, 'grad_norm': 0.6507749557495117, 'learning_rate': 8.140231767357075e-06, 'epoch': 1.07}

 36%|███▌      | 4734/13317 [48:46<1:27:46,  1.63it/s]
 36%|███▌      | 4735/13317 [48:47<1:27:47,  1.63it/s]
                                                      
{'loss': 0.2089, 'grad_norm': 0.6133325695991516, 'learning_rate': 8.139211754693928e-06, 'epoch': 1.07}

 36%|███▌      | 4735/13317 [48:47<1:27:47,  1.63it/s]
 36%|███▌      | 4736/13317 [48:47<1:27:42,  1.63it/s]
                                                      
{'loss': 0.1719, 'grad_norm': 0.5251728892326355, 'learning_rate': 8.138191526333695e-06, 'epoch': 1.07}

 36%|███▌      | 4736/13317 [48:48<1:27:42,  1.63it/s]
 36%|███▌      | 4737/13317 [48:48<1:27:45,  1.63it/s]
                                                      
{'loss': 0.1906, 'grad_norm': 0.480032354593277, 'learning_rate': 8.137171082346474e-06, 'epoch': 1.07}

 36%|███▌      | 4737/13317 [48:48<1:27:45,  1.63it/s]
 36%|███▌      | 4738/13317 [48:49<1:27:44,  1.63it/s]
                                                      
{'loss': 0.2603, 'grad_norm': 0.6397773027420044, 'learning_rate': 8.136150422802384e-06, 'epoch': 1.07}

 36%|███▌      | 4738/13317 [48:49<1:27:44,  1.63it/s]
 36%|███▌      | 4739/13317 [48:49<1:27:42,  1.63it/s]
                                                      
{'loss': 0.1922, 'grad_norm': 0.4780716896057129, 'learning_rate': 8.135129547771557e-06, 'epoch': 1.07}

 36%|███▌      | 4739/13317 [48:49<1:27:42,  1.63it/s]
 36%|███▌      | 4740/13317 [48:50<1:27:47,  1.63it/s]
                                                      
{'loss': 0.268, 'grad_norm': 0.5651827454566956, 'learning_rate': 8.134108457324131e-06, 'epoch': 1.07}

 36%|███▌      | 4740/13317 [48:50<1:27:47,  1.63it/s]
 36%|███▌      | 4741/13317 [48:51<1:27:44,  1.63it/s]
                                                      
{'loss': 0.2023, 'grad_norm': 0.5133894085884094, 'learning_rate': 8.13308715153027e-06, 'epoch': 1.07}

 36%|███▌      | 4741/13317 [48:51<1:27:44,  1.63it/s]
 36%|███▌      | 4742/13317 [48:51<1:27:44,  1.63it/s]
                                                      
{'loss': 0.1524, 'grad_norm': 0.5460180044174194, 'learning_rate': 8.13206563046015e-06, 'epoch': 1.07}

 36%|███▌      | 4742/13317 [48:51<1:27:44,  1.63it/s]
 36%|███▌      | 4743/13317 [48:52<1:27:42,  1.63it/s]
                                                      
{'loss': 0.203, 'grad_norm': 0.6032525300979614, 'learning_rate': 8.131043894183956e-06, 'epoch': 1.07}

 36%|███▌      | 4743/13317 [48:52<1:27:42,  1.63it/s]
 36%|███▌      | 4744/13317 [48:52<1:27:41,  1.63it/s]
                                                      
{'loss': 0.1669, 'grad_norm': 0.5283722281455994, 'learning_rate': 8.130021942771895e-06, 'epoch': 1.07}

 36%|███▌      | 4744/13317 [48:53<1:27:41,  1.63it/s]
 36%|███▌      | 4745/13317 [48:53<1:27:43,  1.63it/s]
                                                      
{'loss': 0.1895, 'grad_norm': 0.5115729570388794, 'learning_rate': 8.128999776294187e-06, 'epoch': 1.07}

 36%|███▌      | 4745/13317 [48:53<1:27:43,  1.63it/s]
 36%|███▌      | 4746/13317 [48:54<1:27:39,  1.63it/s]
                                                      
{'loss': 0.1906, 'grad_norm': 0.5224975943565369, 'learning_rate': 8.127977394821062e-06, 'epoch': 1.07}

 36%|███▌      | 4746/13317 [48:54<1:27:39,  1.63it/s]
 36%|███▌      | 4747/13317 [48:54<1:27:40,  1.63it/s]
                                                      
{'loss': 0.1925, 'grad_norm': 0.5185752511024475, 'learning_rate': 8.126954798422773e-06, 'epoch': 1.07}

 36%|███▌      | 4747/13317 [48:54<1:27:40,  1.63it/s]
 36%|███▌      | 4748/13317 [48:55<1:27:40,  1.63it/s]
                                                      
{'loss': 0.2231, 'grad_norm': 0.5735016465187073, 'learning_rate': 8.125931987169579e-06, 'epoch': 1.07}

 36%|███▌      | 4748/13317 [48:55<1:27:40,  1.63it/s]
 36%|███▌      | 4749/13317 [48:55<1:27:39,  1.63it/s]
                                                      
{'loss': 0.2053, 'grad_norm': 0.4969252943992615, 'learning_rate': 8.124908961131759e-06, 'epoch': 1.07}

 36%|███▌      | 4749/13317 [48:56<1:27:39,  1.63it/s]
 36%|███▌      | 4750/13317 [48:56<1:27:41,  1.63it/s]
                                                      
{'loss': 0.2359, 'grad_norm': 0.5552588105201721, 'learning_rate': 8.123885720379606e-06, 'epoch': 1.07}

 36%|███▌      | 4750/13317 [48:56<1:27:41,  1.63it/s]
 36%|███▌      | 4751/13317 [48:57<1:27:37,  1.63it/s]
                                                      
{'loss': 0.2424, 'grad_norm': 0.6540660262107849, 'learning_rate': 8.122862264983428e-06, 'epoch': 1.07}

 36%|███▌      | 4751/13317 [48:57<1:27:37,  1.63it/s]
 36%|███▌      | 4752/13317 [48:57<1:27:35,  1.63it/s]
                                                      
{'loss': 0.2335, 'grad_norm': 0.6033342480659485, 'learning_rate': 8.121838595013549e-06, 'epoch': 1.07}

 36%|███▌      | 4752/13317 [48:57<1:27:35,  1.63it/s]
 36%|███▌      | 4753/13317 [48:58<1:27:39,  1.63it/s]
                                                      
{'loss': 0.212, 'grad_norm': 0.4985070824623108, 'learning_rate': 8.120814710540302e-06, 'epoch': 1.07}

 36%|███▌      | 4753/13317 [48:58<1:27:39,  1.63it/s]
 36%|███▌      | 4754/13317 [48:59<1:27:36,  1.63it/s]
                                                      
{'loss': 0.1882, 'grad_norm': 0.5016540288925171, 'learning_rate': 8.119790611634042e-06, 'epoch': 1.07}

 36%|███▌      | 4754/13317 [48:59<1:27:36,  1.63it/s]
 36%|███▌      | 4755/13317 [48:59<1:27:40,  1.63it/s]
                                                      
{'loss': 0.2189, 'grad_norm': 0.491219699382782, 'learning_rate': 8.118766298365133e-06, 'epoch': 1.07}

 36%|███▌      | 4755/13317 [48:59<1:27:40,  1.63it/s]
 36%|███▌      | 4756/13317 [49:00<1:27:37,  1.63it/s]
                                                      
{'loss': 0.2637, 'grad_norm': 0.5949490070343018, 'learning_rate': 8.117741770803957e-06, 'epoch': 1.07}

 36%|███▌      | 4756/13317 [49:00<1:27:37,  1.63it/s]
 36%|███▌      | 4757/13317 [49:00<1:27:34,  1.63it/s]
                                                      
{'loss': 0.2112, 'grad_norm': 0.5240169167518616, 'learning_rate': 8.11671702902091e-06, 'epoch': 1.07}

 36%|███▌      | 4757/13317 [49:01<1:27:34,  1.63it/s]
 36%|███▌      | 4758/13317 [49:01<1:27:32,  1.63it/s]
                                                      
{'loss': 0.2184, 'grad_norm': 0.569298505783081, 'learning_rate': 8.115692073086405e-06, 'epoch': 1.07}

 36%|███▌      | 4758/13317 [49:01<1:27:32,  1.63it/s]
 36%|███▌      | 4759/13317 [49:02<1:27:32,  1.63it/s]
                                                      
{'loss': 0.2258, 'grad_norm': 0.5636802315711975, 'learning_rate': 8.114666903070862e-06, 'epoch': 1.07}

 36%|███▌      | 4759/13317 [49:02<1:27:32,  1.63it/s]
 36%|███▌      | 4760/13317 [49:02<1:27:34,  1.63it/s]
                                                      
{'loss': 0.1923, 'grad_norm': 0.5043460130691528, 'learning_rate': 8.113641519044728e-06, 'epoch': 1.07}

 36%|███▌      | 4760/13317 [49:02<1:27:34,  1.63it/s]
 36%|███▌      | 4761/13317 [49:03<1:27:29,  1.63it/s]
                                                      
{'loss': 0.2365, 'grad_norm': 0.5512215495109558, 'learning_rate': 8.11261592107845e-06, 'epoch': 1.07}

 36%|███▌      | 4761/13317 [49:03<1:27:29,  1.63it/s]
 36%|███▌      | 4762/13317 [49:03<1:27:29,  1.63it/s]
                                                      
{'loss': 0.2019, 'grad_norm': 0.5364555716514587, 'learning_rate': 8.111590109242505e-06, 'epoch': 1.07}

 36%|███▌      | 4762/13317 [49:04<1:27:29,  1.63it/s]
 36%|███▌      | 4763/13317 [49:04<1:27:26,  1.63it/s]
                                                      
{'loss': 0.2441, 'grad_norm': 0.5830269455909729, 'learning_rate': 8.110564083607371e-06, 'epoch': 1.07}

 36%|███▌      | 4763/13317 [49:04<1:27:26,  1.63it/s]
 36%|███▌      | 4764/13317 [49:05<1:27:23,  1.63it/s]
                                                      
{'loss': 0.2169, 'grad_norm': 0.5603758096694946, 'learning_rate': 8.10953784424355e-06, 'epoch': 1.07}

 36%|███▌      | 4764/13317 [49:05<1:27:23,  1.63it/s]
 36%|███▌      | 4765/13317 [49:05<1:27:22,  1.63it/s]
                                                      
{'loss': 0.1867, 'grad_norm': 0.4573252499103546, 'learning_rate': 8.108511391221554e-06, 'epoch': 1.07}

 36%|███▌      | 4765/13317 [49:05<1:27:22,  1.63it/s]
 36%|███▌      | 4766/13317 [49:06<1:27:22,  1.63it/s]
                                                      
{'loss': 0.2009, 'grad_norm': 0.6041606068611145, 'learning_rate': 8.107484724611912e-06, 'epoch': 1.07}

 36%|███▌      | 4766/13317 [49:06<1:27:22,  1.63it/s]
 36%|███▌      | 4767/13317 [49:07<1:27:23,  1.63it/s]
                                                      
{'loss': 0.2106, 'grad_norm': 0.539697527885437, 'learning_rate': 8.106457844485167e-06, 'epoch': 1.07}

 36%|███▌      | 4767/13317 [49:07<1:27:23,  1.63it/s]
 36%|███▌      | 4768/13317 [49:07<1:27:23,  1.63it/s]
                                                      
{'loss': 0.1724, 'grad_norm': 0.42588427662849426, 'learning_rate': 8.105430750911878e-06, 'epoch': 1.07}

 36%|███▌      | 4768/13317 [49:07<1:27:23,  1.63it/s]
 36%|███▌      | 4769/13317 [49:08<1:27:23,  1.63it/s]
                                                      
{'loss': 0.2668, 'grad_norm': 0.6389322280883789, 'learning_rate': 8.104403443962614e-06, 'epoch': 1.07}

 36%|███▌      | 4769/13317 [49:08<1:27:23,  1.63it/s]
 36%|███▌      | 4770/13317 [49:08<1:27:21,  1.63it/s]
                                                      
{'loss': 0.199, 'grad_norm': 0.4913303256034851, 'learning_rate': 8.103375923707963e-06, 'epoch': 1.07}

 36%|███▌      | 4770/13317 [49:08<1:27:21,  1.63it/s]
 36%|███▌      | 4771/13317 [49:09<1:27:21,  1.63it/s]
                                                      
{'loss': 0.2644, 'grad_norm': 0.6198660731315613, 'learning_rate': 8.102348190218528e-06, 'epoch': 1.07}

 36%|███▌      | 4771/13317 [49:09<1:27:21,  1.63it/s]
 36%|███▌      | 4772/13317 [49:10<1:27:24,  1.63it/s]
                                                      
{'loss': 0.2037, 'grad_norm': 0.5297249555587769, 'learning_rate': 8.101320243564923e-06, 'epoch': 1.08}

 36%|███▌      | 4772/13317 [49:10<1:27:24,  1.63it/s]
 36%|███▌      | 4773/13317 [49:10<1:27:23,  1.63it/s]
                                                      
{'loss': 0.241, 'grad_norm': 0.5870354771614075, 'learning_rate': 8.100292083817782e-06, 'epoch': 1.08}

 36%|███▌      | 4773/13317 [49:10<1:27:23,  1.63it/s]
 36%|███▌      | 4774/13317 [49:11<1:27:22,  1.63it/s]
                                                      
{'loss': 0.2079, 'grad_norm': 0.4813504219055176, 'learning_rate': 8.099263711047746e-06, 'epoch': 1.08}

 36%|███▌      | 4774/13317 [49:11<1:27:22,  1.63it/s]
 36%|███▌      | 4775/13317 [49:11<1:27:22,  1.63it/s]
                                                      
{'loss': 0.1661, 'grad_norm': 0.6178066730499268, 'learning_rate': 8.09823512532548e-06, 'epoch': 1.08}

 36%|███▌      | 4775/13317 [49:12<1:27:22,  1.63it/s]
 36%|███▌      | 4776/13317 [49:12<1:27:19,  1.63it/s]
                                                      
{'loss': 0.1637, 'grad_norm': 0.4457681179046631, 'learning_rate': 8.097206326721654e-06, 'epoch': 1.08}

 36%|███▌      | 4776/13317 [49:12<1:27:19,  1.63it/s]
 36%|███▌      | 4777/13317 [49:13<1:27:17,  1.63it/s]
                                                      
{'loss': 0.2191, 'grad_norm': 0.5275582075119019, 'learning_rate': 8.096177315306962e-06, 'epoch': 1.08}

 36%|███▌      | 4777/13317 [49:13<1:27:17,  1.63it/s]
 36%|███▌      | 4778/13317 [49:13<1:27:17,  1.63it/s]
                                                      
{'loss': 0.1633, 'grad_norm': 0.43217384815216064, 'learning_rate': 8.095148091152103e-06, 'epoch': 1.08}

 36%|███▌      | 4778/13317 [49:13<1:27:17,  1.63it/s]
 36%|███▌      | 4779/13317 [49:14<1:27:18,  1.63it/s]
                                                      
{'loss': 0.2134, 'grad_norm': 0.523327112197876, 'learning_rate': 8.094118654327799e-06, 'epoch': 1.08}

 36%|███▌      | 4779/13317 [49:14<1:27:18,  1.63it/s]
 36%|███▌      | 4780/13317 [49:14<1:27:18,  1.63it/s]
                                                      
{'loss': 0.2009, 'grad_norm': 0.5541602373123169, 'learning_rate': 8.093089004904784e-06, 'epoch': 1.08}

 36%|███▌      | 4780/13317 [49:15<1:27:18,  1.63it/s]
 36%|███▌      | 4781/13317 [49:15<1:27:17,  1.63it/s]
                                                      
{'loss': 0.2545, 'grad_norm': 0.6938455104827881, 'learning_rate': 8.092059142953803e-06, 'epoch': 1.08}

 36%|███▌      | 4781/13317 [49:15<1:27:17,  1.63it/s]
 36%|███▌      | 4782/13317 [49:16<1:27:18,  1.63it/s]
                                                      
{'loss': 0.145, 'grad_norm': 0.4155624508857727, 'learning_rate': 8.09102906854562e-06, 'epoch': 1.08}

 36%|███▌      | 4782/13317 [49:16<1:27:18,  1.63it/s]
 36%|███▌      | 4783/13317 [49:16<1:27:18,  1.63it/s]
                                                      
{'loss': 0.1912, 'grad_norm': 0.48018863797187805, 'learning_rate': 8.08999878175101e-06, 'epoch': 1.08}

 36%|███▌      | 4783/13317 [49:16<1:27:18,  1.63it/s]
 36%|███▌      | 4784/13317 [49:17<1:27:15,  1.63it/s]
                                                      
{'loss': 0.1715, 'grad_norm': 0.5545921325683594, 'learning_rate': 8.088968282640768e-06, 'epoch': 1.08}

 36%|███▌      | 4784/13317 [49:17<1:27:15,  1.63it/s]
 36%|███▌      | 4785/13317 [49:18<1:27:16,  1.63it/s]
                                                      
{'loss': 0.2325, 'grad_norm': 0.511447548866272, 'learning_rate': 8.087937571285698e-06, 'epoch': 1.08}

 36%|███▌      | 4785/13317 [49:18<1:27:16,  1.63it/s]
 36%|███▌      | 4786/13317 [49:18<1:27:14,  1.63it/s]
                                                      
{'loss': 0.1964, 'grad_norm': 0.538618266582489, 'learning_rate': 8.08690664775662e-06, 'epoch': 1.08}

 36%|███▌      | 4786/13317 [49:18<1:27:14,  1.63it/s]
 36%|███▌      | 4787/13317 [49:19<1:27:14,  1.63it/s]
                                                      
{'loss': 0.2151, 'grad_norm': 0.5567079186439514, 'learning_rate': 8.085875512124373e-06, 'epoch': 1.08}

 36%|███▌      | 4787/13317 [49:19<1:27:14,  1.63it/s]
 36%|███▌      | 4788/13317 [49:19<1:27:12,  1.63it/s]
                                                      
{'loss': 0.1795, 'grad_norm': 0.5855269432067871, 'learning_rate': 8.084844164459803e-06, 'epoch': 1.08}

 36%|███▌      | 4788/13317 [49:20<1:27:12,  1.63it/s]
 36%|███▌      | 4789/13317 [49:20<1:27:13,  1.63it/s]
                                                      
{'loss': 0.2112, 'grad_norm': 0.5438957810401917, 'learning_rate': 8.083812604833776e-06, 'epoch': 1.08}

 36%|███▌      | 4789/13317 [49:20<1:27:13,  1.63it/s]
 36%|███▌      | 4790/13317 [49:21<1:27:12,  1.63it/s]
                                                      
{'loss': 0.1882, 'grad_norm': 0.4947921335697174, 'learning_rate': 8.082780833317172e-06, 'epoch': 1.08}

 36%|███▌      | 4790/13317 [49:21<1:27:12,  1.63it/s]
 36%|███▌      | 4791/13317 [49:21<1:27:12,  1.63it/s]
                                                      
{'loss': 0.2219, 'grad_norm': 0.55435711145401, 'learning_rate': 8.081748849980883e-06, 'epoch': 1.08}

 36%|███▌      | 4791/13317 [49:21<1:27:12,  1.63it/s]
 36%|███▌      | 4792/13317 [49:22<1:27:10,  1.63it/s]
                                                      
{'loss': 0.2062, 'grad_norm': 0.5356965065002441, 'learning_rate': 8.08071665489582e-06, 'epoch': 1.08}

 36%|███▌      | 4792/13317 [49:22<1:27:10,  1.63it/s]
 36%|███▌      | 4793/13317 [49:22<1:27:11,  1.63it/s]
                                                      
{'loss': 0.1983, 'grad_norm': 0.5183491706848145, 'learning_rate': 8.079684248132899e-06, 'epoch': 1.08}

 36%|███▌      | 4793/13317 [49:23<1:27:11,  1.63it/s]
 36%|███▌      | 4794/13317 [49:23<1:27:12,  1.63it/s]
                                                      
{'loss': 0.2393, 'grad_norm': 0.6268991827964783, 'learning_rate': 8.078651629763066e-06, 'epoch': 1.08}

 36%|███▌      | 4794/13317 [49:23<1:27:12,  1.63it/s]
 36%|███▌      | 4795/13317 [49:24<1:27:11,  1.63it/s]
                                                      
{'loss': 0.2006, 'grad_norm': 0.46056100726127625, 'learning_rate': 8.077618799857269e-06, 'epoch': 1.08}

 36%|███▌      | 4795/13317 [49:24<1:27:11,  1.63it/s]
 36%|███▌      | 4796/13317 [49:24<1:27:12,  1.63it/s]
                                                      
{'loss': 0.1583, 'grad_norm': 0.5262221097946167, 'learning_rate': 8.076585758486471e-06, 'epoch': 1.08}

 36%|███▌      | 4796/13317 [49:24<1:27:12,  1.63it/s]
 36%|███▌      | 4797/13317 [49:25<1:27:12,  1.63it/s]
                                                      
{'loss': 0.2075, 'grad_norm': 0.49498558044433594, 'learning_rate': 8.075552505721658e-06, 'epoch': 1.08}

 36%|███▌      | 4797/13317 [49:25<1:27:12,  1.63it/s]
 36%|███▌      | 4798/13317 [49:26<1:27:11,  1.63it/s]
                                                      
{'loss': 0.2052, 'grad_norm': 0.45973584055900574, 'learning_rate': 8.074519041633823e-06, 'epoch': 1.08}

 36%|███▌      | 4798/13317 [49:26<1:27:11,  1.63it/s]
 36%|███▌      | 4799/13317 [49:26<1:27:12,  1.63it/s]
                                                      
{'loss': 0.2034, 'grad_norm': 0.47989898920059204, 'learning_rate': 8.073485366293978e-06, 'epoch': 1.08}

 36%|███▌      | 4799/13317 [49:26<1:27:12,  1.63it/s]
 36%|███▌      | 4800/13317 [49:27<1:27:16,  1.63it/s]
                                                      
{'loss': 0.1609, 'grad_norm': 0.45837995409965515, 'learning_rate': 8.072451479773143e-06, 'epoch': 1.08}

 36%|███▌      | 4800/13317 [49:27<1:27:16,  1.63it/s]
 36%|███▌      | 4801/13317 [49:27<1:27:13,  1.63it/s]
                                                      
{'loss': 0.1855, 'grad_norm': 0.46302103996276855, 'learning_rate': 8.071417382142362e-06, 'epoch': 1.08}

 36%|███▌      | 4801/13317 [49:28<1:27:13,  1.63it/s]
 36%|███▌      | 4802/13317 [49:28<1:27:12,  1.63it/s]
                                                      
{'loss': 0.2114, 'grad_norm': 0.5789350867271423, 'learning_rate': 8.070383073472685e-06, 'epoch': 1.08}

 36%|███▌      | 4802/13317 [49:28<1:27:12,  1.63it/s]
 36%|███▌      | 4803/13317 [49:29<1:27:07,  1.63it/s]
                                                      
{'loss': 0.1948, 'grad_norm': 0.5277194380760193, 'learning_rate': 8.06934855383518e-06, 'epoch': 1.08}

 36%|███▌      | 4803/13317 [49:29<1:27:07,  1.63it/s]
 36%|███▌      | 4804/13317 [49:29<1:27:05,  1.63it/s]
                                                      
{'loss': 0.1822, 'grad_norm': 0.5041102170944214, 'learning_rate': 8.06831382330093e-06, 'epoch': 1.08}

 36%|███▌      | 4804/13317 [49:29<1:27:05,  1.63it/s]
 36%|███▌      | 4805/13317 [49:30<1:27:10,  1.63it/s]
                                                      
{'loss': 0.2199, 'grad_norm': 0.525994062423706, 'learning_rate': 8.067278881941035e-06, 'epoch': 1.08}

 36%|███▌      | 4805/13317 [49:30<1:27:10,  1.63it/s]
 36%|███▌      | 4806/13317 [49:30<1:27:11,  1.63it/s]
                                                      
{'loss': 0.2628, 'grad_norm': 0.6050390601158142, 'learning_rate': 8.066243729826605e-06, 'epoch': 1.08}

 36%|███▌      | 4806/13317 [49:31<1:27:11,  1.63it/s]
 36%|███▌      | 4807/13317 [49:31<1:27:15,  1.63it/s]
                                                      
{'loss': 0.1725, 'grad_norm': 0.4782005846500397, 'learning_rate': 8.065208367028762e-06, 'epoch': 1.08}

 36%|███▌      | 4807/13317 [49:31<1:27:15,  1.63it/s]
 36%|███▌      | 4808/13317 [49:32<1:27:09,  1.63it/s]
                                                      
{'loss': 0.2375, 'grad_norm': 0.5630320310592651, 'learning_rate': 8.06417279361865e-06, 'epoch': 1.08}

 36%|███▌      | 4808/13317 [49:32<1:27:09,  1.63it/s]
 36%|███▌      | 4809/13317 [49:32<1:27:05,  1.63it/s]
                                                      
{'loss': 0.158, 'grad_norm': 0.4409019947052002, 'learning_rate': 8.063137009667423e-06, 'epoch': 1.08}

 36%|███▌      | 4809/13317 [49:32<1:27:05,  1.63it/s]
 36%|███▌      | 4810/13317 [49:33<1:27:09,  1.63it/s]
                                                      
{'loss': 0.2097, 'grad_norm': 0.4846598505973816, 'learning_rate': 8.06210101524625e-06, 'epoch': 1.08}

 36%|███▌      | 4810/13317 [49:33<1:27:09,  1.63it/s]
 36%|███▌      | 4811/13317 [49:34<1:27:05,  1.63it/s]
                                                      
{'loss': 0.2337, 'grad_norm': 0.6078984141349792, 'learning_rate': 8.061064810426317e-06, 'epoch': 1.08}

 36%|███▌      | 4811/13317 [49:34<1:27:05,  1.63it/s]
 36%|███▌      | 4812/13317 [49:34<1:27:01,  1.63it/s]
                                                      
{'loss': 0.1632, 'grad_norm': 0.49496975541114807, 'learning_rate': 8.06002839527882e-06, 'epoch': 1.08}

 36%|███▌      | 4812/13317 [49:34<1:27:01,  1.63it/s]
 36%|███▌      | 4813/13317 [49:35<1:27:00,  1.63it/s]
                                                      
{'loss': 0.2734, 'grad_norm': 0.6337580680847168, 'learning_rate': 8.05899176987497e-06, 'epoch': 1.08}

 36%|███▌      | 4813/13317 [49:35<1:27:00,  1.63it/s]
 36%|███▌      | 4814/13317 [49:35<1:26:56,  1.63it/s]
                                                      
{'loss': 0.1672, 'grad_norm': 0.42010498046875, 'learning_rate': 8.057954934285998e-06, 'epoch': 1.08}

 36%|███▌      | 4814/13317 [49:36<1:26:56,  1.63it/s]
 36%|███▌      | 4815/13317 [49:36<1:27:01,  1.63it/s]
                                                      
{'loss': 0.1707, 'grad_norm': 0.5008965730667114, 'learning_rate': 8.056917888583144e-06, 'epoch': 1.08}

 36%|███▌      | 4815/13317 [49:36<1:27:01,  1.63it/s]
 36%|███▌      | 4816/13317 [49:37<1:27:00,  1.63it/s]
                                                      
{'loss': 0.287, 'grad_norm': 0.6345281600952148, 'learning_rate': 8.055880632837663e-06, 'epoch': 1.08}

 36%|███▌      | 4816/13317 [49:37<1:27:00,  1.63it/s]
 36%|███▌      | 4817/13317 [49:37<1:26:54,  1.63it/s]
                                                      
{'loss': 0.2107, 'grad_norm': 0.5823284387588501, 'learning_rate': 8.054843167120827e-06, 'epoch': 1.09}

 36%|███▌      | 4817/13317 [49:37<1:26:54,  1.63it/s]
 36%|███▌      | 4818/13317 [49:38<1:26:56,  1.63it/s]
                                                      
{'loss': 0.1719, 'grad_norm': 0.48563963174819946, 'learning_rate': 8.053805491503921e-06, 'epoch': 1.09}

 36%|███▌      | 4818/13317 [49:38<1:26:56,  1.63it/s]
 36%|███▌      | 4819/13317 [49:38<1:26:53,  1.63it/s]
                                                      
{'loss': 0.1655, 'grad_norm': 0.4693280756473541, 'learning_rate': 8.052767606058243e-06, 'epoch': 1.09}

 36%|███▌      | 4819/13317 [49:39<1:26:53,  1.63it/s]
 36%|███▌      | 4820/13317 [49:39<1:26:58,  1.63it/s]
                                                      
{'loss': 0.1712, 'grad_norm': 0.4824850261211395, 'learning_rate': 8.051729510855107e-06, 'epoch': 1.09}

 36%|███▌      | 4820/13317 [49:39<1:26:58,  1.63it/s]
 36%|███▌      | 4821/13317 [49:40<1:26:56,  1.63it/s]
                                                      
{'loss': 0.1867, 'grad_norm': 0.48951268196105957, 'learning_rate': 8.05069120596584e-06, 'epoch': 1.09}

 36%|███▌      | 4821/13317 [49:40<1:26:56,  1.63it/s]
 36%|███▌      | 4822/13317 [49:40<1:26:57,  1.63it/s]
                                                      
{'loss': 0.1853, 'grad_norm': 0.47919124364852905, 'learning_rate': 8.049652691461788e-06, 'epoch': 1.09}

 36%|███▌      | 4822/13317 [49:40<1:26:57,  1.63it/s]
 36%|███▌      | 4823/13317 [49:41<1:26:52,  1.63it/s]
                                                      
{'loss': 0.2145, 'grad_norm': 0.62095046043396, 'learning_rate': 8.048613967414305e-06, 'epoch': 1.09}

 36%|███▌      | 4823/13317 [49:41<1:26:52,  1.63it/s]
 36%|███▌      | 4824/13317 [49:42<1:26:52,  1.63it/s]
                                                      
{'loss': 0.186, 'grad_norm': 0.5469053983688354, 'learning_rate': 8.047575033894763e-06, 'epoch': 1.09}

 36%|███▌      | 4824/13317 [49:42<1:26:52,  1.63it/s]
 36%|███▌      | 4825/13317 [49:42<1:26:52,  1.63it/s]
                                                      
{'loss': 0.2162, 'grad_norm': 0.6449478268623352, 'learning_rate': 8.046535890974547e-06, 'epoch': 1.09}

 36%|███▌      | 4825/13317 [49:42<1:26:52,  1.63it/s]
 36%|███▌      | 4826/13317 [49:43<1:26:52,  1.63it/s]
                                                      
{'loss': 0.2049, 'grad_norm': 0.5703588128089905, 'learning_rate': 8.04549653872506e-06, 'epoch': 1.09}

 36%|███▌      | 4826/13317 [49:43<1:26:52,  1.63it/s]
 36%|███▌      | 4827/13317 [49:43<1:26:50,  1.63it/s]
                                                      
{'loss': 0.2204, 'grad_norm': 0.5718799829483032, 'learning_rate': 8.044456977217714e-06, 'epoch': 1.09}

 36%|███▌      | 4827/13317 [49:43<1:26:50,  1.63it/s]
 36%|███▋      | 4828/13317 [49:44<1:26:50,  1.63it/s]
                                                      
{'loss': 0.1981, 'grad_norm': 0.5182607173919678, 'learning_rate': 8.043417206523937e-06, 'epoch': 1.09}

 36%|███▋      | 4828/13317 [49:44<1:26:50,  1.63it/s]
 36%|███▋      | 4829/13317 [49:45<1:26:45,  1.63it/s]
                                                      
{'loss': 0.1941, 'grad_norm': 0.6372060775756836, 'learning_rate': 8.042377226715177e-06, 'epoch': 1.09}

 36%|███▋      | 4829/13317 [49:45<1:26:45,  1.63it/s]
 36%|███▋      | 4830/13317 [49:45<1:26:47,  1.63it/s]
                                                      
{'loss': 0.1909, 'grad_norm': 0.5097138285636902, 'learning_rate': 8.041337037862884e-06, 'epoch': 1.09}

 36%|███▋      | 4830/13317 [49:45<1:26:47,  1.63it/s]
 36%|███▋      | 4831/13317 [49:46<1:26:47,  1.63it/s]
                                                      
{'loss': 0.2477, 'grad_norm': 0.5917192101478577, 'learning_rate': 8.040296640038538e-06, 'epoch': 1.09}

 36%|███▋      | 4831/13317 [49:46<1:26:47,  1.63it/s]
 36%|███▋      | 4832/13317 [49:46<1:26:48,  1.63it/s]
                                                      
{'loss': 0.2708, 'grad_norm': 0.6620301008224487, 'learning_rate': 8.03925603331362e-06, 'epoch': 1.09}

 36%|███▋      | 4832/13317 [49:47<1:26:48,  1.63it/s]
 36%|███▋      | 4833/13317 [49:47<1:26:44,  1.63it/s]
                                                      
{'loss': 0.2021, 'grad_norm': 0.505016565322876, 'learning_rate': 8.038215217759633e-06, 'epoch': 1.09}

 36%|███▋      | 4833/13317 [49:47<1:26:44,  1.63it/s]
 36%|███▋      | 4834/13317 [49:48<1:26:42,  1.63it/s]
                                                      
{'loss': 0.1958, 'grad_norm': 0.5348223447799683, 'learning_rate': 8.03717419344809e-06, 'epoch': 1.09}

 36%|███▋      | 4834/13317 [49:48<1:26:42,  1.63it/s]
 36%|███▋      | 4835/13317 [49:48<1:26:42,  1.63it/s]
                                                      
{'loss': 0.2179, 'grad_norm': 0.5905929803848267, 'learning_rate': 8.036132960450522e-06, 'epoch': 1.09}

 36%|███▋      | 4835/13317 [49:48<1:26:42,  1.63it/s]
 36%|███▋      | 4836/13317 [49:49<1:26:40,  1.63it/s]
                                                      
{'loss': 0.1635, 'grad_norm': 0.48944205045700073, 'learning_rate': 8.035091518838474e-06, 'epoch': 1.09}

 36%|███▋      | 4836/13317 [49:49<1:26:40,  1.63it/s]
 36%|███▋      | 4837/13317 [49:49<1:26:40,  1.63it/s]
                                                      
{'loss': 0.2333, 'grad_norm': 0.5256756544113159, 'learning_rate': 8.034049868683503e-06, 'epoch': 1.09}

 36%|███▋      | 4837/13317 [49:50<1:26:40,  1.63it/s]
 36%|███▋      | 4838/13317 [49:50<1:26:42,  1.63it/s]
                                                      
{'loss': 0.1616, 'grad_norm': 0.42145049571990967, 'learning_rate': 8.033008010057179e-06, 'epoch': 1.09}

 36%|███▋      | 4838/13317 [49:50<1:26:42,  1.63it/s]
 36%|███▋      | 4839/13317 [49:51<1:26:40,  1.63it/s]
                                                      
{'loss': 0.1592, 'grad_norm': 0.43093588948249817, 'learning_rate': 8.031965943031092e-06, 'epoch': 1.09}

 36%|███▋      | 4839/13317 [49:51<1:26:40,  1.63it/s]
 36%|███▋      | 4840/13317 [49:51<1:26:38,  1.63it/s]
                                                      
{'loss': 0.1945, 'grad_norm': 0.5070554614067078, 'learning_rate': 8.03092366767684e-06, 'epoch': 1.09}

 36%|███▋      | 4840/13317 [49:51<1:26:38,  1.63it/s]
 36%|███▋      | 4841/13317 [49:52<1:26:39,  1.63it/s]
                                                      
{'loss': 0.26, 'grad_norm': 0.6066389083862305, 'learning_rate': 8.029881184066042e-06, 'epoch': 1.09}

 36%|███▋      | 4841/13317 [49:52<1:26:39,  1.63it/s]
 36%|███▋      | 4842/13317 [49:53<1:26:40,  1.63it/s]
                                                      
{'loss': 0.1948, 'grad_norm': 0.5052110552787781, 'learning_rate': 8.028838492270325e-06, 'epoch': 1.09}

 36%|███▋      | 4842/13317 [49:53<1:26:40,  1.63it/s]
 36%|███▋      | 4843/13317 [49:53<1:26:40,  1.63it/s]
                                                      
{'loss': 0.1751, 'grad_norm': 0.5022056698799133, 'learning_rate': 8.027795592361334e-06, 'epoch': 1.09}

 36%|███▋      | 4843/13317 [49:53<1:26:40,  1.63it/s]
 36%|███▋      | 4844/13317 [49:54<1:26:41,  1.63it/s]
                                                      
{'loss': 0.2732, 'grad_norm': 0.6178392767906189, 'learning_rate': 8.026752484410727e-06, 'epoch': 1.09}

 36%|███▋      | 4844/13317 [49:54<1:26:41,  1.63it/s]
 36%|███▋      | 4845/13317 [49:54<1:26:37,  1.63it/s]
                                                      
{'loss': 0.2006, 'grad_norm': 0.5131925940513611, 'learning_rate': 8.025709168490177e-06, 'epoch': 1.09}

 36%|███▋      | 4845/13317 [49:55<1:26:37,  1.63it/s]
 36%|███▋      | 4846/13317 [49:55<1:26:37,  1.63it/s]
                                                      
{'loss': 0.1887, 'grad_norm': 0.4793148338794708, 'learning_rate': 8.02466564467137e-06, 'epoch': 1.09}

 36%|███▋      | 4846/13317 [49:55<1:26:37,  1.63it/s]
 36%|███▋      | 4847/13317 [49:56<1:26:37,  1.63it/s]
                                                      
{'loss': 0.2122, 'grad_norm': 0.5700826644897461, 'learning_rate': 8.023621913026006e-06, 'epoch': 1.09}

 36%|███▋      | 4847/13317 [49:56<1:26:37,  1.63it/s]
 36%|███▋      | 4848/13317 [49:56<1:26:38,  1.63it/s]
                                                      
{'loss': 0.1831, 'grad_norm': 0.4821555018424988, 'learning_rate': 8.022577973625803e-06, 'epoch': 1.09}

 36%|███▋      | 4848/13317 [49:56<1:26:38,  1.63it/s]
 36%|███▋      | 4849/13317 [49:57<1:26:34,  1.63it/s]
                                                      
{'loss': 0.1895, 'grad_norm': 0.5003646016120911, 'learning_rate': 8.02153382654249e-06, 'epoch': 1.09}

 36%|███▋      | 4849/13317 [49:57<1:26:34,  1.63it/s]
 36%|███▋      | 4850/13317 [49:57<1:26:36,  1.63it/s]
                                                      
{'loss': 0.2088, 'grad_norm': 0.48592662811279297, 'learning_rate': 8.020489471847812e-06, 'epoch': 1.09}

 36%|███▋      | 4850/13317 [49:58<1:26:36,  1.63it/s]
 36%|███▋      | 4851/13317 [49:58<1:26:35,  1.63it/s]
                                                      
{'loss': 0.2198, 'grad_norm': 0.5355643630027771, 'learning_rate': 8.019444909613524e-06, 'epoch': 1.09}

 36%|███▋      | 4851/13317 [49:58<1:26:35,  1.63it/s]
 36%|███▋      | 4852/13317 [49:59<1:26:32,  1.63it/s]
                                                      
{'loss': 0.1661, 'grad_norm': 0.47960352897644043, 'learning_rate': 8.018400139911399e-06, 'epoch': 1.09}

 36%|███▋      | 4852/13317 [49:59<1:26:32,  1.63it/s]
 36%|███▋      | 4853/13317 [49:59<1:26:31,  1.63it/s]
                                                      
{'loss': 0.19, 'grad_norm': 0.4984211325645447, 'learning_rate': 8.017355162813228e-06, 'epoch': 1.09}

 36%|███▋      | 4853/13317 [49:59<1:26:31,  1.63it/s]
 36%|███▋      | 4854/13317 [50:00<1:26:32,  1.63it/s]
                                                      
{'loss': 0.1909, 'grad_norm': 0.5203539133071899, 'learning_rate': 8.016309978390808e-06, 'epoch': 1.09}

 36%|███▋      | 4854/13317 [50:00<1:26:32,  1.63it/s]
 36%|███▋      | 4855/13317 [50:01<1:26:30,  1.63it/s]
                                                      
{'loss': 0.1614, 'grad_norm': 0.4658195972442627, 'learning_rate': 8.015264586715954e-06, 'epoch': 1.09}

 36%|███▋      | 4855/13317 [50:01<1:26:30,  1.63it/s]
 36%|███▋      | 4856/13317 [50:01<1:26:34,  1.63it/s]
                                                      
{'loss': 0.1752, 'grad_norm': 0.5146451592445374, 'learning_rate': 8.014218987860499e-06, 'epoch': 1.09}

 36%|███▋      | 4856/13317 [50:01<1:26:34,  1.63it/s]
 36%|███▋      | 4857/13317 [50:02<1:26:31,  1.63it/s]
                                                      
{'loss': 0.2735, 'grad_norm': 0.6842280626296997, 'learning_rate': 8.013173181896283e-06, 'epoch': 1.09}

 36%|███▋      | 4857/13317 [50:02<1:26:31,  1.63it/s]
 36%|███▋      | 4858/13317 [50:02<1:26:29,  1.63it/s]
                                                      
{'loss': 0.1768, 'grad_norm': 0.47905200719833374, 'learning_rate': 8.012127168895166e-06, 'epoch': 1.09}

 36%|███▋      | 4858/13317 [50:03<1:26:29,  1.63it/s]
 36%|███▋      | 4859/13317 [50:03<1:26:27,  1.63it/s]
                                                      
{'loss': 0.2087, 'grad_norm': 0.5615405440330505, 'learning_rate': 8.011080948929017e-06, 'epoch': 1.09}

 36%|███▋      | 4859/13317 [50:03<1:26:27,  1.63it/s]
 36%|███▋      | 4860/13317 [50:04<1:26:26,  1.63it/s]
                                                      
{'loss': 0.2578, 'grad_norm': 0.6917605996131897, 'learning_rate': 8.01003452206973e-06, 'epoch': 1.09}

 36%|███▋      | 4860/13317 [50:04<1:26:26,  1.63it/s]
 37%|███▋      | 4861/13317 [50:04<1:26:25,  1.63it/s]
                                                      
{'loss': 0.2279, 'grad_norm': 0.5557433366775513, 'learning_rate': 8.008987888389197e-06, 'epoch': 1.1}

 37%|███▋      | 4861/13317 [50:04<1:26:25,  1.63it/s]
 37%|███▋      | 4862/13317 [50:05<1:26:26,  1.63it/s]
                                                      
{'loss': 0.1599, 'grad_norm': 0.4144836664199829, 'learning_rate': 8.007941047959338e-06, 'epoch': 1.1}

 37%|███▋      | 4862/13317 [50:05<1:26:26,  1.63it/s]
 37%|███▋      | 4863/13317 [50:05<1:26:27,  1.63it/s]
                                                      
{'loss': 0.2053, 'grad_norm': 0.5369203686714172, 'learning_rate': 8.00689400085208e-06, 'epoch': 1.1}

 37%|███▋      | 4863/13317 [50:06<1:26:27,  1.63it/s]
 37%|███▋      | 4864/13317 [50:06<1:26:25,  1.63it/s]
                                                      
{'loss': 0.1758, 'grad_norm': 0.5039169788360596, 'learning_rate': 8.005846747139365e-06, 'epoch': 1.1}

 37%|███▋      | 4864/13317 [50:06<1:26:25,  1.63it/s]
 37%|███▋      | 4865/13317 [50:07<1:26:24,  1.63it/s]
                                                      
{'loss': 0.2093, 'grad_norm': 0.5369598269462585, 'learning_rate': 8.004799286893153e-06, 'epoch': 1.1}

 37%|███▋      | 4865/13317 [50:07<1:26:24,  1.63it/s]
 37%|███▋      | 4866/13317 [50:07<1:26:23,  1.63it/s]
                                                      
{'loss': 0.2506, 'grad_norm': 0.6035506129264832, 'learning_rate': 8.003751620185415e-06, 'epoch': 1.1}

 37%|███▋      | 4866/13317 [50:07<1:26:23,  1.63it/s]
 37%|███▋      | 4867/13317 [50:08<1:26:20,  1.63it/s]
                                                      
{'loss': 0.2139, 'grad_norm': 0.5244039297103882, 'learning_rate': 8.002703747088137e-06, 'epoch': 1.1}

 37%|███▋      | 4867/13317 [50:08<1:26:20,  1.63it/s]
 37%|███▋      | 4868/13317 [50:09<1:26:21,  1.63it/s]
                                                      
{'loss': 0.2946, 'grad_norm': 0.692434549331665, 'learning_rate': 8.001655667673318e-06, 'epoch': 1.1}

 37%|███▋      | 4868/13317 [50:09<1:26:21,  1.63it/s]
 37%|███▋      | 4869/13317 [50:09<1:26:20,  1.63it/s]
                                                      
{'loss': 0.2602, 'grad_norm': 0.5823192000389099, 'learning_rate': 8.000607382012972e-06, 'epoch': 1.1}

 37%|███▋      | 4869/13317 [50:09<1:26:20,  1.63it/s]
 37%|███▋      | 4870/13317 [50:10<1:26:19,  1.63it/s]
                                                      
{'loss': 0.223, 'grad_norm': 0.5739058256149292, 'learning_rate': 7.999558890179129e-06, 'epoch': 1.1}

 37%|███▋      | 4870/13317 [50:10<1:26:19,  1.63it/s]
 37%|███▋      | 4871/13317 [50:10<1:26:19,  1.63it/s]
                                                      
{'loss': 0.209, 'grad_norm': 0.5407153964042664, 'learning_rate': 7.99851019224383e-06, 'epoch': 1.1}

 37%|███▋      | 4871/13317 [50:10<1:26:19,  1.63it/s]
 37%|███▋      | 4872/13317 [50:11<1:26:23,  1.63it/s]
                                                      
{'loss': 0.1948, 'grad_norm': 0.5076795816421509, 'learning_rate': 7.997461288279132e-06, 'epoch': 1.1}

 37%|███▋      | 4872/13317 [50:11<1:26:23,  1.63it/s]
 37%|███▋      | 4873/13317 [50:12<1:26:21,  1.63it/s]
                                                      
{'loss': 0.145, 'grad_norm': 0.39981362223625183, 'learning_rate': 7.996412178357105e-06, 'epoch': 1.1}

 37%|███▋      | 4873/13317 [50:12<1:26:21,  1.63it/s]
 37%|███▋      | 4874/13317 [50:12<1:26:21,  1.63it/s]
                                                      
{'loss': 0.2642, 'grad_norm': 0.5908222794532776, 'learning_rate': 7.995362862549836e-06, 'epoch': 1.1}

 37%|███▋      | 4874/13317 [50:12<1:26:21,  1.63it/s]
 37%|███▋      | 4875/13317 [50:13<1:26:21,  1.63it/s]
                                                      
{'loss': 0.1791, 'grad_norm': 0.45983919501304626, 'learning_rate': 7.994313340929423e-06, 'epoch': 1.1}

 37%|███▋      | 4875/13317 [50:13<1:26:21,  1.63it/s]
 37%|███▋      | 4876/13317 [50:13<1:26:20,  1.63it/s]
                                                      
{'loss': 0.1766, 'grad_norm': 0.4984082579612732, 'learning_rate': 7.993263613567977e-06, 'epoch': 1.1}

 37%|███▋      | 4876/13317 [50:14<1:26:20,  1.63it/s]
 37%|███▋      | 4877/13317 [50:14<1:26:17,  1.63it/s]
                                                      
{'loss': 0.1811, 'grad_norm': 0.48129764199256897, 'learning_rate': 7.99221368053763e-06, 'epoch': 1.1}

 37%|███▋      | 4877/13317 [50:14<1:26:17,  1.63it/s]
 37%|███▋      | 4878/13317 [50:15<1:26:19,  1.63it/s]
                                                      
{'loss': 0.2117, 'grad_norm': 0.5434082746505737, 'learning_rate': 7.99116354191052e-06, 'epoch': 1.1}

 37%|███▋      | 4878/13317 [50:15<1:26:19,  1.63it/s]
 37%|███▋      | 4879/13317 [50:15<1:26:16,  1.63it/s]
                                                      
{'loss': 0.1951, 'grad_norm': 0.5133165717124939, 'learning_rate': 7.990113197758804e-06, 'epoch': 1.1}

 37%|███▋      | 4879/13317 [50:15<1:26:16,  1.63it/s]
 37%|███▋      | 4880/13317 [50:16<1:26:20,  1.63it/s]
                                                      
{'loss': 0.2016, 'grad_norm': 0.5411083102226257, 'learning_rate': 7.989062648154653e-06, 'epoch': 1.1}

 37%|███▋      | 4880/13317 [50:16<1:26:20,  1.63it/s]
 37%|███▋      | 4881/13317 [50:16<1:26:19,  1.63it/s]
                                                      
{'loss': 0.2515, 'grad_norm': 0.5990210771560669, 'learning_rate': 7.988011893170246e-06, 'epoch': 1.1}

 37%|███▋      | 4881/13317 [50:17<1:26:19,  1.63it/s]
 37%|███▋      | 4882/13317 [50:17<1:26:18,  1.63it/s]
                                                      
{'loss': 0.2046, 'grad_norm': 0.5426397323608398, 'learning_rate': 7.986960932877787e-06, 'epoch': 1.1}

 37%|███▋      | 4882/13317 [50:17<1:26:18,  1.63it/s]
 37%|███▋      | 4883/13317 [50:18<1:26:20,  1.63it/s]
                                                      
{'loss': 0.2224, 'grad_norm': 0.6016702055931091, 'learning_rate': 7.985909767349484e-06, 'epoch': 1.1}

 37%|███▋      | 4883/13317 [50:18<1:26:20,  1.63it/s]
 37%|███▋      | 4884/13317 [50:18<1:26:17,  1.63it/s]
                                                      
{'loss': 0.2821, 'grad_norm': 0.6103573441505432, 'learning_rate': 7.984858396657566e-06, 'epoch': 1.1}

 37%|███▋      | 4884/13317 [50:18<1:26:17,  1.63it/s]
 37%|███▋      | 4885/13317 [50:19<1:26:18,  1.63it/s]
                                                      
{'loss': 0.2236, 'grad_norm': 0.5112893581390381, 'learning_rate': 7.983806820874272e-06, 'epoch': 1.1}

 37%|███▋      | 4885/13317 [50:19<1:26:18,  1.63it/s]
 37%|███▋      | 4886/13317 [50:20<1:26:20,  1.63it/s]
                                                      
{'loss': 0.2237, 'grad_norm': 0.5258054137229919, 'learning_rate': 7.982755040071854e-06, 'epoch': 1.1}

 37%|███▋      | 4886/13317 [50:20<1:26:20,  1.63it/s]
 37%|███▋      | 4887/13317 [50:20<1:26:19,  1.63it/s]
                                                      
{'loss': 0.1652, 'grad_norm': 0.4536302983760834, 'learning_rate': 7.981703054322586e-06, 'epoch': 1.1}

 37%|███▋      | 4887/13317 [50:20<1:26:19,  1.63it/s]
 37%|███▋      | 4888/13317 [50:21<1:26:17,  1.63it/s]
                                                      
{'loss': 0.1862, 'grad_norm': 0.44897016882896423, 'learning_rate': 7.980650863698745e-06, 'epoch': 1.1}

 37%|███▋      | 4888/13317 [50:21<1:26:17,  1.63it/s]
 37%|███▋      | 4889/13317 [50:21<1:26:15,  1.63it/s]
                                                      
{'loss': 0.1653, 'grad_norm': 0.4480290710926056, 'learning_rate': 7.979598468272631e-06, 'epoch': 1.1}

 37%|███▋      | 4889/13317 [50:22<1:26:15,  1.63it/s]
 37%|███▋      | 4890/13317 [50:22<1:26:14,  1.63it/s]
                                                      
{'loss': 0.1952, 'grad_norm': 0.4736599922180176, 'learning_rate': 7.978545868116556e-06, 'epoch': 1.1}

 37%|███▋      | 4890/13317 [50:22<1:26:14,  1.63it/s]
 37%|███▋      | 4891/13317 [50:23<1:26:14,  1.63it/s]
                                                      
{'loss': 0.2672, 'grad_norm': 0.6014149188995361, 'learning_rate': 7.97749306330284e-06, 'epoch': 1.1}

 37%|███▋      | 4891/13317 [50:23<1:26:14,  1.63it/s]
 37%|███▋      | 4892/13317 [50:23<1:26:10,  1.63it/s]
                                                      
{'loss': 0.257, 'grad_norm': 0.559394359588623, 'learning_rate': 7.976440053903823e-06, 'epoch': 1.1}

 37%|███▋      | 4892/13317 [50:23<1:26:10,  1.63it/s]
 37%|███▋      | 4893/13317 [50:24<1:26:11,  1.63it/s]
                                                      
{'loss': 0.1977, 'grad_norm': 0.5065642595291138, 'learning_rate': 7.97538683999186e-06, 'epoch': 1.1}

 37%|███▋      | 4893/13317 [50:24<1:26:11,  1.63it/s]
 37%|███▋      | 4894/13317 [50:25<1:30:40,  1.55it/s]
                                                      
{'loss': 0.2039, 'grad_norm': 0.5522657632827759, 'learning_rate': 7.97433342163932e-06, 'epoch': 1.1}

 37%|███▋      | 4894/13317 [50:25<1:30:40,  1.55it/s]
 37%|███▋      | 4895/13317 [50:25<1:29:25,  1.57it/s]
                                                      
{'loss': 0.1607, 'grad_norm': 0.45312607288360596, 'learning_rate': 7.973279798918579e-06, 'epoch': 1.1}

 37%|███▋      | 4895/13317 [50:25<1:29:25,  1.57it/s]
 37%|███▋      | 4896/13317 [50:26<1:28:24,  1.59it/s]
                                                      
{'loss': 0.2167, 'grad_norm': 0.5369633436203003, 'learning_rate': 7.972225971902034e-06, 'epoch': 1.1}

 37%|███▋      | 4896/13317 [50:26<1:28:24,  1.59it/s]
 37%|███▋      | 4897/13317 [50:26<1:27:47,  1.60it/s]
                                                      
{'loss': 0.207, 'grad_norm': 0.5107353925704956, 'learning_rate': 7.971171940662095e-06, 'epoch': 1.1}

 37%|███▋      | 4897/13317 [50:27<1:27:47,  1.60it/s]
 37%|███▋      | 4898/13317 [50:27<1:27:15,  1.61it/s]
                                                      
{'loss': 0.2406, 'grad_norm': 0.6017702221870422, 'learning_rate': 7.970117705271184e-06, 'epoch': 1.1}

 37%|███▋      | 4898/13317 [50:27<1:27:15,  1.61it/s]
 37%|███▋      | 4899/13317 [50:28<1:26:54,  1.61it/s]
                                                      
{'loss': 0.2029, 'grad_norm': 0.5052012801170349, 'learning_rate': 7.969063265801738e-06, 'epoch': 1.1}

 37%|███▋      | 4899/13317 [50:28<1:26:54,  1.61it/s]
 37%|███▋      | 4900/13317 [50:28<1:31:52,  1.53it/s]
                                                      
{'loss': 0.2476, 'grad_norm': 0.5820426940917969, 'learning_rate': 7.96800862232621e-06, 'epoch': 1.1}

 37%|███▋      | 4900/13317 [50:29<1:31:52,  1.53it/s]
 37%|███▋      | 4901/13317 [50:29<1:30:04,  1.56it/s]
                                                      
{'loss': 0.2096, 'grad_norm': 0.5247533917427063, 'learning_rate': 7.966953774917062e-06, 'epoch': 1.1}

 37%|███▋      | 4901/13317 [50:29<1:30:04,  1.56it/s]
 37%|███▋      | 4902/13317 [50:30<1:28:51,  1.58it/s]
                                                      
{'loss': 0.2297, 'grad_norm': 0.5202381610870361, 'learning_rate': 7.965898723646777e-06, 'epoch': 1.1}

 37%|███▋      | 4902/13317 [50:30<1:28:51,  1.58it/s]
 37%|███▋      | 4903/13317 [50:30<1:27:59,  1.59it/s]
                                                      
{'loss': 0.1753, 'grad_norm': 0.44180697202682495, 'learning_rate': 7.964843468587845e-06, 'epoch': 1.1}

 37%|███▋      | 4903/13317 [50:30<1:27:59,  1.59it/s]
 37%|███▋      | 4904/13317 [50:31<1:27:22,  1.60it/s]
                                                      
{'loss': 0.2531, 'grad_norm': 0.6489229798316956, 'learning_rate': 7.963788009812775e-06, 'epoch': 1.1}

 37%|███▋      | 4904/13317 [50:31<1:27:22,  1.60it/s]
 37%|███▋      | 4905/13317 [50:31<1:26:58,  1.61it/s]
                                                      
{'loss': 0.2089, 'grad_norm': 0.6086112856864929, 'learning_rate': 7.962732347394089e-06, 'epoch': 1.1}

 37%|███▋      | 4905/13317 [50:32<1:26:58,  1.61it/s]
 37%|███▋      | 4906/13317 [50:32<1:26:41,  1.62it/s]
                                                      
{'loss': 0.2074, 'grad_norm': 0.5845323204994202, 'learning_rate': 7.96167648140432e-06, 'epoch': 1.11}

 37%|███▋      | 4906/13317 [50:32<1:26:41,  1.62it/s]
 37%|███▋      | 4907/13317 [50:33<1:26:25,  1.62it/s]
                                                      
{'loss': 0.1726, 'grad_norm': 0.5009586811065674, 'learning_rate': 7.960620411916016e-06, 'epoch': 1.11}

 37%|███▋      | 4907/13317 [50:33<1:26:25,  1.62it/s]
 37%|███▋      | 4908/13317 [50:33<1:26:15,  1.62it/s]
                                                      
{'loss': 0.2052, 'grad_norm': 0.6279234886169434, 'learning_rate': 7.959564139001743e-06, 'epoch': 1.11}

 37%|███▋      | 4908/13317 [50:33<1:26:15,  1.62it/s]
 37%|███▋      | 4909/13317 [50:34<1:26:11,  1.63it/s]
                                                      
{'loss': 0.1723, 'grad_norm': 0.45459502935409546, 'learning_rate': 7.958507662734078e-06, 'epoch': 1.11}

 37%|███▋      | 4909/13317 [50:34<1:26:11,  1.63it/s]
 37%|███▋      | 4910/13317 [50:35<1:26:08,  1.63it/s]
                                                      
{'loss': 0.2494, 'grad_norm': 0.5568063259124756, 'learning_rate': 7.95745098318561e-06, 'epoch': 1.11}

 37%|███▋      | 4910/13317 [50:35<1:26:08,  1.63it/s]
 37%|███▋      | 4911/13317 [50:35<1:26:03,  1.63it/s]
                                                      
{'loss': 0.1898, 'grad_norm': 0.5232956409454346, 'learning_rate': 7.956394100428948e-06, 'epoch': 1.11}

 37%|███▋      | 4911/13317 [50:35<1:26:03,  1.63it/s]
 37%|███▋      | 4912/13317 [50:36<1:26:01,  1.63it/s]
                                                      
{'loss': 0.2272, 'grad_norm': 0.4863044321537018, 'learning_rate': 7.955337014536708e-06, 'epoch': 1.11}

 37%|███▋      | 4912/13317 [50:36<1:26:01,  1.63it/s]
 37%|███▋      | 4913/13317 [50:36<1:25:57,  1.63it/s]
                                                      
{'loss': 0.2174, 'grad_norm': 0.48622986674308777, 'learning_rate': 7.954279725581521e-06, 'epoch': 1.11}

 37%|███▋      | 4913/13317 [50:36<1:25:57,  1.63it/s]
 37%|███▋      | 4914/13317 [50:37<1:25:58,  1.63it/s]
                                                      
{'loss': 0.1905, 'grad_norm': 0.503668487071991, 'learning_rate': 7.953222233636038e-06, 'epoch': 1.11}

 37%|███▋      | 4914/13317 [50:37<1:25:58,  1.63it/s]
 37%|███▋      | 4915/13317 [50:38<1:25:55,  1.63it/s]
                                                      
{'loss': 0.209, 'grad_norm': 0.5600008368492126, 'learning_rate': 7.952164538772919e-06, 'epoch': 1.11}

 37%|███▋      | 4915/13317 [50:38<1:25:55,  1.63it/s]
 37%|███▋      | 4916/13317 [50:38<1:25:53,  1.63it/s]
                                                      
{'loss': 0.2563, 'grad_norm': 0.6262803673744202, 'learning_rate': 7.951106641064839e-06, 'epoch': 1.11}

 37%|███▋      | 4916/13317 [50:38<1:25:53,  1.63it/s]
 37%|███▋      | 4917/13317 [50:39<1:25:52,  1.63it/s]
                                                      
{'loss': 0.1488, 'grad_norm': 0.40987464785575867, 'learning_rate': 7.950048540584484e-06, 'epoch': 1.11}

 37%|███▋      | 4917/13317 [50:39<1:25:52,  1.63it/s]
 37%|███▋      | 4918/13317 [50:39<1:25:53,  1.63it/s]
                                                      
{'loss': 0.2584, 'grad_norm': 0.5764193534851074, 'learning_rate': 7.948990237404559e-06, 'epoch': 1.11}

 37%|███▋      | 4918/13317 [50:40<1:25:53,  1.63it/s]
 37%|███▋      | 4919/13317 [50:40<1:25:53,  1.63it/s]
                                                      
{'loss': 0.2289, 'grad_norm': 0.5885605216026306, 'learning_rate': 7.94793173159778e-06, 'epoch': 1.11}

 37%|███▋      | 4919/13317 [50:40<1:25:53,  1.63it/s]
 37%|███▋      | 4920/13317 [50:41<1:25:51,  1.63it/s]
                                                      
{'loss': 0.1692, 'grad_norm': 0.5158630013465881, 'learning_rate': 7.946873023236878e-06, 'epoch': 1.11}

 37%|███▋      | 4920/13317 [50:41<1:25:51,  1.63it/s]
 37%|███▋      | 4921/13317 [50:41<1:26:06,  1.63it/s]
                                                      
{'loss': 0.1514, 'grad_norm': 0.47077566385269165, 'learning_rate': 7.945814112394599e-06, 'epoch': 1.11}

 37%|███▋      | 4921/13317 [50:41<1:26:06,  1.63it/s]
 37%|███▋      | 4922/13317 [50:42<1:25:59,  1.63it/s]
                                                      
{'loss': 0.16, 'grad_norm': 0.4408859610557556, 'learning_rate': 7.944754999143698e-06, 'epoch': 1.11}

 37%|███▋      | 4922/13317 [50:42<1:25:59,  1.63it/s]
 37%|███▋      | 4923/13317 [50:42<1:25:57,  1.63it/s]
                                                      
{'loss': 0.2461, 'grad_norm': 0.5734688639640808, 'learning_rate': 7.943695683556948e-06, 'epoch': 1.11}

 37%|███▋      | 4923/13317 [50:43<1:25:57,  1.63it/s]
 37%|███▋      | 4924/13317 [50:43<1:25:55,  1.63it/s]
                                                      
{'loss': 0.1953, 'grad_norm': 0.5350326895713806, 'learning_rate': 7.942636165707138e-06, 'epoch': 1.11}

 37%|███▋      | 4924/13317 [50:43<1:25:55,  1.63it/s]
 37%|███▋      | 4925/13317 [50:44<1:25:58,  1.63it/s]
                                                      
{'loss': 0.1418, 'grad_norm': 0.38593557476997375, 'learning_rate': 7.941576445667066e-06, 'epoch': 1.11}

 37%|███▋      | 4925/13317 [50:44<1:25:58,  1.63it/s]
 37%|███▋      | 4926/13317 [50:44<1:25:55,  1.63it/s]
                                                      
{'loss': 0.2521, 'grad_norm': 0.5936880707740784, 'learning_rate': 7.940516523509547e-06, 'epoch': 1.11}

 37%|███▋      | 4926/13317 [50:44<1:25:55,  1.63it/s]
 37%|███▋      | 4927/13317 [50:45<1:25:54,  1.63it/s]
                                                      
{'loss': 0.1479, 'grad_norm': 0.40719351172447205, 'learning_rate': 7.939456399307407e-06, 'epoch': 1.11}

 37%|███▋      | 4927/13317 [50:45<1:25:54,  1.63it/s]
 37%|███▋      | 4928/13317 [50:46<1:25:52,  1.63it/s]
                                                      
{'loss': 0.2173, 'grad_norm': 0.5519824028015137, 'learning_rate': 7.938396073133488e-06, 'epoch': 1.11}

 37%|███▋      | 4928/13317 [50:46<1:25:52,  1.63it/s]
 37%|███▋      | 4929/13317 [50:46<1:25:49,  1.63it/s]
                                                      
{'loss': 0.2054, 'grad_norm': 0.5584433674812317, 'learning_rate': 7.937335545060648e-06, 'epoch': 1.11}

 37%|███▋      | 4929/13317 [50:46<1:25:49,  1.63it/s]
 37%|███▋      | 4930/13317 [50:47<1:25:54,  1.63it/s]
                                                      
{'loss': 0.1891, 'grad_norm': 0.4779050052165985, 'learning_rate': 7.936274815161755e-06, 'epoch': 1.11}

 37%|███▋      | 4930/13317 [50:47<1:25:54,  1.63it/s]
 37%|███▋      | 4931/13317 [50:47<1:25:52,  1.63it/s]
                                                      
{'loss': 0.1818, 'grad_norm': 0.4998318552970886, 'learning_rate': 7.935213883509691e-06, 'epoch': 1.11}

 37%|███▋      | 4931/13317 [50:48<1:25:52,  1.63it/s]
 37%|███▋      | 4932/13317 [50:48<1:25:50,  1.63it/s]
                                                      
{'loss': 0.2277, 'grad_norm': 0.49879589676856995, 'learning_rate': 7.934152750177357e-06, 'epoch': 1.11}

 37%|███▋      | 4932/13317 [50:48<1:25:50,  1.63it/s]
 37%|███▋      | 4933/13317 [50:49<1:25:48,  1.63it/s]
                                                      
{'loss': 0.2, 'grad_norm': 0.520417332649231, 'learning_rate': 7.93309141523766e-06, 'epoch': 1.11}

 37%|███▋      | 4933/13317 [50:49<1:25:48,  1.63it/s]
 37%|███▋      | 4934/13317 [50:49<1:25:45,  1.63it/s]
                                                      
{'loss': 0.1766, 'grad_norm': 0.4744150936603546, 'learning_rate': 7.932029878763527e-06, 'epoch': 1.11}

 37%|███▋      | 4934/13317 [50:49<1:25:45,  1.63it/s]
 37%|███▋      | 4935/13317 [50:50<1:25:52,  1.63it/s]
                                                      
{'loss': 0.1763, 'grad_norm': 0.5006439685821533, 'learning_rate': 7.930968140827898e-06, 'epoch': 1.11}

 37%|███▋      | 4935/13317 [50:50<1:25:52,  1.63it/s]
 37%|███▋      | 4936/13317 [50:50<1:25:49,  1.63it/s]
                                                      
{'loss': 0.1501, 'grad_norm': 0.42886391282081604, 'learning_rate': 7.929906201503722e-06, 'epoch': 1.11}

 37%|███▋      | 4936/13317 [50:51<1:25:49,  1.63it/s]
 37%|███▋      | 4937/13317 [50:51<1:25:50,  1.63it/s]
                                                      
{'loss': 0.3078, 'grad_norm': 0.6965404748916626, 'learning_rate': 7.928844060863969e-06, 'epoch': 1.11}

 37%|███▋      | 4937/13317 [50:51<1:25:50,  1.63it/s]
 37%|███▋      | 4938/13317 [50:52<1:25:43,  1.63it/s]
                                                      
{'loss': 0.2347, 'grad_norm': 0.6474852561950684, 'learning_rate': 7.927781718981617e-06, 'epoch': 1.11}

 37%|███▋      | 4938/13317 [50:52<1:25:43,  1.63it/s]
 37%|███▋      | 4939/13317 [50:52<1:25:43,  1.63it/s]
                                                      
{'loss': 0.2125, 'grad_norm': 0.5104877352714539, 'learning_rate': 7.926719175929663e-06, 'epoch': 1.11}

 37%|███▋      | 4939/13317 [50:52<1:25:43,  1.63it/s]
 37%|███▋      | 4940/13317 [50:53<1:25:45,  1.63it/s]
                                                      
{'loss': 0.2391, 'grad_norm': 0.5741419196128845, 'learning_rate': 7.92565643178111e-06, 'epoch': 1.11}

 37%|███▋      | 4940/13317 [50:53<1:25:45,  1.63it/s]
 37%|███▋      | 4941/13317 [50:54<1:25:42,  1.63it/s]
                                                      
{'loss': 0.1804, 'grad_norm': 0.4909672439098358, 'learning_rate': 7.924593486608984e-06, 'epoch': 1.11}

 37%|███▋      | 4941/13317 [50:54<1:25:42,  1.63it/s]
 37%|███▋      | 4942/13317 [50:54<1:25:51,  1.63it/s]
                                                      
{'loss': 0.1789, 'grad_norm': 0.4830353856086731, 'learning_rate': 7.92353034048632e-06, 'epoch': 1.11}

 37%|███▋      | 4942/13317 [50:54<1:25:51,  1.63it/s]
 37%|███▋      | 4943/13317 [50:55<1:25:53,  1.62it/s]
                                                      
{'loss': 0.2774, 'grad_norm': 0.6532754898071289, 'learning_rate': 7.922466993486166e-06, 'epoch': 1.11}

 37%|███▋      | 4943/13317 [50:55<1:25:53,  1.62it/s]
 37%|███▋      | 4944/13317 [50:55<1:25:47,  1.63it/s]
                                                      
{'loss': 0.2311, 'grad_norm': 0.5035069584846497, 'learning_rate': 7.921403445681588e-06, 'epoch': 1.11}

 37%|███▋      | 4944/13317 [50:56<1:25:47,  1.63it/s]
 37%|███▋      | 4945/13317 [50:56<1:25:44,  1.63it/s]
                                                      
{'loss': 0.2365, 'grad_norm': 0.5595019459724426, 'learning_rate': 7.920339697145659e-06, 'epoch': 1.11}

 37%|███▋      | 4945/13317 [50:56<1:25:44,  1.63it/s]
 37%|███▋      | 4946/13317 [50:57<1:25:40,  1.63it/s]
                                                      
{'loss': 0.2057, 'grad_norm': 0.5266584753990173, 'learning_rate': 7.919275747951472e-06, 'epoch': 1.11}

 37%|███▋      | 4946/13317 [50:57<1:25:40,  1.63it/s]
 37%|███▋      | 4947/13317 [50:57<1:26:10,  1.62it/s]
                                                      
{'loss': 0.1932, 'grad_norm': 0.4955552816390991, 'learning_rate': 7.91821159817213e-06, 'epoch': 1.11}

 37%|███▋      | 4947/13317 [50:57<1:26:10,  1.62it/s]
 37%|███▋      | 4948/13317 [50:58<1:26:06,  1.62it/s]
                                                      
{'loss': 0.2131, 'grad_norm': 0.6121083498001099, 'learning_rate': 7.917147247880754e-06, 'epoch': 1.11}

 37%|███▋      | 4948/13317 [50:58<1:26:06,  1.62it/s]
 37%|███▋      | 4949/13317 [50:58<1:25:57,  1.62it/s]
                                                      
{'loss': 0.1452, 'grad_norm': 0.4390018880367279, 'learning_rate': 7.916082697150475e-06, 'epoch': 1.11}

 37%|███▋      | 4949/13317 [50:59<1:25:57,  1.62it/s]
 37%|███▋      | 4950/13317 [50:59<1:25:54,  1.62it/s]
                                                      
{'loss': 0.2075, 'grad_norm': 0.5044956803321838, 'learning_rate': 7.915017946054439e-06, 'epoch': 1.12}

 37%|███▋      | 4950/13317 [50:59<1:25:54,  1.62it/s]
 37%|███▋      | 4951/13317 [51:00<1:25:49,  1.62it/s]
                                                      
{'loss': 0.2233, 'grad_norm': 0.5457475185394287, 'learning_rate': 7.913952994665805e-06, 'epoch': 1.12}

 37%|███▋      | 4951/13317 [51:00<1:25:49,  1.62it/s]
 37%|███▋      | 4952/13317 [51:00<1:25:50,  1.62it/s]
                                                      
{'loss': 0.1917, 'grad_norm': 0.5030430555343628, 'learning_rate': 7.912887843057747e-06, 'epoch': 1.12}

 37%|███▋      | 4952/13317 [51:00<1:25:50,  1.62it/s]
 37%|███▋      | 4953/13317 [51:01<1:25:48,  1.62it/s]
                                                      
{'loss': 0.1918, 'grad_norm': 0.4975908696651459, 'learning_rate': 7.911822491303453e-06, 'epoch': 1.12}

 37%|███▋      | 4953/13317 [51:01<1:25:48,  1.62it/s]
 37%|███▋      | 4954/13317 [51:02<1:25:46,  1.62it/s]
                                                      
{'loss': 0.2534, 'grad_norm': 0.6022861003875732, 'learning_rate': 7.910756939476121e-06, 'epoch': 1.12}

 37%|███▋      | 4954/13317 [51:02<1:25:46,  1.62it/s]
 37%|███▋      | 4955/13317 [51:02<1:25:46,  1.62it/s]
                                                      
{'loss': 0.2042, 'grad_norm': 0.4939626157283783, 'learning_rate': 7.909691187648969e-06, 'epoch': 1.12}

 37%|███▋      | 4955/13317 [51:02<1:25:46,  1.62it/s]
 37%|███▋      | 4956/13317 [51:03<1:25:41,  1.63it/s]
                                                      
{'loss': 0.193, 'grad_norm': 0.5055279731750488, 'learning_rate': 7.908625235895224e-06, 'epoch': 1.12}

 37%|███▋      | 4956/13317 [51:03<1:25:41,  1.63it/s]
 37%|███▋      | 4957/13317 [51:03<1:25:45,  1.62it/s]
                                                      
{'loss': 0.2761, 'grad_norm': 0.7433663010597229, 'learning_rate': 7.907559084288128e-06, 'epoch': 1.12}

 37%|███▋      | 4957/13317 [51:04<1:25:45,  1.62it/s]
 37%|███▋      | 4958/13317 [51:04<1:25:47,  1.62it/s]
                                                      
{'loss': 0.196, 'grad_norm': 0.5313215255737305, 'learning_rate': 7.906492732900937e-06, 'epoch': 1.12}

 37%|███▋      | 4958/13317 [51:04<1:25:47,  1.62it/s]
 37%|███▋      | 4959/13317 [51:05<1:25:43,  1.63it/s]
                                                      
{'loss': 0.2013, 'grad_norm': 0.45424339175224304, 'learning_rate': 7.905426181806921e-06, 'epoch': 1.12}

 37%|███▋      | 4959/13317 [51:05<1:25:43,  1.63it/s]
 37%|███▋      | 4960/13317 [51:05<1:25:47,  1.62it/s]
                                                      
{'loss': 0.2557, 'grad_norm': 0.6216340065002441, 'learning_rate': 7.904359431079363e-06, 'epoch': 1.12}

 37%|███▋      | 4960/13317 [51:05<1:25:47,  1.62it/s]
 37%|███▋      | 4961/13317 [51:06<1:25:41,  1.63it/s]
                                                      
{'loss': 0.1912, 'grad_norm': 0.539266049861908, 'learning_rate': 7.90329248079156e-06, 'epoch': 1.12}

 37%|███▋      | 4961/13317 [51:06<1:25:41,  1.63it/s]
 37%|███▋      | 4962/13317 [51:06<1:25:34,  1.63it/s]
                                                      
{'loss': 0.2071, 'grad_norm': 0.5371692776679993, 'learning_rate': 7.902225331016824e-06, 'epoch': 1.12}

 37%|███▋      | 4962/13317 [51:07<1:25:34,  1.63it/s]
 37%|███▋      | 4963/13317 [51:07<1:25:36,  1.63it/s]
                                                      
{'loss': 0.1709, 'grad_norm': 0.43123406171798706, 'learning_rate': 7.901157981828478e-06, 'epoch': 1.12}

 37%|███▋      | 4963/13317 [51:07<1:25:36,  1.63it/s]
 37%|███▋      | 4964/13317 [51:08<1:25:34,  1.63it/s]
                                                      
{'loss': 0.2022, 'grad_norm': 0.5455217361450195, 'learning_rate': 7.90009043329986e-06, 'epoch': 1.12}

 37%|███▋      | 4964/13317 [51:08<1:25:34,  1.63it/s]
 37%|███▋      | 4965/13317 [51:08<1:25:36,  1.63it/s]
                                                      
{'loss': 0.1955, 'grad_norm': 0.6042534708976746, 'learning_rate': 7.899022685504324e-06, 'epoch': 1.12}

 37%|███▋      | 4965/13317 [51:08<1:25:36,  1.63it/s]
 37%|███▋      | 4966/13317 [51:09<1:25:27,  1.63it/s]
                                                      
{'loss': 0.2223, 'grad_norm': 0.5148909687995911, 'learning_rate': 7.897954738515232e-06, 'epoch': 1.12}

 37%|███▋      | 4966/13317 [51:09<1:25:27,  1.63it/s]
 37%|███▋      | 4967/13317 [51:10<1:25:28,  1.63it/s]
                                                      
{'loss': 0.2321, 'grad_norm': 0.5593107342720032, 'learning_rate': 7.896886592405967e-06, 'epoch': 1.12}

 37%|███▋      | 4967/13317 [51:10<1:25:28,  1.63it/s]
 37%|███▋      | 4968/13317 [51:10<1:25:28,  1.63it/s]
                                                      
{'loss': 0.2391, 'grad_norm': 0.5778020620346069, 'learning_rate': 7.89581824724992e-06, 'epoch': 1.12}

 37%|███▋      | 4968/13317 [51:10<1:25:28,  1.63it/s]
 37%|███▋      | 4969/13317 [51:11<1:25:27,  1.63it/s]
                                                      
{'loss': 0.2354, 'grad_norm': 0.6417771577835083, 'learning_rate': 7.8947497031205e-06, 'epoch': 1.12}

 37%|███▋      | 4969/13317 [51:11<1:25:27,  1.63it/s]
 37%|███▋      | 4970/13317 [51:11<1:25:36,  1.63it/s]
                                                      
{'loss': 0.2405, 'grad_norm': 0.548954963684082, 'learning_rate': 7.893680960091121e-06, 'epoch': 1.12}

 37%|███▋      | 4970/13317 [51:12<1:25:36,  1.63it/s]
 37%|███▋      | 4971/13317 [51:12<1:25:30,  1.63it/s]
                                                      
{'loss': 0.2101, 'grad_norm': 0.5195525288581848, 'learning_rate': 7.892612018235224e-06, 'epoch': 1.12}

 37%|███▋      | 4971/13317 [51:12<1:25:30,  1.63it/s]
 37%|███▋      | 4972/13317 [51:13<1:25:27,  1.63it/s]
                                                      
{'loss': 0.1873, 'grad_norm': 0.48598429560661316, 'learning_rate': 7.891542877626255e-06, 'epoch': 1.12}

 37%|███▋      | 4972/13317 [51:13<1:25:27,  1.63it/s]
 37%|███▋      | 4973/13317 [51:13<1:25:29,  1.63it/s]
                                                      
{'loss': 0.1939, 'grad_norm': 0.5411356687545776, 'learning_rate': 7.890473538337673e-06, 'epoch': 1.12}

 37%|███▋      | 4973/13317 [51:13<1:25:29,  1.63it/s]
 37%|███▋      | 4974/13317 [51:14<1:25:23,  1.63it/s]
                                                      
{'loss': 0.2243, 'grad_norm': 0.5439515113830566, 'learning_rate': 7.889404000442953e-06, 'epoch': 1.12}

 37%|███▋      | 4974/13317 [51:14<1:25:23,  1.63it/s]
 37%|███▋      | 4975/13317 [51:14<1:25:28,  1.63it/s]
                                                      
{'loss': 0.2045, 'grad_norm': 0.5065094232559204, 'learning_rate': 7.888334264015584e-06, 'epoch': 1.12}

 37%|███▋      | 4975/13317 [51:15<1:25:28,  1.63it/s]
 37%|███▋      | 4976/13317 [51:15<1:25:26,  1.63it/s]
                                                      
{'loss': 0.1816, 'grad_norm': 0.5020540952682495, 'learning_rate': 7.887264329129071e-06, 'epoch': 1.12}

 37%|███▋      | 4976/13317 [51:15<1:25:26,  1.63it/s]
 37%|███▋      | 4977/13317 [51:16<1:25:23,  1.63it/s]
                                                      
{'loss': 0.1466, 'grad_norm': 0.43928977847099304, 'learning_rate': 7.886194195856927e-06, 'epoch': 1.12}

 37%|███▋      | 4977/13317 [51:16<1:25:23,  1.63it/s]
 37%|███▋      | 4978/13317 [51:16<1:25:20,  1.63it/s]
                                                      
{'loss': 0.246, 'grad_norm': 0.6548168063163757, 'learning_rate': 7.885123864272683e-06, 'epoch': 1.12}

 37%|███▋      | 4978/13317 [51:16<1:25:20,  1.63it/s]
 37%|███▋      | 4979/13317 [51:17<1:25:17,  1.63it/s]
                                                      
{'loss': 0.2012, 'grad_norm': 0.5428364276885986, 'learning_rate': 7.884053334449881e-06, 'epoch': 1.12}

 37%|███▋      | 4979/13317 [51:17<1:25:17,  1.63it/s]
 37%|███▋      | 4980/13317 [51:18<1:25:20,  1.63it/s]
                                                      
{'loss': 0.2116, 'grad_norm': 0.6479126811027527, 'learning_rate': 7.882982606462078e-06, 'epoch': 1.12}

 37%|███▋      | 4980/13317 [51:18<1:25:20,  1.63it/s]
 37%|███▋      | 4981/13317 [51:18<1:25:16,  1.63it/s]
                                                      
{'loss': 0.1654, 'grad_norm': 0.49162960052490234, 'learning_rate': 7.881911680382844e-06, 'epoch': 1.12}

 37%|███▋      | 4981/13317 [51:18<1:25:16,  1.63it/s]
 37%|███▋      | 4982/13317 [51:19<1:25:15,  1.63it/s]
                                                      
{'loss': 0.1957, 'grad_norm': 0.529255211353302, 'learning_rate': 7.880840556285763e-06, 'epoch': 1.12}

 37%|███▋      | 4982/13317 [51:19<1:25:15,  1.63it/s]
 37%|███▋      | 4983/13317 [51:19<1:25:42,  1.62it/s]
                                                      
{'loss': 0.2579, 'grad_norm': 0.5962173342704773, 'learning_rate': 7.879769234244435e-06, 'epoch': 1.12}

 37%|███▋      | 4983/13317 [51:20<1:25:42,  1.62it/s]
 37%|███▋      | 4984/13317 [51:20<1:25:34,  1.62it/s]
                                                      
{'loss': 0.2088, 'grad_norm': 0.4757121503353119, 'learning_rate': 7.878697714332466e-06, 'epoch': 1.12}

 37%|███▋      | 4984/13317 [51:20<1:25:34,  1.62it/s]
 37%|███▋      | 4985/13317 [51:21<1:25:30,  1.62it/s]
                                                      
{'loss': 0.1882, 'grad_norm': 0.46399667859077454, 'learning_rate': 7.877625996623486e-06, 'epoch': 1.12}

 37%|███▋      | 4985/13317 [51:21<1:25:30,  1.62it/s]
 37%|███▋      | 4986/13317 [51:21<1:25:27,  1.62it/s]
                                                      
{'loss': 0.1831, 'grad_norm': 0.5051536560058594, 'learning_rate': 7.87655408119113e-06, 'epoch': 1.12}

 37%|███▋      | 4986/13317 [51:21<1:25:27,  1.62it/s]
 37%|███▋      | 4987/13317 [51:22<1:25:21,  1.63it/s]
                                                      
{'loss': 0.2136, 'grad_norm': 0.5397459268569946, 'learning_rate': 7.875481968109052e-06, 'epoch': 1.12}

 37%|███▋      | 4987/13317 [51:22<1:25:21,  1.63it/s]
 37%|███▋      | 4988/13317 [51:22<1:25:16,  1.63it/s]
                                                      
{'loss': 0.1714, 'grad_norm': 0.43107837438583374, 'learning_rate': 7.874409657450917e-06, 'epoch': 1.12}

 37%|███▋      | 4988/13317 [51:23<1:25:16,  1.63it/s]
 37%|███▋      | 4989/13317 [51:23<1:25:14,  1.63it/s]
                                                      
{'loss': 0.2453, 'grad_norm': 0.6165768504142761, 'learning_rate': 7.873337149290403e-06, 'epoch': 1.12}

 37%|███▋      | 4989/13317 [51:23<1:25:14,  1.63it/s]
 37%|███▋      | 4990/13317 [51:24<1:25:36,  1.62it/s]
                                                      
{'loss': 0.3252, 'grad_norm': 0.6359604597091675, 'learning_rate': 7.872264443701205e-06, 'epoch': 1.12}

 37%|███▋      | 4990/13317 [51:24<1:25:36,  1.62it/s]
 37%|███▋      | 4991/13317 [51:24<1:25:28,  1.62it/s]
                                                      
{'loss': 0.248, 'grad_norm': 0.5905756950378418, 'learning_rate': 7.871191540757027e-06, 'epoch': 1.12}

 37%|███▋      | 4991/13317 [51:24<1:25:28,  1.62it/s]
 37%|███▋      | 4992/13317 [51:25<1:25:22,  1.63it/s]
                                                      
{'loss': 0.2087, 'grad_norm': 0.5233577489852905, 'learning_rate': 7.870118440531589e-06, 'epoch': 1.12}

 37%|███▋      | 4992/13317 [51:25<1:25:22,  1.63it/s]
 37%|███▋      | 4993/13317 [51:26<1:25:24,  1.62it/s]
                                                      
{'loss': 0.1897, 'grad_norm': 0.502128541469574, 'learning_rate': 7.869045143098624e-06, 'epoch': 1.12}

 37%|███▋      | 4993/13317 [51:26<1:25:24,  1.62it/s]
 38%|███▊      | 4994/13317 [51:26<1:25:17,  1.63it/s]
                                                      
{'loss': 0.1398, 'grad_norm': 0.42175430059432983, 'learning_rate': 7.86797164853188e-06, 'epoch': 1.13}

 38%|███▊      | 4994/13317 [51:26<1:25:17,  1.63it/s]
 38%|███▊      | 4995/13317 [51:27<1:25:41,  1.62it/s]
                                                      
{'loss': 0.2443, 'grad_norm': 0.5680059194564819, 'learning_rate': 7.866897956905119e-06, 'epoch': 1.13}

 38%|███▊      | 4995/13317 [51:27<1:25:41,  1.62it/s]
 38%|███▊      | 4996/13317 [51:27<1:25:28,  1.62it/s]
                                                      
{'loss': 0.2878, 'grad_norm': 0.677178144454956, 'learning_rate': 7.865824068292112e-06, 'epoch': 1.13}

 38%|███▊      | 4996/13317 [51:28<1:25:28,  1.62it/s]
 38%|███▊      | 4997/13317 [51:28<1:25:18,  1.63it/s]
                                                      
{'loss': 0.1567, 'grad_norm': 0.44045278429985046, 'learning_rate': 7.864749982766648e-06, 'epoch': 1.13}

 38%|███▊      | 4997/13317 [51:28<1:25:18,  1.63it/s]
 38%|███▊      | 4998/13317 [51:29<1:25:16,  1.63it/s]
                                                      
{'loss': 0.1501, 'grad_norm': 0.42429324984550476, 'learning_rate': 7.863675700402527e-06, 'epoch': 1.13}

 38%|███▊      | 4998/13317 [51:29<1:25:16,  1.63it/s]
 38%|███▊      | 4999/13317 [51:29<1:25:10,  1.63it/s]
                                                      
{'loss': 0.1429, 'grad_norm': 0.426752507686615, 'learning_rate': 7.862601221273565e-06, 'epoch': 1.13}

 38%|███▊      | 4999/13317 [51:29<1:25:10,  1.63it/s]
 38%|███▊      | 5000/13317 [51:30<1:25:16,  1.63it/s]
                                                      
{'loss': 0.2213, 'grad_norm': 0.4728797376155853, 'learning_rate': 7.861526545453589e-06, 'epoch': 1.13}

 38%|███▊      | 5000/13317 [51:30<1:25:16,  1.63it/s]

  0%|          | 0/13 [00:00<?, ?it/s]

 15%|█▌        | 2/13 [00:00<00:00, 13.37it/s]

 31%|███       | 4/13 [00:00<00:01,  8.38it/s]

 38%|███▊      | 5/13 [00:00<00:01,  7.79it/s]

 46%|████▌     | 6/13 [00:00<00:00,  7.41it/s]

 54%|█████▍    | 7/13 [00:00<00:00,  7.17it/s]

 62%|██████▏   | 8/13 [00:01<00:00,  7.01it/s]

 69%|██████▉   | 9/13 [00:01<00:00,  6.90it/s]

 77%|███████▋  | 10/13 [00:01<00:00,  6.80it/s]

 85%|████████▍ | 11/13 [00:01<00:00,  6.76it/s]

 92%|█████████▏| 12/13 [00:01<00:00,  6.73it/s]

100%|██████████| 13/13 [00:01<00:00,  6.75it/s]
                                                      

                                               
{'eval_loss': 0.6126519441604614, 'eval_runtime': 1.9558, 'eval_samples_per_second': 102.262, 'eval_steps_per_second': 6.647, 'epoch': 1.13}

 38%|███▊      | 5000/13317 [51:32<1:25:16,  1.63it/s]

100%|██████████| 13/13 [00:01<00:00,  6.75it/s]

                                               
 38%|███▊      | 5001/13317 [51:32<2:46:46,  1.20s/it]
                                                      
{'loss': 0.1691, 'grad_norm': 0.44416213035583496, 'learning_rate': 7.86045167301644e-06, 'epoch': 1.13}

 38%|███▊      | 5001/13317 [51:33<2:46:46,  1.20s/it]
 38%|███▊      | 5002/13317 [51:33<2:22:15,  1.03s/it]
                                                      
{'loss': 0.215, 'grad_norm': 0.5072044730186462, 'learning_rate': 7.859376604035975e-06, 'epoch': 1.13}

 38%|███▊      | 5002/13317 [51:33<2:22:15,  1.03s/it]
 38%|███▊      | 5003/13317 [51:34<2:05:25,  1.10it/s]
                                                      
{'loss': 0.196, 'grad_norm': 0.5101170539855957, 'learning_rate': 7.858301338586062e-06, 'epoch': 1.13}

 38%|███▊      | 5003/13317 [51:34<2:05:25,  1.10it/s]
 38%|███▊      | 5004/13317 [51:34<1:53:16,  1.22it/s]
                                                      
{'loss': 0.2072, 'grad_norm': 0.4880150258541107, 'learning_rate': 7.857225876740585e-06, 'epoch': 1.13}

 38%|███▊      | 5004/13317 [51:34<1:53:16,  1.22it/s]
 38%|███▊      | 5005/13317 [51:35<1:44:45,  1.32it/s]
                                                      
{'loss': 0.179, 'grad_norm': 0.47474485635757446, 'learning_rate': 7.856150218573434e-06, 'epoch': 1.13}

 38%|███▊      | 5005/13317 [51:35<1:44:45,  1.32it/s]
 38%|███▊      | 5006/13317 [51:35<1:38:46,  1.40it/s]
                                                      
{'loss': 0.2463, 'grad_norm': 0.5937788486480713, 'learning_rate': 7.855074364158523e-06, 'epoch': 1.13}

 38%|███▊      | 5006/13317 [51:36<1:38:46,  1.40it/s]
 38%|███▊      | 5007/13317 [51:36<1:34:45,  1.46it/s]
                                                      
{'loss': 0.2347, 'grad_norm': 0.5067310929298401, 'learning_rate': 7.853998313569773e-06, 'epoch': 1.13}

 38%|███▊      | 5007/13317 [51:36<1:34:45,  1.46it/s]
 38%|███▊      | 5008/13317 [51:37<1:31:57,  1.51it/s]
                                                      
{'loss': 0.2552, 'grad_norm': 0.46756234765052795, 'learning_rate': 7.852922066881121e-06, 'epoch': 1.13}

 38%|███▊      | 5008/13317 [51:37<1:31:57,  1.51it/s]
 38%|███▊      | 5009/13317 [51:37<1:29:51,  1.54it/s]
                                                      
{'loss': 0.1593, 'grad_norm': 0.40953201055526733, 'learning_rate': 7.851845624166513e-06, 'epoch': 1.13}

 38%|███▊      | 5009/13317 [51:37<1:29:51,  1.54it/s]
 38%|███▊      | 5010/13317 [51:38<1:28:30,  1.56it/s]
                                                      
{'loss': 0.2073, 'grad_norm': 0.5361944437026978, 'learning_rate': 7.850768985499919e-06, 'epoch': 1.13}

 38%|███▊      | 5010/13317 [51:38<1:28:30,  1.56it/s]
 38%|███▊      | 5011/13317 [51:39<1:27:30,  1.58it/s]
                                                      
{'loss': 0.1838, 'grad_norm': 0.495160847902298, 'learning_rate': 7.849692150955308e-06, 'epoch': 1.13}

 38%|███▊      | 5011/13317 [51:39<1:27:30,  1.58it/s]
 38%|███▊      | 5012/13317 [51:39<1:26:45,  1.60it/s]
                                                      
{'loss': 0.2343, 'grad_norm': 0.5696114897727966, 'learning_rate': 7.848615120606676e-06, 'epoch': 1.13}

 38%|███▊      | 5012/13317 [51:39<1:26:45,  1.60it/s]
 38%|███▊      | 5013/13317 [51:40<1:26:19,  1.60it/s]
                                                      
{'loss': 0.2011, 'grad_norm': 0.5092751383781433, 'learning_rate': 7.84753789452802e-06, 'epoch': 1.13}

 38%|███▊      | 5013/13317 [51:40<1:26:19,  1.60it/s]
 38%|███▊      | 5014/13317 [51:40<1:26:00,  1.61it/s]
                                                      
{'loss': 0.1938, 'grad_norm': 0.49583128094673157, 'learning_rate': 7.846460472793363e-06, 'epoch': 1.13}

 38%|███▊      | 5014/13317 [51:41<1:26:00,  1.61it/s]
 38%|███▊      | 5015/13317 [51:41<1:25:42,  1.61it/s]
                                                      
{'loss': 0.1954, 'grad_norm': 0.5159274935722351, 'learning_rate': 7.845382855476731e-06, 'epoch': 1.13}

 38%|███▊      | 5015/13317 [51:41<1:25:42,  1.61it/s]
 38%|███▊      | 5016/13317 [51:42<1:25:24,  1.62it/s]
                                                      
{'loss': 0.1445, 'grad_norm': 0.45044374465942383, 'learning_rate': 7.844305042652172e-06, 'epoch': 1.13}

 38%|███▊      | 5016/13317 [51:42<1:25:24,  1.62it/s]
 38%|███▊      | 5017/13317 [51:42<1:25:20,  1.62it/s]
                                                      
{'loss': 0.214, 'grad_norm': 0.5589297413825989, 'learning_rate': 7.84322703439374e-06, 'epoch': 1.13}

 38%|███▊      | 5017/13317 [51:42<1:25:20,  1.62it/s]
 38%|███▊      | 5018/13317 [51:43<1:25:21,  1.62it/s]
                                                      
{'loss': 0.2005, 'grad_norm': 0.5139644742012024, 'learning_rate': 7.842148830775504e-06, 'epoch': 1.13}

 38%|███▊      | 5018/13317 [51:43<1:25:21,  1.62it/s]
 38%|███▊      | 5019/13317 [51:43<1:25:11,  1.62it/s]
                                                      
{'loss': 0.1964, 'grad_norm': 0.4812833070755005, 'learning_rate': 7.841070431871552e-06, 'epoch': 1.13}

 38%|███▊      | 5019/13317 [51:44<1:25:11,  1.62it/s]
 38%|███▊      | 5020/13317 [51:44<1:25:13,  1.62it/s]
                                                      
{'loss': 0.212, 'grad_norm': 0.5325605273246765, 'learning_rate': 7.839991837755978e-06, 'epoch': 1.13}

 38%|███▊      | 5020/13317 [51:44<1:25:13,  1.62it/s]
 38%|███▊      | 5021/13317 [51:45<1:25:09,  1.62it/s]
                                                      
{'loss': 0.2376, 'grad_norm': 0.5068824887275696, 'learning_rate': 7.838913048502895e-06, 'epoch': 1.13}

 38%|███▊      | 5021/13317 [51:45<1:25:09,  1.62it/s]
 38%|███▊      | 5022/13317 [51:45<1:25:09,  1.62it/s]
                                                      
{'loss': 0.1998, 'grad_norm': 0.46985873579978943, 'learning_rate': 7.837834064186427e-06, 'epoch': 1.13}

 38%|███▊      | 5022/13317 [51:45<1:25:09,  1.62it/s]
 38%|███▊      | 5023/13317 [51:46<1:25:10,  1.62it/s]
                                                      
{'loss': 0.1798, 'grad_norm': 0.5506662130355835, 'learning_rate': 7.836754884880712e-06, 'epoch': 1.13}

 38%|███▊      | 5023/13317 [51:46<1:25:10,  1.62it/s]
 38%|███▊      | 5024/13317 [51:47<1:25:08,  1.62it/s]
                                                      
{'loss': 0.2119, 'grad_norm': 0.541058361530304, 'learning_rate': 7.8356755106599e-06, 'epoch': 1.13}

 38%|███▊      | 5024/13317 [51:47<1:25:08,  1.62it/s]
 38%|███▊      | 5025/13317 [51:47<1:25:15,  1.62it/s]
                                                      
{'loss': 0.2231, 'grad_norm': 0.5398669242858887, 'learning_rate': 7.834595941598156e-06, 'epoch': 1.13}

 38%|███▊      | 5025/13317 [51:47<1:25:15,  1.62it/s]
 38%|███▊      | 5026/13317 [51:48<1:25:08,  1.62it/s]
                                                      
{'loss': 0.1756, 'grad_norm': 0.47218117117881775, 'learning_rate': 7.833516177769657e-06, 'epoch': 1.13}

 38%|███▊      | 5026/13317 [51:48<1:25:08,  1.62it/s]
 38%|███▊      | 5027/13317 [51:48<1:25:09,  1.62it/s]
                                                      
{'loss': 0.1537, 'grad_norm': 0.47089332342147827, 'learning_rate': 7.832436219248596e-06, 'epoch': 1.13}

 38%|███▊      | 5027/13317 [51:49<1:25:09,  1.62it/s]
 38%|███▊      | 5028/13317 [51:49<1:25:10,  1.62it/s]
                                                      
{'loss': 0.2035, 'grad_norm': 0.5029439330101013, 'learning_rate': 7.831356066109176e-06, 'epoch': 1.13}

 38%|███▊      | 5028/13317 [51:49<1:25:10,  1.62it/s]
 38%|███▊      | 5029/13317 [51:50<1:25:06,  1.62it/s]
                                                      
{'loss': 0.1941, 'grad_norm': 0.4866565465927124, 'learning_rate': 7.830275718425612e-06, 'epoch': 1.13}

 38%|███▊      | 5029/13317 [51:50<1:25:06,  1.62it/s]
 38%|███▊      | 5030/13317 [51:50<1:25:08,  1.62it/s]
                                                      
{'loss': 0.2062, 'grad_norm': 0.5710649490356445, 'learning_rate': 7.829195176272141e-06, 'epoch': 1.13}

 38%|███▊      | 5030/13317 [51:50<1:25:08,  1.62it/s]
 38%|███▊      | 5031/13317 [51:51<1:25:00,  1.62it/s]
                                                      
{'loss': 0.273, 'grad_norm': 0.6315731406211853, 'learning_rate': 7.828114439723006e-06, 'epoch': 1.13}

 38%|███▊      | 5031/13317 [51:51<1:25:00,  1.62it/s]
 38%|███▊      | 5032/13317 [51:52<1:24:59,  1.62it/s]
                                                      
{'loss': 0.2047, 'grad_norm': 0.5418707728385925, 'learning_rate': 7.827033508852466e-06, 'epoch': 1.13}

 38%|███▊      | 5032/13317 [51:52<1:24:59,  1.62it/s]
 38%|███▊      | 5033/13317 [51:52<1:24:59,  1.62it/s]
                                                      
{'loss': 0.2299, 'grad_norm': 0.5224904417991638, 'learning_rate': 7.82595238373479e-06, 'epoch': 1.13}

 38%|███▊      | 5033/13317 [51:52<1:24:59,  1.62it/s]
 38%|███▊      | 5034/13317 [51:53<1:24:51,  1.63it/s]
                                                      
{'loss': 0.2055, 'grad_norm': 0.5406088829040527, 'learning_rate': 7.824871064444263e-06, 'epoch': 1.13}

 38%|███▊      | 5034/13317 [51:53<1:24:51,  1.63it/s]
 38%|███▊      | 5035/13317 [51:53<1:24:55,  1.63it/s]
                                                      
{'loss': 0.2614, 'grad_norm': 0.6210120320320129, 'learning_rate': 7.823789551055183e-06, 'epoch': 1.13}

 38%|███▊      | 5035/13317 [51:53<1:24:55,  1.63it/s]
 38%|███▊      | 5036/13317 [51:54<1:24:51,  1.63it/s]
                                                      
{'loss': 0.1992, 'grad_norm': 0.5067441463470459, 'learning_rate': 7.822707843641862e-06, 'epoch': 1.13}

 38%|███▊      | 5036/13317 [51:54<1:24:51,  1.63it/s]
 38%|███▊      | 5037/13317 [51:55<1:24:55,  1.62it/s]
                                                      
{'loss': 0.2087, 'grad_norm': 0.6037783026695251, 'learning_rate': 7.821625942278626e-06, 'epoch': 1.13}

 38%|███▊      | 5037/13317 [51:55<1:24:55,  1.62it/s]
 38%|███▊      | 5038/13317 [51:55<1:24:57,  1.62it/s]
                                                      
{'loss': 0.2274, 'grad_norm': 0.6161059141159058, 'learning_rate': 7.820543847039812e-06, 'epoch': 1.13}

 38%|███▊      | 5038/13317 [51:55<1:24:57,  1.62it/s]
 38%|███▊      | 5039/13317 [51:56<1:24:49,  1.63it/s]
                                                      
{'loss': 0.2125, 'grad_norm': 0.4829203188419342, 'learning_rate': 7.81946155799977e-06, 'epoch': 1.14}

 38%|███▊      | 5039/13317 [51:56<1:24:49,  1.63it/s]
 38%|███▊      | 5040/13317 [51:56<1:24:47,  1.63it/s]
                                                      
{'loss': 0.2065, 'grad_norm': 0.5607085227966309, 'learning_rate': 7.818379075232865e-06, 'epoch': 1.14}

 38%|███▊      | 5040/13317 [51:57<1:24:47,  1.63it/s]
 38%|███▊      | 5041/13317 [51:57<1:24:44,  1.63it/s]
                                                      
{'loss': 0.2276, 'grad_norm': 0.5185084342956543, 'learning_rate': 7.817296398813479e-06, 'epoch': 1.14}

 38%|███▊      | 5041/13317 [51:57<1:24:44,  1.63it/s]
 38%|███▊      | 5042/13317 [51:58<1:24:47,  1.63it/s]
                                                      
{'loss': 0.2131, 'grad_norm': 0.6135324239730835, 'learning_rate': 7.816213528815998e-06, 'epoch': 1.14}

 38%|███▊      | 5042/13317 [51:58<1:24:47,  1.63it/s]
 38%|███▊      | 5043/13317 [51:58<1:25:06,  1.62it/s]
                                                      
{'loss': 0.199, 'grad_norm': 0.47512322664260864, 'learning_rate': 7.81513046531483e-06, 'epoch': 1.14}

 38%|███▊      | 5043/13317 [51:58<1:25:06,  1.62it/s]
 38%|███▊      | 5044/13317 [51:59<1:24:58,  1.62it/s]
                                                      
{'loss': 0.1852, 'grad_norm': 0.5530169606208801, 'learning_rate': 7.814047208384395e-06, 'epoch': 1.14}

 38%|███▊      | 5044/13317 [51:59<1:24:58,  1.62it/s]
 38%|███▊      | 5045/13317 [52:00<1:24:56,  1.62it/s]
                                                      
{'loss': 0.1925, 'grad_norm': 0.5453214645385742, 'learning_rate': 7.812963758099118e-06, 'epoch': 1.14}

 38%|███▊      | 5045/13317 [52:00<1:24:56,  1.62it/s]
 38%|███▊      | 5046/13317 [52:00<1:24:50,  1.62it/s]
                                                      
{'loss': 0.2432, 'grad_norm': 0.6239911913871765, 'learning_rate': 7.811880114533446e-06, 'epoch': 1.14}

 38%|███▊      | 5046/13317 [52:00<1:24:50,  1.62it/s]
 38%|███▊      | 5047/13317 [52:01<1:24:51,  1.62it/s]
                                                      
{'loss': 0.2016, 'grad_norm': 0.5337732434272766, 'learning_rate': 7.810796277761838e-06, 'epoch': 1.14}

 38%|███▊      | 5047/13317 [52:01<1:24:51,  1.62it/s]
 38%|███▊      | 5048/13317 [52:01<1:24:46,  1.63it/s]
                                                      
{'loss': 0.2112, 'grad_norm': 0.5269044041633606, 'learning_rate': 7.809712247858764e-06, 'epoch': 1.14}

 38%|███▊      | 5048/13317 [52:01<1:24:46,  1.63it/s]
 38%|███▊      | 5049/13317 [52:02<1:24:41,  1.63it/s]
                                                      
{'loss': 0.2035, 'grad_norm': 0.49993598461151123, 'learning_rate': 7.80862802489871e-06, 'epoch': 1.14}

 38%|███▊      | 5049/13317 [52:02<1:24:41,  1.63it/s]
 38%|███▊      | 5050/13317 [52:03<1:24:42,  1.63it/s]
                                                      
{'loss': 0.1759, 'grad_norm': 0.5137647986412048, 'learning_rate': 7.807543608956172e-06, 'epoch': 1.14}

 38%|███▊      | 5050/13317 [52:03<1:24:42,  1.63it/s]
 38%|███▊      | 5051/13317 [52:03<1:24:42,  1.63it/s]
                                                      
{'loss': 0.2781, 'grad_norm': 0.6710174679756165, 'learning_rate': 7.806459000105662e-06, 'epoch': 1.14}

 38%|███▊      | 5051/13317 [52:03<1:24:42,  1.63it/s]
 38%|███▊      | 5052/13317 [52:04<1:24:41,  1.63it/s]
                                                      
{'loss': 0.1692, 'grad_norm': 0.46386563777923584, 'learning_rate': 7.805374198421702e-06, 'epoch': 1.14}

 38%|███▊      | 5052/13317 [52:04<1:24:41,  1.63it/s]
 38%|███▊      | 5053/13317 [52:04<1:24:37,  1.63it/s]
                                                      
{'loss': 0.176, 'grad_norm': 0.48329290747642517, 'learning_rate': 7.804289203978833e-06, 'epoch': 1.14}

 38%|███▊      | 5053/13317 [52:05<1:24:37,  1.63it/s]
 38%|███▊      | 5054/13317 [52:05<1:24:35,  1.63it/s]
                                                      
{'loss': 0.1924, 'grad_norm': 0.4599270820617676, 'learning_rate': 7.803204016851603e-06, 'epoch': 1.14}

 38%|███▊      | 5054/13317 [52:05<1:24:35,  1.63it/s]
 38%|███▊      | 5055/13317 [52:06<1:24:37,  1.63it/s]
                                                      
{'loss': 0.2363, 'grad_norm': 0.554645299911499, 'learning_rate': 7.802118637114575e-06, 'epoch': 1.14}

 38%|███▊      | 5055/13317 [52:06<1:24:37,  1.63it/s]
 38%|███▊      | 5056/13317 [52:06<1:24:33,  1.63it/s]
                                                      
{'loss': 0.2158, 'grad_norm': 0.48455625772476196, 'learning_rate': 7.801033064842328e-06, 'epoch': 1.14}

 38%|███▊      | 5056/13317 [52:06<1:24:33,  1.63it/s]
 38%|███▊      | 5057/13317 [52:07<1:24:32,  1.63it/s]
                                                      
{'loss': 0.1941, 'grad_norm': 0.5006222128868103, 'learning_rate': 7.79994730010945e-06, 'epoch': 1.14}

 38%|███▊      | 5057/13317 [52:07<1:24:32,  1.63it/s]
 38%|███▊      | 5058/13317 [52:08<1:24:58,  1.62it/s]
                                                      
{'loss': 0.1693, 'grad_norm': 0.47029295563697815, 'learning_rate': 7.798861342990548e-06, 'epoch': 1.14}

 38%|███▊      | 5058/13317 [52:08<1:24:58,  1.62it/s]
 38%|███▊      | 5059/13317 [52:08<1:24:49,  1.62it/s]
                                                      
{'loss': 0.1893, 'grad_norm': 0.5381361246109009, 'learning_rate': 7.797775193560237e-06, 'epoch': 1.14}

 38%|███▊      | 5059/13317 [52:08<1:24:49,  1.62it/s]
 38%|███▊      | 5060/13317 [52:09<1:24:50,  1.62it/s]
                                                      
{'loss': 0.2271, 'grad_norm': 0.5855047106742859, 'learning_rate': 7.796688851893146e-06, 'epoch': 1.14}

 38%|███▊      | 5060/13317 [52:09<1:24:50,  1.62it/s]
 38%|███▊      | 5061/13317 [52:09<1:24:43,  1.62it/s]
                                                      
{'loss': 0.1617, 'grad_norm': 0.4039457440376282, 'learning_rate': 7.79560231806392e-06, 'epoch': 1.14}

 38%|███▊      | 5061/13317 [52:09<1:24:43,  1.62it/s]
 38%|███▊      | 5062/13317 [52:10<1:24:35,  1.63it/s]
                                                      
{'loss': 0.2034, 'grad_norm': 0.514007568359375, 'learning_rate': 7.794515592147212e-06, 'epoch': 1.14}

 38%|███▊      | 5062/13317 [52:10<1:24:35,  1.63it/s]
 38%|███▊      | 5063/13317 [52:11<1:24:36,  1.63it/s]
                                                      
{'loss': 0.239, 'grad_norm': 0.5271636843681335, 'learning_rate': 7.793428674217696e-06, 'epoch': 1.14}

 38%|███▊      | 5063/13317 [52:11<1:24:36,  1.63it/s]
 38%|███▊      | 5064/13317 [52:11<1:24:33,  1.63it/s]
                                                      
{'loss': 0.1996, 'grad_norm': 0.589052140712738, 'learning_rate': 7.792341564350054e-06, 'epoch': 1.14}

 38%|███▊      | 5064/13317 [52:11<1:24:33,  1.63it/s]
 38%|███▊      | 5065/13317 [52:12<1:24:58,  1.62it/s]
                                                      
{'loss': 0.1593, 'grad_norm': 0.4269770085811615, 'learning_rate': 7.79125426261898e-06, 'epoch': 1.14}

 38%|███▊      | 5065/13317 [52:12<1:24:58,  1.62it/s]
 38%|███▊      | 5066/13317 [52:12<1:24:49,  1.62it/s]
                                                      
{'loss': 0.2408, 'grad_norm': 0.548892080783844, 'learning_rate': 7.790166769099183e-06, 'epoch': 1.14}

 38%|███▊      | 5066/13317 [52:13<1:24:49,  1.62it/s]
 38%|███▊      | 5067/13317 [52:13<1:24:44,  1.62it/s]
                                                      
{'loss': 0.2013, 'grad_norm': 0.44884830713272095, 'learning_rate': 7.789079083865385e-06, 'epoch': 1.14}

 38%|███▊      | 5067/13317 [52:13<1:24:44,  1.62it/s]
 38%|███▊      | 5068/13317 [52:14<1:24:40,  1.62it/s]
                                                      
{'loss': 0.2877, 'grad_norm': 0.5771863460540771, 'learning_rate': 7.787991206992324e-06, 'epoch': 1.14}

 38%|███▊      | 5068/13317 [52:14<1:24:40,  1.62it/s]
 38%|███▊      | 5069/13317 [52:14<1:24:31,  1.63it/s]
                                                      
{'loss': 0.2241, 'grad_norm': 0.5631236433982849, 'learning_rate': 7.786903138554747e-06, 'epoch': 1.14}

 38%|███▊      | 5069/13317 [52:14<1:24:31,  1.63it/s]
 38%|███▊      | 5070/13317 [52:15<1:24:33,  1.63it/s]
                                                      
{'loss': 0.199, 'grad_norm': 0.5127972364425659, 'learning_rate': 7.785814878627418e-06, 'epoch': 1.14}

 38%|███▊      | 5070/13317 [52:15<1:24:33,  1.63it/s]
 38%|███▊      | 5071/13317 [52:16<1:24:29,  1.63it/s]
                                                      
{'loss': 0.2271, 'grad_norm': 0.6094288229942322, 'learning_rate': 7.78472642728511e-06, 'epoch': 1.14}

 38%|███▊      | 5071/13317 [52:16<1:24:29,  1.63it/s]
 38%|███▊      | 5072/13317 [52:16<1:24:25,  1.63it/s]
                                                      
{'loss': 0.2122, 'grad_norm': 0.5793483853340149, 'learning_rate': 7.783637784602608e-06, 'epoch': 1.14}

 38%|███▊      | 5072/13317 [52:16<1:24:25,  1.63it/s]
 38%|███▊      | 5073/13317 [52:17<1:24:28,  1.63it/s]
                                                      
{'loss': 0.2065, 'grad_norm': 0.540442168712616, 'learning_rate': 7.78254895065472e-06, 'epoch': 1.14}

 38%|███▊      | 5073/13317 [52:17<1:24:28,  1.63it/s]
 38%|███▊      | 5074/13317 [52:17<1:24:46,  1.62it/s]
                                                      
{'loss': 0.1899, 'grad_norm': 0.48222580552101135, 'learning_rate': 7.781459925516255e-06, 'epoch': 1.14}

 38%|███▊      | 5074/13317 [52:17<1:24:46,  1.62it/s]
 38%|███▊      | 5075/13317 [52:18<1:24:41,  1.62it/s]
                                                      
{'loss': 0.2182, 'grad_norm': 0.4781748652458191, 'learning_rate': 7.780370709262044e-06, 'epoch': 1.14}

 38%|███▊      | 5075/13317 [52:18<1:24:41,  1.62it/s]
 38%|███▊      | 5076/13317 [52:19<1:24:37,  1.62it/s]
                                                      
{'loss': 0.2347, 'grad_norm': 0.5956448316574097, 'learning_rate': 7.779281301966924e-06, 'epoch': 1.14}

 38%|███▊      | 5076/13317 [52:19<1:24:37,  1.62it/s]
 38%|███▊      | 5077/13317 [52:19<1:24:31,  1.62it/s]
                                                      
{'loss': 0.183, 'grad_norm': 0.47589626908302307, 'learning_rate': 7.778191703705753e-06, 'epoch': 1.14}

 38%|███▊      | 5077/13317 [52:19<1:24:31,  1.62it/s]
 38%|███▊      | 5078/13317 [52:20<1:24:32,  1.62it/s]
                                                      
{'loss': 0.223, 'grad_norm': 0.5694820284843445, 'learning_rate': 7.777101914553396e-06, 'epoch': 1.14}

 38%|███▊      | 5078/13317 [52:20<1:24:32,  1.62it/s]
 38%|███▊      | 5079/13317 [52:20<1:24:25,  1.63it/s]
                                                      
{'loss': 0.2313, 'grad_norm': 0.5968425869941711, 'learning_rate': 7.776011934584732e-06, 'epoch': 1.14}

 38%|███▊      | 5079/13317 [52:21<1:24:25,  1.63it/s]
 38%|███▊      | 5080/13317 [52:21<1:24:30,  1.62it/s]
                                                      
{'loss': 0.1563, 'grad_norm': 0.45284080505371094, 'learning_rate': 7.774921763874657e-06, 'epoch': 1.14}

 38%|███▊      | 5080/13317 [52:21<1:24:30,  1.62it/s]
 38%|███▊      | 5081/13317 [52:22<1:24:25,  1.63it/s]
                                                      
{'loss': 0.178, 'grad_norm': 0.5567383766174316, 'learning_rate': 7.773831402498073e-06, 'epoch': 1.14}

 38%|███▊      | 5081/13317 [52:22<1:24:25,  1.63it/s]
 38%|███▊      | 5082/13317 [52:22<1:24:20,  1.63it/s]
                                                      
{'loss': 0.2599, 'grad_norm': 0.6665673851966858, 'learning_rate': 7.772740850529903e-06, 'epoch': 1.14}

 38%|███▊      | 5082/13317 [52:22<1:24:20,  1.63it/s]
 38%|███▊      | 5083/13317 [52:23<1:24:22,  1.63it/s]
                                                      
{'loss': 0.1921, 'grad_norm': 0.4792543053627014, 'learning_rate': 7.771650108045079e-06, 'epoch': 1.15}

 38%|███▊      | 5083/13317 [52:23<1:24:22,  1.63it/s]
 38%|███▊      | 5084/13317 [52:24<1:24:18,  1.63it/s]
                                                      
{'loss': 0.1816, 'grad_norm': 0.5263941884040833, 'learning_rate': 7.770559175118545e-06, 'epoch': 1.15}

 38%|███▊      | 5084/13317 [52:24<1:24:18,  1.63it/s]
 38%|███▊      | 5085/13317 [52:24<1:24:20,  1.63it/s]
                                                      
{'loss': 0.1953, 'grad_norm': 0.5223544836044312, 'learning_rate': 7.769468051825262e-06, 'epoch': 1.15}

 38%|███▊      | 5085/13317 [52:24<1:24:20,  1.63it/s]
 38%|███▊      | 5086/13317 [52:25<1:24:17,  1.63it/s]
                                                      
{'loss': 0.1854, 'grad_norm': 0.5183048248291016, 'learning_rate': 7.768376738240198e-06, 'epoch': 1.15}

 38%|███▊      | 5086/13317 [52:25<1:24:17,  1.63it/s]
 38%|███▊      | 5087/13317 [52:25<1:24:20,  1.63it/s]
                                                      
{'loss': 0.197, 'grad_norm': 0.5096584558486938, 'learning_rate': 7.76728523443834e-06, 'epoch': 1.15}

 38%|███▊      | 5087/13317 [52:25<1:24:20,  1.63it/s]
 38%|███▊      | 5088/13317 [52:26<1:24:19,  1.63it/s]
                                                      
{'loss': 0.178, 'grad_norm': 0.4807375967502594, 'learning_rate': 7.76619354049469e-06, 'epoch': 1.15}

 38%|███▊      | 5088/13317 [52:26<1:24:19,  1.63it/s]
 38%|███▊      | 5089/13317 [52:27<1:24:15,  1.63it/s]
                                                      
{'loss': 0.1401, 'grad_norm': 0.3944408893585205, 'learning_rate': 7.76510165648425e-06, 'epoch': 1.15}

 38%|███▊      | 5089/13317 [52:27<1:24:15,  1.63it/s]
 38%|███▊      | 5090/13317 [52:27<1:24:19,  1.63it/s]
                                                      
{'loss': 0.3054, 'grad_norm': 0.6317303776741028, 'learning_rate': 7.76400958248205e-06, 'epoch': 1.15}

 38%|███▊      | 5090/13317 [52:27<1:24:19,  1.63it/s]
 38%|███▊      | 5091/13317 [52:28<1:24:12,  1.63it/s]
                                                      
{'loss': 0.1854, 'grad_norm': 0.47378939390182495, 'learning_rate': 7.762917318563126e-06, 'epoch': 1.15}

 38%|███▊      | 5091/13317 [52:28<1:24:12,  1.63it/s]
 38%|███▊      | 5092/13317 [52:28<1:24:16,  1.63it/s]
                                                      
{'loss': 0.1601, 'grad_norm': 0.45456916093826294, 'learning_rate': 7.76182486480253e-06, 'epoch': 1.15}

 38%|███▊      | 5092/13317 [52:29<1:24:16,  1.63it/s]
 38%|███▊      | 5093/13317 [52:29<1:24:15,  1.63it/s]
                                                      
{'loss': 0.2443, 'grad_norm': 0.5610646605491638, 'learning_rate': 7.760732221275322e-06, 'epoch': 1.15}

 38%|███▊      | 5093/13317 [52:29<1:24:15,  1.63it/s]
 38%|███▊      | 5094/13317 [52:30<1:24:14,  1.63it/s]
                                                      
{'loss': 0.21, 'grad_norm': 0.5319033265113831, 'learning_rate': 7.759639388056579e-06, 'epoch': 1.15}

 38%|███▊      | 5094/13317 [52:30<1:24:14,  1.63it/s]
 38%|███▊      | 5095/13317 [52:30<1:24:20,  1.62it/s]
                                                      
{'loss': 0.2403, 'grad_norm': 0.5403506755828857, 'learning_rate': 7.75854636522139e-06, 'epoch': 1.15}

 38%|███▊      | 5095/13317 [52:30<1:24:20,  1.62it/s]
 38%|███▊      | 5096/13317 [52:31<1:24:18,  1.63it/s]
                                                      
{'loss': 0.1948, 'grad_norm': 0.5041813850402832, 'learning_rate': 7.757453152844859e-06, 'epoch': 1.15}

 38%|███▊      | 5096/13317 [52:31<1:24:18,  1.63it/s]
 38%|███▊      | 5097/13317 [52:31<1:24:19,  1.62it/s]
                                                      
{'loss': 0.2017, 'grad_norm': 0.5512658953666687, 'learning_rate': 7.756359751002098e-06, 'epoch': 1.15}

 38%|███▊      | 5097/13317 [52:32<1:24:19,  1.62it/s]
 38%|███▊      | 5098/13317 [52:32<1:24:22,  1.62it/s]
                                                      
{'loss': 0.262, 'grad_norm': 0.6115124821662903, 'learning_rate': 7.75526615976824e-06, 'epoch': 1.15}

 38%|███▊      | 5098/13317 [52:32<1:24:22,  1.62it/s]
 38%|███▊      | 5099/13317 [52:33<1:24:13,  1.63it/s]
                                                      
{'loss': 0.2359, 'grad_norm': 0.6152074933052063, 'learning_rate': 7.75417237921842e-06, 'epoch': 1.15}

 38%|███▊      | 5099/13317 [52:33<1:24:13,  1.63it/s]
 38%|███▊      | 5100/13317 [52:33<1:24:17,  1.62it/s]
                                                      
{'loss': 0.2358, 'grad_norm': 0.5692088603973389, 'learning_rate': 7.753078409427799e-06, 'epoch': 1.15}

 38%|███▊      | 5100/13317 [52:33<1:24:17,  1.62it/s]
 38%|███▊      | 5101/13317 [52:34<1:24:10,  1.63it/s]
                                                      
{'loss': 0.1847, 'grad_norm': 0.506322979927063, 'learning_rate': 7.75198425047154e-06, 'epoch': 1.15}

 38%|███▊      | 5101/13317 [52:34<1:24:10,  1.63it/s]
 38%|███▊      | 5102/13317 [52:35<1:24:15,  1.62it/s]
                                                      
{'loss': 0.1983, 'grad_norm': 0.5069110989570618, 'learning_rate': 7.750889902424825e-06, 'epoch': 1.15}

 38%|███▊      | 5102/13317 [52:35<1:24:15,  1.62it/s]
 38%|███▊      | 5103/13317 [52:35<1:24:16,  1.62it/s]
                                                      
{'loss': 0.1932, 'grad_norm': 0.4933864176273346, 'learning_rate': 7.749795365362847e-06, 'epoch': 1.15}

 38%|███▊      | 5103/13317 [52:35<1:24:16,  1.62it/s]
 38%|███▊      | 5104/13317 [52:36<1:24:13,  1.63it/s]
                                                      
{'loss': 0.237, 'grad_norm': 0.5738479495048523, 'learning_rate': 7.748700639360812e-06, 'epoch': 1.15}

 38%|███▊      | 5104/13317 [52:36<1:24:13,  1.63it/s]
 38%|███▊      | 5105/13317 [52:36<1:24:14,  1.62it/s]
                                                      
{'loss': 0.2064, 'grad_norm': 0.5306676030158997, 'learning_rate': 7.747605724493938e-06, 'epoch': 1.15}

 38%|███▊      | 5105/13317 [52:37<1:24:14,  1.62it/s]
 38%|███▊      | 5106/13317 [52:37<1:24:11,  1.63it/s]
                                                      
{'loss': 0.1774, 'grad_norm': 0.49842938780784607, 'learning_rate': 7.74651062083746e-06, 'epoch': 1.15}

 38%|███▊      | 5106/13317 [52:37<1:24:11,  1.63it/s]
 38%|███▊      | 5107/13317 [52:38<1:24:13,  1.62it/s]
                                                      
{'loss': 0.1845, 'grad_norm': 0.5520543456077576, 'learning_rate': 7.74541532846662e-06, 'epoch': 1.15}

 38%|███▊      | 5107/13317 [52:38<1:24:13,  1.62it/s]
 38%|███▊      | 5108/13317 [52:38<1:24:08,  1.63it/s]
                                                      
{'loss': 0.1842, 'grad_norm': 0.4750032424926758, 'learning_rate': 7.74431984745668e-06, 'epoch': 1.15}

 38%|███▊      | 5108/13317 [52:38<1:24:08,  1.63it/s]
 38%|███▊      | 5109/13317 [52:39<1:24:03,  1.63it/s]
                                                      
{'loss': 0.246, 'grad_norm': 0.5837498903274536, 'learning_rate': 7.74322417788291e-06, 'epoch': 1.15}

 38%|███▊      | 5109/13317 [52:39<1:24:03,  1.63it/s]
 38%|███▊      | 5110/13317 [52:39<1:24:04,  1.63it/s]
                                                      
{'loss': 0.173, 'grad_norm': 0.4543144404888153, 'learning_rate': 7.742128319820593e-06, 'epoch': 1.15}

 38%|███▊      | 5110/13317 [52:40<1:24:04,  1.63it/s]
 38%|███▊      | 5111/13317 [52:40<1:24:01,  1.63it/s]
                                                      
{'loss': 0.195, 'grad_norm': 0.4894198775291443, 'learning_rate': 7.741032273345023e-06, 'epoch': 1.15}

 38%|███▊      | 5111/13317 [52:40<1:24:01,  1.63it/s]
 38%|███▊      | 5112/13317 [52:41<1:23:55,  1.63it/s]
                                                      
{'loss': 0.2019, 'grad_norm': 0.49744364619255066, 'learning_rate': 7.739936038531517e-06, 'epoch': 1.15}

 38%|███▊      | 5112/13317 [52:41<1:23:55,  1.63it/s]
 38%|███▊      | 5113/13317 [52:41<1:24:00,  1.63it/s]
                                                      
{'loss': 0.1683, 'grad_norm': 0.43812066316604614, 'learning_rate': 7.738839615455391e-06, 'epoch': 1.15}

 38%|███▊      | 5113/13317 [52:41<1:24:00,  1.63it/s]
 38%|███▊      | 5114/13317 [52:42<1:23:56,  1.63it/s]
                                                      
{'loss': 0.2368, 'grad_norm': 0.6080548167228699, 'learning_rate': 7.737743004191986e-06, 'epoch': 1.15}

 38%|███▊      | 5114/13317 [52:42<1:23:56,  1.63it/s]
 38%|███▊      | 5115/13317 [52:43<1:24:00,  1.63it/s]
                                                      
{'loss': 0.1648, 'grad_norm': 0.4598134458065033, 'learning_rate': 7.736646204816648e-06, 'epoch': 1.15}

 38%|███▊      | 5115/13317 [52:43<1:24:00,  1.63it/s]
 38%|███▊      | 5116/13317 [52:43<1:23:56,  1.63it/s]
                                                      
{'loss': 0.1743, 'grad_norm': 0.46998125314712524, 'learning_rate': 7.735549217404741e-06, 'epoch': 1.15}

 38%|███▊      | 5116/13317 [52:43<1:23:56,  1.63it/s]
 38%|███▊      | 5117/13317 [52:44<1:23:53,  1.63it/s]
                                                      
{'loss': 0.2029, 'grad_norm': 0.5111919045448303, 'learning_rate': 7.734452042031639e-06, 'epoch': 1.15}

 38%|███▊      | 5117/13317 [52:44<1:23:53,  1.63it/s]
 38%|███▊      | 5118/13317 [52:44<1:23:56,  1.63it/s]
                                                      
{'loss': 0.1952, 'grad_norm': 0.5212923884391785, 'learning_rate': 7.733354678772726e-06, 'epoch': 1.15}

 38%|███▊      | 5118/13317 [52:45<1:23:56,  1.63it/s]
 38%|███▊      | 5119/13317 [52:45<1:23:53,  1.63it/s]
                                                      
{'loss': 0.1681, 'grad_norm': 0.47161492705345154, 'learning_rate': 7.732257127703408e-06, 'epoch': 1.15}

 38%|███▊      | 5119/13317 [52:45<1:23:53,  1.63it/s]
 38%|███▊      | 5120/13317 [52:46<1:23:56,  1.63it/s]
                                                      
{'loss': 0.1559, 'grad_norm': 0.45795199275016785, 'learning_rate': 7.731159388899094e-06, 'epoch': 1.15}

 38%|███▊      | 5120/13317 [52:46<1:23:56,  1.63it/s]
 38%|███▊      | 5121/13317 [52:46<1:23:57,  1.63it/s]
                                                      
{'loss': 0.2268, 'grad_norm': 0.5331576466560364, 'learning_rate': 7.730061462435214e-06, 'epoch': 1.15}

 38%|███▊      | 5121/13317 [52:46<1:23:57,  1.63it/s]
 38%|███▊      | 5122/13317 [52:47<1:23:56,  1.63it/s]
                                                      
{'loss': 0.212, 'grad_norm': 0.49380385875701904, 'learning_rate': 7.728963348387205e-06, 'epoch': 1.15}

 38%|███▊      | 5122/13317 [52:47<1:23:56,  1.63it/s]
 38%|███▊      | 5123/13317 [52:47<1:23:58,  1.63it/s]
                                                      
{'loss': 0.2197, 'grad_norm': 0.5589454174041748, 'learning_rate': 7.727865046830517e-06, 'epoch': 1.15}

 38%|███▊      | 5123/13317 [52:48<1:23:58,  1.63it/s]
 38%|███▊      | 5124/13317 [52:48<1:23:55,  1.63it/s]
                                                      
{'loss': 0.1964, 'grad_norm': 0.5275042057037354, 'learning_rate': 7.72676655784062e-06, 'epoch': 1.15}

 38%|███▊      | 5124/13317 [52:48<1:23:55,  1.63it/s]
 38%|███▊      | 5125/13317 [52:49<1:23:57,  1.63it/s]
                                                      
{'loss': 0.2289, 'grad_norm': 0.627875030040741, 'learning_rate': 7.725667881492984e-06, 'epoch': 1.15}

 38%|███▊      | 5125/13317 [52:49<1:23:57,  1.63it/s]
 38%|███▊      | 5126/13317 [52:49<1:23:54,  1.63it/s]
                                                      
{'loss': 0.2094, 'grad_norm': 0.5948286056518555, 'learning_rate': 7.72456901786311e-06, 'epoch': 1.15}

 38%|███▊      | 5126/13317 [52:49<1:23:54,  1.63it/s]
 38%|███▊      | 5127/13317 [52:50<1:23:51,  1.63it/s]
                                                      
{'loss': 0.2213, 'grad_norm': 0.5888522863388062, 'learning_rate': 7.723469967026494e-06, 'epoch': 1.15}

 38%|███▊      | 5127/13317 [52:50<1:23:51,  1.63it/s]
 39%|███▊      | 5128/13317 [52:51<1:26:05,  1.59it/s]
                                                      
{'loss': 0.1776, 'grad_norm': 0.5229619145393372, 'learning_rate': 7.722370729058654e-06, 'epoch': 1.16}

 39%|███▊      | 5128/13317 [52:51<1:26:05,  1.59it/s]
 39%|███▊      | 5129/13317 [52:51<1:25:20,  1.60it/s]
                                                      
{'loss': 0.1958, 'grad_norm': 0.5136141777038574, 'learning_rate': 7.721271304035123e-06, 'epoch': 1.16}

 39%|███▊      | 5129/13317 [52:51<1:25:20,  1.60it/s]
 39%|███▊      | 5130/13317 [52:52<1:25:24,  1.60it/s]
                                                      
{'loss': 0.2038, 'grad_norm': 0.5117580890655518, 'learning_rate': 7.720171692031437e-06, 'epoch': 1.16}

 39%|███▊      | 5130/13317 [52:52<1:25:24,  1.60it/s]
 39%|███▊      | 5131/13317 [52:52<1:24:54,  1.61it/s]
                                                      
{'loss': 0.1875, 'grad_norm': 0.49613022804260254, 'learning_rate': 7.719071893123155e-06, 'epoch': 1.16}

 39%|███▊      | 5131/13317 [52:53<1:24:54,  1.61it/s]
 39%|███▊      | 5132/13317 [52:53<1:24:36,  1.61it/s]
                                                      
{'loss': 0.1814, 'grad_norm': 0.49505510926246643, 'learning_rate': 7.717971907385844e-06, 'epoch': 1.16}

 39%|███▊      | 5132/13317 [52:53<1:24:36,  1.61it/s]
 39%|███▊      | 5133/13317 [52:54<1:24:21,  1.62it/s]
                                                      
{'loss': 0.2145, 'grad_norm': 0.6560090184211731, 'learning_rate': 7.716871734895087e-06, 'epoch': 1.16}

 39%|███▊      | 5133/13317 [52:54<1:24:21,  1.62it/s]
 39%|███▊      | 5134/13317 [52:54<1:24:08,  1.62it/s]
                                                      
{'loss': 0.191, 'grad_norm': 0.5087390542030334, 'learning_rate': 7.715771375726472e-06, 'epoch': 1.16}

 39%|███▊      | 5134/13317 [52:54<1:24:08,  1.62it/s]
 39%|███▊      | 5135/13317 [52:55<1:24:19,  1.62it/s]
                                                      
{'loss': 0.2292, 'grad_norm': 0.573222279548645, 'learning_rate': 7.714670829955608e-06, 'epoch': 1.16}

 39%|███▊      | 5135/13317 [52:55<1:24:19,  1.62it/s]
 39%|███▊      | 5136/13317 [52:56<1:24:07,  1.62it/s]
                                                      
{'loss': 0.1902, 'grad_norm': 0.5517352223396301, 'learning_rate': 7.713570097658116e-06, 'epoch': 1.16}

 39%|███▊      | 5136/13317 [52:56<1:24:07,  1.62it/s]
 39%|███▊      | 5137/13317 [52:56<1:24:05,  1.62it/s]
                                                      
{'loss': 0.2186, 'grad_norm': 0.5132723450660706, 'learning_rate': 7.712469178909628e-06, 'epoch': 1.16}

 39%|███▊      | 5137/13317 [52:56<1:24:05,  1.62it/s]
 39%|███▊      | 5138/13317 [52:57<1:23:58,  1.62it/s]
                                                      
{'loss': 0.2229, 'grad_norm': 0.5641118884086609, 'learning_rate': 7.711368073785785e-06, 'epoch': 1.16}

 39%|███▊      | 5138/13317 [52:57<1:23:58,  1.62it/s]
 39%|███▊      | 5139/13317 [52:57<1:23:50,  1.63it/s]
                                                      
{'loss': 0.1994, 'grad_norm': 0.5146247744560242, 'learning_rate': 7.710266782362248e-06, 'epoch': 1.16}

 39%|███▊      | 5139/13317 [52:58<1:23:50,  1.63it/s]
 39%|███▊      | 5140/13317 [52:58<1:23:52,  1.62it/s]
                                                      
{'loss': 0.2355, 'grad_norm': 0.5895352363586426, 'learning_rate': 7.709165304714685e-06, 'epoch': 1.16}

 39%|███▊      | 5140/13317 [52:58<1:23:52,  1.62it/s]
 39%|███▊      | 5141/13317 [52:59<1:23:49,  1.63it/s]
                                                      
{'loss': 0.1944, 'grad_norm': 0.5171108841896057, 'learning_rate': 7.708063640918781e-06, 'epoch': 1.16}

 39%|███▊      | 5141/13317 [52:59<1:23:49,  1.63it/s]
 39%|███▊      | 5142/13317 [52:59<1:23:51,  1.62it/s]
                                                      
{'loss': 0.1748, 'grad_norm': 0.5028201341629028, 'learning_rate': 7.706961791050233e-06, 'epoch': 1.16}

 39%|███▊      | 5142/13317 [52:59<1:23:51,  1.62it/s]
 39%|███▊      | 5143/13317 [53:00<1:23:45,  1.63it/s]
                                                      
{'loss': 0.2423, 'grad_norm': 0.6180912256240845, 'learning_rate': 7.705859755184746e-06, 'epoch': 1.16}

 39%|███▊      | 5143/13317 [53:00<1:23:45,  1.63it/s]
 39%|███▊      | 5144/13317 [53:00<1:23:44,  1.63it/s]
                                                      
{'loss': 0.1805, 'grad_norm': 0.43748679757118225, 'learning_rate': 7.704757533398047e-06, 'epoch': 1.16}

 39%|███▊      | 5144/13317 [53:01<1:23:44,  1.63it/s]
 39%|███▊      | 5145/13317 [53:01<1:23:50,  1.62it/s]
                                                      
{'loss': 0.2363, 'grad_norm': 0.5447813272476196, 'learning_rate': 7.703655125765864e-06, 'epoch': 1.16}

 39%|███▊      | 5145/13317 [53:01<1:23:50,  1.62it/s]
 39%|███▊      | 5146/13317 [53:02<1:23:43,  1.63it/s]
                                                      
{'loss': 0.2409, 'grad_norm': 0.5754573941230774, 'learning_rate': 7.702552532363948e-06, 'epoch': 1.16}

 39%|███▊      | 5146/13317 [53:02<1:23:43,  1.63it/s]
 39%|███▊      | 5147/13317 [53:02<1:23:44,  1.63it/s]
                                                      
{'loss': 0.1754, 'grad_norm': 0.4812919795513153, 'learning_rate': 7.701449753268058e-06, 'epoch': 1.16}

 39%|███▊      | 5147/13317 [53:02<1:23:44,  1.63it/s]
 39%|███▊      | 5148/13317 [53:03<1:23:44,  1.63it/s]
                                                      
{'loss': 0.211, 'grad_norm': 0.5384246110916138, 'learning_rate': 7.700346788553966e-06, 'epoch': 1.16}

 39%|███▊      | 5148/13317 [53:03<1:23:44,  1.63it/s]
 39%|███▊      | 5149/13317 [53:04<1:23:42,  1.63it/s]
                                                      
{'loss': 0.2973, 'grad_norm': 0.6848861575126648, 'learning_rate': 7.699243638297458e-06, 'epoch': 1.16}

 39%|███▊      | 5149/13317 [53:04<1:23:42,  1.63it/s]
 39%|███▊      | 5150/13317 [53:04<1:23:46,  1.62it/s]
                                                      
{'loss': 0.2088, 'grad_norm': 0.6013614535331726, 'learning_rate': 7.698140302574332e-06, 'epoch': 1.16}

 39%|███▊      | 5150/13317 [53:04<1:23:46,  1.62it/s]
 39%|███▊      | 5151/13317 [53:05<1:23:42,  1.63it/s]
                                                      
{'loss': 0.2354, 'grad_norm': 0.558944046497345, 'learning_rate': 7.6970367814604e-06, 'epoch': 1.16}

 39%|███▊      | 5151/13317 [53:05<1:23:42,  1.63it/s]
 39%|███▊      | 5152/13317 [53:05<1:23:42,  1.63it/s]
                                                      
{'loss': 0.2198, 'grad_norm': 0.4960601031780243, 'learning_rate': 7.695933075031483e-06, 'epoch': 1.16}

 39%|███▊      | 5152/13317 [53:06<1:23:42,  1.63it/s]
 39%|███▊      | 5153/13317 [53:06<1:23:43,  1.63it/s]
                                                      
{'loss': 0.2714, 'grad_norm': 0.5826730728149414, 'learning_rate': 7.69482918336342e-06, 'epoch': 1.16}

 39%|███▊      | 5153/13317 [53:06<1:23:43,  1.63it/s]
 39%|███▊      | 5154/13317 [53:07<1:23:39,  1.63it/s]
                                                      
{'loss': 0.2025, 'grad_norm': 0.5696575045585632, 'learning_rate': 7.693725106532058e-06, 'epoch': 1.16}

 39%|███▊      | 5154/13317 [53:07<1:23:39,  1.63it/s]
 39%|███▊      | 5155/13317 [53:07<1:23:44,  1.62it/s]
                                                      
{'loss': 0.1802, 'grad_norm': 0.4734730124473572, 'learning_rate': 7.69262084461326e-06, 'epoch': 1.16}

 39%|███▊      | 5155/13317 [53:07<1:23:44,  1.62it/s]
 39%|███▊      | 5156/13317 [53:08<1:23:37,  1.63it/s]
                                                      
{'loss': 0.2258, 'grad_norm': 0.6119878888130188, 'learning_rate': 7.691516397682898e-06, 'epoch': 1.16}

 39%|███▊      | 5156/13317 [53:08<1:23:37,  1.63it/s]
 39%|███▊      | 5157/13317 [53:08<1:23:37,  1.63it/s]
                                                      
{'loss': 0.2033, 'grad_norm': 0.47475773096084595, 'learning_rate': 7.690411765816864e-06, 'epoch': 1.16}

 39%|███▊      | 5157/13317 [53:09<1:23:37,  1.63it/s]
 39%|███▊      | 5158/13317 [53:09<1:23:38,  1.63it/s]
                                                      
{'loss': 0.2464, 'grad_norm': 0.5868075489997864, 'learning_rate': 7.689306949091055e-06, 'epoch': 1.16}

 39%|███▊      | 5158/13317 [53:09<1:23:38,  1.63it/s]
 39%|███▊      | 5159/13317 [53:10<1:23:32,  1.63it/s]
                                                      
{'loss': 0.1527, 'grad_norm': 0.44144463539123535, 'learning_rate': 7.688201947581382e-06, 'epoch': 1.16}

 39%|███▊      | 5159/13317 [53:10<1:23:32,  1.63it/s]
 39%|███▊      | 5160/13317 [53:10<1:23:34,  1.63it/s]
                                                      
{'loss': 0.1712, 'grad_norm': 0.4709995985031128, 'learning_rate': 7.687096761363773e-06, 'epoch': 1.16}

 39%|███▊      | 5160/13317 [53:10<1:23:34,  1.63it/s]
 39%|███▉      | 5161/13317 [53:11<1:23:30,  1.63it/s]
                                                      
{'loss': 0.1912, 'grad_norm': 0.5071984529495239, 'learning_rate': 7.685991390514165e-06, 'epoch': 1.16}

 39%|███▉      | 5161/13317 [53:11<1:23:30,  1.63it/s]
 39%|███▉      | 5162/13317 [53:12<1:23:26,  1.63it/s]
                                                      
{'loss': 0.1786, 'grad_norm': 0.5257363319396973, 'learning_rate': 7.68488583510851e-06, 'epoch': 1.16}

 39%|███▉      | 5162/13317 [53:12<1:23:26,  1.63it/s]
 39%|███▉      | 5163/13317 [53:12<1:23:27,  1.63it/s]
                                                      
{'loss': 0.2247, 'grad_norm': 0.46509453654289246, 'learning_rate': 7.683780095222767e-06, 'epoch': 1.16}

 39%|███▉      | 5163/13317 [53:12<1:23:27,  1.63it/s]
 39%|███▉      | 5164/13317 [53:13<1:23:27,  1.63it/s]
                                                      
{'loss': 0.1598, 'grad_norm': 0.46113717555999756, 'learning_rate': 7.682674170932918e-06, 'epoch': 1.16}

 39%|███▉      | 5164/13317 [53:13<1:23:27,  1.63it/s]
 39%|███▉      | 5165/13317 [53:13<1:23:25,  1.63it/s]
                                                      
{'loss': 0.202, 'grad_norm': 0.5053879618644714, 'learning_rate': 7.681568062314949e-06, 'epoch': 1.16}

 39%|███▉      | 5165/13317 [53:13<1:23:25,  1.63it/s]
 39%|███▉      | 5166/13317 [53:14<1:23:20,  1.63it/s]
                                                      
{'loss': 0.1824, 'grad_norm': 0.4996494650840759, 'learning_rate': 7.68046176944486e-06, 'epoch': 1.16}

 39%|███▉      | 5166/13317 [53:14<1:23:20,  1.63it/s]
 39%|███▉      | 5167/13317 [53:15<1:23:20,  1.63it/s]
                                                      
{'loss': 0.2064, 'grad_norm': 0.4750921428203583, 'learning_rate': 7.679355292398665e-06, 'epoch': 1.16}

 39%|███▉      | 5167/13317 [53:15<1:23:20,  1.63it/s]
 39%|███▉      | 5168/13317 [53:15<1:23:26,  1.63it/s]
                                                      
{'loss': 0.19, 'grad_norm': 0.5495511293411255, 'learning_rate': 7.678248631252393e-06, 'epoch': 1.16}

 39%|███▉      | 5168/13317 [53:15<1:23:26,  1.63it/s]
 39%|███▉      | 5169/13317 [53:16<1:23:24,  1.63it/s]
                                                      
{'loss': 0.1487, 'grad_norm': 0.47329428791999817, 'learning_rate': 7.677141786082083e-06, 'epoch': 1.16}

 39%|███▉      | 5169/13317 [53:16<1:23:24,  1.63it/s]
 39%|███▉      | 5170/13317 [53:16<1:23:21,  1.63it/s]
                                                      
{'loss': 0.1893, 'grad_norm': 0.5084052085876465, 'learning_rate': 7.676034756963786e-06, 'epoch': 1.16}

 39%|███▉      | 5170/13317 [53:17<1:23:21,  1.63it/s]
 39%|███▉      | 5171/13317 [53:17<1:23:19,  1.63it/s]
                                                      
{'loss': 0.1818, 'grad_norm': 0.47758418321609497, 'learning_rate': 7.674927543973567e-06, 'epoch': 1.16}

 39%|███▉      | 5171/13317 [53:17<1:23:19,  1.63it/s]
 39%|███▉      | 5172/13317 [53:18<1:23:16,  1.63it/s]
                                                      
{'loss': 0.1722, 'grad_norm': 0.46482178568840027, 'learning_rate': 7.673820147187503e-06, 'epoch': 1.17}

 39%|███▉      | 5172/13317 [53:18<1:23:16,  1.63it/s]
 39%|███▉      | 5173/13317 [53:18<1:23:19,  1.63it/s]
                                                      
{'loss': 0.1912, 'grad_norm': 0.5527383685112, 'learning_rate': 7.672712566681682e-06, 'epoch': 1.17}

 39%|███▉      | 5173/13317 [53:18<1:23:19,  1.63it/s]
 39%|███▉      | 5174/13317 [53:19<1:23:15,  1.63it/s]
                                                      
{'loss': 0.2006, 'grad_norm': 0.5210579037666321, 'learning_rate': 7.671604802532212e-06, 'epoch': 1.17}

 39%|███▉      | 5174/13317 [53:19<1:23:15,  1.63it/s]
 39%|███▉      | 5175/13317 [53:20<1:23:16,  1.63it/s]
                                                      
{'loss': 0.1699, 'grad_norm': 0.5275606513023376, 'learning_rate': 7.670496854815204e-06, 'epoch': 1.17}

 39%|███▉      | 5175/13317 [53:20<1:23:16,  1.63it/s]
 39%|███▉      | 5176/13317 [53:20<1:23:16,  1.63it/s]
                                                      
{'loss': 0.1563, 'grad_norm': 0.48912757635116577, 'learning_rate': 7.669388723606784e-06, 'epoch': 1.17}

 39%|███▉      | 5176/13317 [53:20<1:23:16,  1.63it/s]
 39%|███▉      | 5177/13317 [53:21<1:23:13,  1.63it/s]
                                                      
{'loss': 0.2343, 'grad_norm': 0.5823351144790649, 'learning_rate': 7.668280408983097e-06, 'epoch': 1.17}

 39%|███▉      | 5177/13317 [53:21<1:23:13,  1.63it/s]
 39%|███▉      | 5178/13317 [53:21<1:23:18,  1.63it/s]
                                                      
{'loss': 0.2094, 'grad_norm': 0.5235173106193542, 'learning_rate': 7.667171911020292e-06, 'epoch': 1.17}

 39%|███▉      | 5178/13317 [53:21<1:23:18,  1.63it/s]
 39%|███▉      | 5179/13317 [53:22<1:23:15,  1.63it/s]
                                                      
{'loss': 0.1862, 'grad_norm': 0.4813014268875122, 'learning_rate': 7.666063229794538e-06, 'epoch': 1.17}

 39%|███▉      | 5179/13317 [53:22<1:23:15,  1.63it/s]
 39%|███▉      | 5180/13317 [53:23<1:23:14,  1.63it/s]
                                                      
{'loss': 0.1686, 'grad_norm': 0.4922415316104889, 'learning_rate': 7.664954365382008e-06, 'epoch': 1.17}

 39%|███▉      | 5180/13317 [53:23<1:23:14,  1.63it/s]
 39%|███▉      | 5181/13317 [53:23<1:23:16,  1.63it/s]
                                                      
{'loss': 0.1673, 'grad_norm': 0.4659979045391083, 'learning_rate': 7.663845317858901e-06, 'epoch': 1.17}

 39%|███▉      | 5181/13317 [53:23<1:23:16,  1.63it/s]
 39%|███▉      | 5182/13317 [53:24<1:23:16,  1.63it/s]
                                                      
{'loss': 0.2255, 'grad_norm': 0.5625462532043457, 'learning_rate': 7.662736087301413e-06, 'epoch': 1.17}

 39%|███▉      | 5182/13317 [53:24<1:23:16,  1.63it/s]
 39%|███▉      | 5183/13317 [53:24<1:23:18,  1.63it/s]
                                                      
{'loss': 0.2275, 'grad_norm': 0.5687175989151001, 'learning_rate': 7.661626673785762e-06, 'epoch': 1.17}

 39%|███▉      | 5183/13317 [53:25<1:23:18,  1.63it/s]
 39%|███▉      | 5184/13317 [53:25<1:23:27,  1.62it/s]
                                                      
{'loss': 0.176, 'grad_norm': 0.4981166124343872, 'learning_rate': 7.660517077388176e-06, 'epoch': 1.17}

 39%|███▉      | 5184/13317 [53:25<1:23:27,  1.62it/s]
 39%|███▉      | 5185/13317 [53:26<1:23:22,  1.63it/s]
                                                      
{'loss': 0.1637, 'grad_norm': 0.5251676440238953, 'learning_rate': 7.659407298184899e-06, 'epoch': 1.17}

 39%|███▉      | 5185/13317 [53:26<1:23:22,  1.63it/s]
 39%|███▉      | 5186/13317 [53:26<1:23:19,  1.63it/s]
                                                      
{'loss': 0.2627, 'grad_norm': 0.5849807262420654, 'learning_rate': 7.658297336252181e-06, 'epoch': 1.17}

 39%|███▉      | 5186/13317 [53:26<1:23:19,  1.63it/s]
 39%|███▉      | 5187/13317 [53:27<1:23:16,  1.63it/s]
                                                      
{'loss': 0.2015, 'grad_norm': 0.567481517791748, 'learning_rate': 7.65718719166629e-06, 'epoch': 1.17}

 39%|███▉      | 5187/13317 [53:27<1:23:16,  1.63it/s]
 39%|███▉      | 5188/13317 [53:27<1:23:19,  1.63it/s]
                                                      
{'loss': 0.2198, 'grad_norm': 0.6658103466033936, 'learning_rate': 7.656076864503504e-06, 'epoch': 1.17}

 39%|███▉      | 5188/13317 [53:28<1:23:19,  1.63it/s]
 39%|███▉      | 5189/13317 [53:28<1:23:17,  1.63it/s]
                                                      
{'loss': 0.2108, 'grad_norm': 0.5380929708480835, 'learning_rate': 7.654966354840114e-06, 'epoch': 1.17}

 39%|███▉      | 5189/13317 [53:28<1:23:17,  1.63it/s]
 39%|███▉      | 5190/13317 [53:29<1:23:19,  1.63it/s]
                                                      
{'loss': 0.1764, 'grad_norm': 0.5015786290168762, 'learning_rate': 7.653855662752425e-06, 'epoch': 1.17}

 39%|███▉      | 5190/13317 [53:29<1:23:19,  1.63it/s]
 39%|███▉      | 5191/13317 [53:29<1:23:13,  1.63it/s]
                                                      
{'loss': 0.1696, 'grad_norm': 0.4625073969364166, 'learning_rate': 7.652744788316753e-06, 'epoch': 1.17}

 39%|███▉      | 5191/13317 [53:29<1:23:13,  1.63it/s]
 39%|███▉      | 5192/13317 [53:30<1:23:14,  1.63it/s]
                                                      
{'loss': 0.2149, 'grad_norm': 0.5293362736701965, 'learning_rate': 7.651633731609424e-06, 'epoch': 1.17}

 39%|███▉      | 5192/13317 [53:30<1:23:14,  1.63it/s]
 39%|███▉      | 5193/13317 [53:31<1:23:17,  1.63it/s]
                                                      
{'loss': 0.1807, 'grad_norm': 0.4245104193687439, 'learning_rate': 7.650522492706784e-06, 'epoch': 1.17}

 39%|███▉      | 5193/13317 [53:31<1:23:17,  1.63it/s]
 39%|███▉      | 5194/13317 [53:31<1:23:11,  1.63it/s]
                                                      
{'loss': 0.2009, 'grad_norm': 0.5194981694221497, 'learning_rate': 7.649411071685183e-06, 'epoch': 1.17}

 39%|███▉      | 5194/13317 [53:31<1:23:11,  1.63it/s]
 39%|███▉      | 5195/13317 [53:32<1:23:18,  1.62it/s]
                                                      
{'loss': 0.2303, 'grad_norm': 0.4759531617164612, 'learning_rate': 7.64829946862099e-06, 'epoch': 1.17}

 39%|███▉      | 5195/13317 [53:32<1:23:18,  1.62it/s]
 39%|███▉      | 5196/13317 [53:32<1:23:12,  1.63it/s]
                                                      
{'loss': 0.2232, 'grad_norm': 0.5502264499664307, 'learning_rate': 7.64718768359058e-06, 'epoch': 1.17}

 39%|███▉      | 5196/13317 [53:33<1:23:12,  1.63it/s]
 39%|███▉      | 5197/13317 [53:33<1:23:08,  1.63it/s]
                                                      
{'loss': 0.2429, 'grad_norm': 0.6582298874855042, 'learning_rate': 7.646075716670351e-06, 'epoch': 1.17}

 39%|███▉      | 5197/13317 [53:33<1:23:08,  1.63it/s]
 39%|███▉      | 5198/13317 [53:34<1:23:13,  1.63it/s]
                                                      
{'loss': 0.2034, 'grad_norm': 0.5010103583335876, 'learning_rate': 7.644963567936703e-06, 'epoch': 1.17}

 39%|███▉      | 5198/13317 [53:34<1:23:13,  1.63it/s]
 39%|███▉      | 5199/13317 [53:34<1:23:08,  1.63it/s]
                                                      
{'loss': 0.1762, 'grad_norm': 0.43072471022605896, 'learning_rate': 7.643851237466052e-06, 'epoch': 1.17}

 39%|███▉      | 5199/13317 [53:34<1:23:08,  1.63it/s]
 39%|███▉      | 5200/13317 [53:35<1:23:11,  1.63it/s]
                                                      
{'loss': 0.2306, 'grad_norm': 0.4955955445766449, 'learning_rate': 7.642738725334826e-06, 'epoch': 1.17}

 39%|███▉      | 5200/13317 [53:35<1:23:11,  1.63it/s]
 39%|███▉      | 5201/13317 [53:35<1:23:09,  1.63it/s]
                                                      
{'loss': 0.1949, 'grad_norm': 0.4827199876308441, 'learning_rate': 7.64162603161947e-06, 'epoch': 1.17}

 39%|███▉      | 5201/13317 [53:36<1:23:09,  1.63it/s]
 39%|███▉      | 5202/13317 [53:36<1:23:05,  1.63it/s]
                                                      
{'loss': 0.1557, 'grad_norm': 0.4489825367927551, 'learning_rate': 7.640513156396434e-06, 'epoch': 1.17}

 39%|███▉      | 5202/13317 [53:36<1:23:05,  1.63it/s]
 39%|███▉      | 5203/13317 [53:37<1:23:05,  1.63it/s]
                                                      
{'loss': 0.2614, 'grad_norm': 0.6173912882804871, 'learning_rate': 7.639400099742185e-06, 'epoch': 1.17}

 39%|███▉      | 5203/13317 [53:37<1:23:05,  1.63it/s]
 39%|███▉      | 5204/13317 [53:37<1:23:02,  1.63it/s]
                                                      
{'loss': 0.216, 'grad_norm': 0.5239077210426331, 'learning_rate': 7.638286861733203e-06, 'epoch': 1.17}

 39%|███▉      | 5204/13317 [53:37<1:23:02,  1.63it/s]
 39%|███▉      | 5205/13317 [53:38<1:23:25,  1.62it/s]
                                                      
{'loss': 0.2129, 'grad_norm': 0.5175224542617798, 'learning_rate': 7.637173442445983e-06, 'epoch': 1.17}

 39%|███▉      | 5205/13317 [53:38<1:23:25,  1.62it/s]
 39%|███▉      | 5206/13317 [53:39<1:23:14,  1.62it/s]
                                                      
{'loss': 0.2237, 'grad_norm': 0.5094324350357056, 'learning_rate': 7.63605984195702e-06, 'epoch': 1.17}

 39%|███▉      | 5206/13317 [53:39<1:23:14,  1.62it/s]
 39%|███▉      | 5207/13317 [53:39<1:23:06,  1.63it/s]
                                                      
{'loss': 0.1405, 'grad_norm': 0.47388070821762085, 'learning_rate': 7.634946060342837e-06, 'epoch': 1.17}

 39%|███▉      | 5207/13317 [53:39<1:23:06,  1.63it/s]
 39%|███▉      | 5208/13317 [53:40<1:23:07,  1.63it/s]
                                                      
{'loss': 0.1962, 'grad_norm': 0.5937091112136841, 'learning_rate': 7.633832097679959e-06, 'epoch': 1.17}

 39%|███▉      | 5208/13317 [53:40<1:23:07,  1.63it/s]
 39%|███▉      | 5209/13317 [53:40<1:23:03,  1.63it/s]
                                                      
{'loss': 0.1599, 'grad_norm': 0.4215715229511261, 'learning_rate': 7.632717954044929e-06, 'epoch': 1.17}

 39%|███▉      | 5209/13317 [53:41<1:23:03,  1.63it/s]
 39%|███▉      | 5210/13317 [53:41<1:23:05,  1.63it/s]
                                                      
{'loss': 0.2089, 'grad_norm': 0.5850642323493958, 'learning_rate': 7.631603629514302e-06, 'epoch': 1.17}

 39%|███▉      | 5210/13317 [53:41<1:23:05,  1.63it/s]
 39%|███▉      | 5211/13317 [53:42<1:23:00,  1.63it/s]
                                                      
{'loss': 0.1444, 'grad_norm': 0.44539085030555725, 'learning_rate': 7.63048912416464e-06, 'epoch': 1.17}

 39%|███▉      | 5211/13317 [53:42<1:23:00,  1.63it/s]
 39%|███▉      | 5212/13317 [53:42<1:23:01,  1.63it/s]
                                                      
{'loss': 0.1816, 'grad_norm': 0.5194801092147827, 'learning_rate': 7.629374438072521e-06, 'epoch': 1.17}

 39%|███▉      | 5212/13317 [53:42<1:23:01,  1.63it/s]
 39%|███▉      | 5213/13317 [53:43<1:23:05,  1.63it/s]
                                                      
{'loss': 0.1987, 'grad_norm': 0.5421858429908752, 'learning_rate': 7.62825957131454e-06, 'epoch': 1.17}

 39%|███▉      | 5213/13317 [53:43<1:23:05,  1.63it/s]
 39%|███▉      | 5214/13317 [53:43<1:23:02,  1.63it/s]
                                                      
{'loss': 0.1803, 'grad_norm': 0.5054981112480164, 'learning_rate': 7.627144523967298e-06, 'epoch': 1.17}

 39%|███▉      | 5214/13317 [53:44<1:23:02,  1.63it/s]
 39%|███▉      | 5215/13317 [53:44<1:23:05,  1.63it/s]
                                                      
{'loss': 0.2773, 'grad_norm': 0.6390038132667542, 'learning_rate': 7.626029296107411e-06, 'epoch': 1.17}

 39%|███▉      | 5215/13317 [53:44<1:23:05,  1.63it/s]
 39%|███▉      | 5216/13317 [53:45<1:23:02,  1.63it/s]
                                                      
{'loss': 0.186, 'grad_norm': 0.562642514705658, 'learning_rate': 7.624913887811505e-06, 'epoch': 1.18}

 39%|███▉      | 5216/13317 [53:45<1:23:02,  1.63it/s]
 39%|███▉      | 5217/13317 [53:45<1:23:00,  1.63it/s]
                                                      
{'loss': 0.2107, 'grad_norm': 0.6012914180755615, 'learning_rate': 7.623798299156225e-06, 'epoch': 1.18}

 39%|███▉      | 5217/13317 [53:45<1:23:00,  1.63it/s]
 39%|███▉      | 5218/13317 [53:46<1:23:00,  1.63it/s]
                                                      
{'loss': 0.1921, 'grad_norm': 0.48305264115333557, 'learning_rate': 7.6226825302182175e-06, 'epoch': 1.18}

 39%|███▉      | 5218/13317 [53:46<1:23:00,  1.63it/s]
 39%|███▉      | 5219/13317 [53:47<1:22:57,  1.63it/s]
                                                      
{'loss': 0.2248, 'grad_norm': 0.5983714461326599, 'learning_rate': 7.621566581074152e-06, 'epoch': 1.18}

 39%|███▉      | 5219/13317 [53:47<1:22:57,  1.63it/s]
 39%|███▉      | 5220/13317 [53:47<1:22:57,  1.63it/s]
                                                      
{'loss': 0.2233, 'grad_norm': 0.5991196036338806, 'learning_rate': 7.620450451800704e-06, 'epoch': 1.18}

 39%|███▉      | 5220/13317 [53:47<1:22:57,  1.63it/s]
 39%|███▉      | 5221/13317 [53:48<1:22:54,  1.63it/s]
                                                      
{'loss': 0.1523, 'grad_norm': 0.4794026017189026, 'learning_rate': 7.619334142474563e-06, 'epoch': 1.18}

 39%|███▉      | 5221/13317 [53:48<1:22:54,  1.63it/s]
 39%|███▉      | 5222/13317 [53:48<1:22:53,  1.63it/s]
                                                      
{'loss': 0.1673, 'grad_norm': 0.447312593460083, 'learning_rate': 7.618217653172436e-06, 'epoch': 1.18}

 39%|███▉      | 5222/13317 [53:49<1:22:53,  1.63it/s]
 39%|███▉      | 5223/13317 [53:49<1:23:08,  1.62it/s]
                                                      
{'loss': 0.2615, 'grad_norm': 0.6628432869911194, 'learning_rate': 7.617100983971031e-06, 'epoch': 1.18}

 39%|███▉      | 5223/13317 [53:49<1:23:08,  1.62it/s]
 39%|███▉      | 5224/13317 [53:50<1:22:58,  1.63it/s]
                                                      
{'loss': 0.1677, 'grad_norm': 0.4604329466819763, 'learning_rate': 7.615984134947081e-06, 'epoch': 1.18}

 39%|███▉      | 5224/13317 [53:50<1:22:58,  1.63it/s]
 39%|███▉      | 5225/13317 [53:50<1:23:01,  1.62it/s]
                                                      
{'loss': 0.238, 'grad_norm': 0.5384863615036011, 'learning_rate': 7.614867106177319e-06, 'epoch': 1.18}

 39%|███▉      | 5225/13317 [53:50<1:23:01,  1.62it/s]
 39%|███▉      | 5226/13317 [53:51<1:22:54,  1.63it/s]
                                                      
{'loss': 0.1458, 'grad_norm': 0.4152052104473114, 'learning_rate': 7.613749897738502e-06, 'epoch': 1.18}

 39%|███▉      | 5226/13317 [53:51<1:22:54,  1.63it/s]
 39%|███▉      | 5227/13317 [53:51<1:22:55,  1.63it/s]
                                                      
{'loss': 0.2266, 'grad_norm': 0.5728364586830139, 'learning_rate': 7.6126325097073935e-06, 'epoch': 1.18}

 39%|███▉      | 5227/13317 [53:52<1:22:55,  1.63it/s]
 39%|███▉      | 5228/13317 [53:52<1:22:54,  1.63it/s]
                                                      
{'loss': 0.1833, 'grad_norm': 0.4678460359573364, 'learning_rate': 7.611514942160767e-06, 'epoch': 1.18}

 39%|███▉      | 5228/13317 [53:52<1:22:54,  1.63it/s]
 39%|███▉      | 5229/13317 [53:53<1:22:54,  1.63it/s]
                                                      
{'loss': 0.2215, 'grad_norm': 0.5400174260139465, 'learning_rate': 7.610397195175414e-06, 'epoch': 1.18}

 39%|███▉      | 5229/13317 [53:53<1:22:54,  1.63it/s]
 39%|███▉      | 5230/13317 [53:53<1:22:57,  1.62it/s]
                                                      
{'loss': 0.2641, 'grad_norm': 0.6569886207580566, 'learning_rate': 7.609279268828131e-06, 'epoch': 1.18}

 39%|███▉      | 5230/13317 [53:53<1:22:57,  1.62it/s]
 39%|███▉      | 5231/13317 [53:54<1:22:51,  1.63it/s]
                                                      
{'loss': 0.1466, 'grad_norm': 0.42827796936035156, 'learning_rate': 7.608161163195738e-06, 'epoch': 1.18}

 39%|███▉      | 5231/13317 [53:54<1:22:51,  1.63it/s]
 39%|███▉      | 5232/13317 [53:55<1:22:50,  1.63it/s]
                                                      
{'loss': 0.1763, 'grad_norm': 0.5992549657821655, 'learning_rate': 7.607042878355058e-06, 'epoch': 1.18}

 39%|███▉      | 5232/13317 [53:55<1:22:50,  1.63it/s]
 39%|███▉      | 5233/13317 [53:55<1:22:50,  1.63it/s]
                                                      
{'loss': 0.1438, 'grad_norm': 0.4211927056312561, 'learning_rate': 7.605924414382926e-06, 'epoch': 1.18}

 39%|███▉      | 5233/13317 [53:55<1:22:50,  1.63it/s]
 39%|███▉      | 5234/13317 [53:56<1:22:46,  1.63it/s]
                                                      
{'loss': 0.1392, 'grad_norm': 0.4181504547595978, 'learning_rate': 7.604805771356196e-06, 'epoch': 1.18}

 39%|███▉      | 5234/13317 [53:56<1:22:46,  1.63it/s]
 39%|███▉      | 5235/13317 [53:56<1:22:47,  1.63it/s]
                                                      
{'loss': 0.22, 'grad_norm': 0.5385854244232178, 'learning_rate': 7.603686949351729e-06, 'epoch': 1.18}

 39%|███▉      | 5235/13317 [53:57<1:22:47,  1.63it/s]
 39%|███▉      | 5236/13317 [53:57<1:22:45,  1.63it/s]
                                                      
{'loss': 0.227, 'grad_norm': 0.5348944664001465, 'learning_rate': 7.6025679484464e-06, 'epoch': 1.18}

 39%|███▉      | 5236/13317 [53:57<1:22:45,  1.63it/s]
 39%|███▉      | 5237/13317 [53:58<1:22:45,  1.63it/s]
                                                      
{'loss': 0.1607, 'grad_norm': 0.45039814710617065, 'learning_rate': 7.601448768717098e-06, 'epoch': 1.18}

 39%|███▉      | 5237/13317 [53:58<1:22:45,  1.63it/s]
 39%|███▉      | 5238/13317 [53:58<1:22:45,  1.63it/s]
                                                      
{'loss': 0.2305, 'grad_norm': 0.600490152835846, 'learning_rate': 7.60032941024072e-06, 'epoch': 1.18}

 39%|███▉      | 5238/13317 [53:58<1:22:45,  1.63it/s]
 39%|███▉      | 5239/13317 [53:59<1:22:41,  1.63it/s]
                                                      
{'loss': 0.1481, 'grad_norm': 0.4097258448600769, 'learning_rate': 7.599209873094178e-06, 'epoch': 1.18}

 39%|███▉      | 5239/13317 [53:59<1:22:41,  1.63it/s]
 39%|███▉      | 5240/13317 [53:59<1:22:38,  1.63it/s]
                                                      
{'loss': 0.2299, 'grad_norm': 0.5837856531143188, 'learning_rate': 7.5980901573543985e-06, 'epoch': 1.18}

 39%|███▉      | 5240/13317 [54:00<1:22:38,  1.63it/s]
 39%|███▉      | 5241/13317 [54:00<1:22:38,  1.63it/s]
                                                      
{'loss': 0.2237, 'grad_norm': 0.5392671227455139, 'learning_rate': 7.596970263098315e-06, 'epoch': 1.18}

 39%|███▉      | 5241/13317 [54:00<1:22:38,  1.63it/s]
 39%|███▉      | 5242/13317 [54:01<1:23:01,  1.62it/s]
                                                      
{'loss': 0.2428, 'grad_norm': 0.530426025390625, 'learning_rate': 7.595850190402877e-06, 'epoch': 1.18}

 39%|███▉      | 5242/13317 [54:01<1:23:01,  1.62it/s]
 39%|███▉      | 5243/13317 [54:01<1:22:48,  1.62it/s]
                                                      
{'loss': 0.1881, 'grad_norm': 0.4772510826587677, 'learning_rate': 7.594729939345045e-06, 'epoch': 1.18}

 39%|███▉      | 5243/13317 [54:01<1:22:48,  1.62it/s]
 39%|███▉      | 5244/13317 [54:02<1:22:43,  1.63it/s]
                                                      
{'loss': 0.1625, 'grad_norm': 0.4832664430141449, 'learning_rate': 7.593609510001796e-06, 'epoch': 1.18}

 39%|███▉      | 5244/13317 [54:02<1:22:43,  1.63it/s]
 39%|███▉      | 5245/13317 [54:03<1:22:45,  1.63it/s]
                                                      
{'loss': 0.2289, 'grad_norm': 0.6056073904037476, 'learning_rate': 7.59248890245011e-06, 'epoch': 1.18}

 39%|███▉      | 5245/13317 [54:03<1:22:45,  1.63it/s]
 39%|███▉      | 5246/13317 [54:03<1:22:39,  1.63it/s]
                                                      
{'loss': 0.1826, 'grad_norm': 0.4885769188404083, 'learning_rate': 7.5913681167669865e-06, 'epoch': 1.18}

 39%|███▉      | 5246/13317 [54:03<1:22:39,  1.63it/s]
 39%|███▉      | 5247/13317 [54:04<1:22:37,  1.63it/s]
                                                      
{'loss': 0.1919, 'grad_norm': 0.45524126291275024, 'learning_rate': 7.590247153029435e-06, 'epoch': 1.18}

 39%|███▉      | 5247/13317 [54:04<1:22:37,  1.63it/s]
 39%|███▉      | 5248/13317 [54:04<1:22:35,  1.63it/s]
                                                      
{'loss': 0.161, 'grad_norm': 0.4365697503089905, 'learning_rate': 7.589126011314478e-06, 'epoch': 1.18}

 39%|███▉      | 5248/13317 [54:05<1:22:35,  1.63it/s]
 39%|███▉      | 5249/13317 [54:05<1:22:32,  1.63it/s]
                                                      
{'loss': 0.2234, 'grad_norm': 0.49719464778900146, 'learning_rate': 7.588004691699152e-06, 'epoch': 1.18}

 39%|███▉      | 5249/13317 [54:05<1:22:32,  1.63it/s]
 39%|███▉      | 5250/13317 [54:06<1:22:30,  1.63it/s]
                                                      
{'loss': 0.2086, 'grad_norm': 0.5787680149078369, 'learning_rate': 7.586883194260499e-06, 'epoch': 1.18}

 39%|███▉      | 5250/13317 [54:06<1:22:30,  1.63it/s]
 39%|███▉      | 5251/13317 [54:06<1:22:32,  1.63it/s]
                                                      
{'loss': 0.2576, 'grad_norm': 0.4892262816429138, 'learning_rate': 7.585761519075583e-06, 'epoch': 1.18}

 39%|███▉      | 5251/13317 [54:06<1:22:32,  1.63it/s]
 39%|███▉      | 5252/13317 [54:07<1:22:27,  1.63it/s]
                                                      
{'loss': 0.1844, 'grad_norm': 0.45320993661880493, 'learning_rate': 7.584639666221469e-06, 'epoch': 1.18}

 39%|███▉      | 5252/13317 [54:07<1:22:27,  1.63it/s]
 39%|███▉      | 5253/13317 [54:07<1:22:26,  1.63it/s]
                                                      
{'loss': 0.241, 'grad_norm': 0.6310199499130249, 'learning_rate': 7.583517635775246e-06, 'epoch': 1.18}

 39%|███▉      | 5253/13317 [54:08<1:22:26,  1.63it/s]
 39%|███▉      | 5254/13317 [54:08<1:22:23,  1.63it/s]
                                                      
{'loss': 0.1923, 'grad_norm': 0.489910364151001, 'learning_rate': 7.582395427814005e-06, 'epoch': 1.18}

 39%|███▉      | 5254/13317 [54:08<1:22:23,  1.63it/s]
 39%|███▉      | 5255/13317 [54:09<1:22:22,  1.63it/s]
                                                      
{'loss': 0.2053, 'grad_norm': 0.5243050456047058, 'learning_rate': 7.581273042414856e-06, 'epoch': 1.18}

 39%|███▉      | 5255/13317 [54:09<1:22:22,  1.63it/s]
 39%|███▉      | 5256/13317 [54:09<1:22:22,  1.63it/s]
                                                      
{'loss': 0.2186, 'grad_norm': 0.5652254819869995, 'learning_rate': 7.580150479654919e-06, 'epoch': 1.18}

 39%|███▉      | 5256/13317 [54:09<1:22:22,  1.63it/s]
 39%|███▉      | 5257/13317 [54:10<1:22:20,  1.63it/s]
                                                      
{'loss': 0.187, 'grad_norm': 0.5088774561882019, 'learning_rate': 7.579027739611322e-06, 'epoch': 1.18}

 39%|███▉      | 5257/13317 [54:10<1:22:20,  1.63it/s]
 39%|███▉      | 5258/13317 [54:11<1:22:22,  1.63it/s]
                                                      
{'loss': 0.2041, 'grad_norm': 0.5568956136703491, 'learning_rate': 7.577904822361215e-06, 'epoch': 1.18}

 39%|███▉      | 5258/13317 [54:11<1:22:22,  1.63it/s]
 39%|███▉      | 5259/13317 [54:11<1:22:20,  1.63it/s]
                                                      
{'loss': 0.2116, 'grad_norm': 0.5467513799667358, 'learning_rate': 7.5767817279817505e-06, 'epoch': 1.18}

 39%|███▉      | 5259/13317 [54:11<1:22:20,  1.63it/s]
 39%|███▉      | 5260/13317 [54:12<1:22:21,  1.63it/s]
                                                      
{'loss': 0.2279, 'grad_norm': 0.516986072063446, 'learning_rate': 7.5756584565500965e-06, 'epoch': 1.18}

 39%|███▉      | 5260/13317 [54:12<1:22:21,  1.63it/s]
 40%|███▉      | 5261/13317 [54:12<1:22:19,  1.63it/s]
                                                      
{'loss': 0.292, 'grad_norm': 0.49185478687286377, 'learning_rate': 7.574535008143437e-06, 'epoch': 1.19}

 40%|███▉      | 5261/13317 [54:12<1:22:19,  1.63it/s]
 40%|███▉      | 5262/13317 [54:13<1:22:17,  1.63it/s]
                                                      
{'loss': 0.1668, 'grad_norm': 0.4273604452610016, 'learning_rate': 7.573411382838962e-06, 'epoch': 1.19}

 40%|███▉      | 5262/13317 [54:13<1:22:17,  1.63it/s]
 40%|███▉      | 5263/13317 [54:14<1:22:19,  1.63it/s]
                                                      
{'loss': 0.2267, 'grad_norm': 0.5522693991661072, 'learning_rate': 7.572287580713877e-06, 'epoch': 1.19}

 40%|███▉      | 5263/13317 [54:14<1:22:19,  1.63it/s]
 40%|███▉      | 5264/13317 [54:14<1:22:19,  1.63it/s]
                                                      
{'loss': 0.1796, 'grad_norm': 0.5372216105461121, 'learning_rate': 7.571163601845399e-06, 'epoch': 1.19}

 40%|███▉      | 5264/13317 [54:14<1:22:19,  1.63it/s]
 40%|███▉      | 5265/13317 [54:15<1:22:17,  1.63it/s]
                                                      
{'loss': 0.2325, 'grad_norm': 0.5475426316261292, 'learning_rate': 7.57003944631076e-06, 'epoch': 1.19}

 40%|███▉      | 5265/13317 [54:15<1:22:17,  1.63it/s]
 40%|███▉      | 5266/13317 [54:15<1:22:22,  1.63it/s]
                                                      
{'loss': 0.2474, 'grad_norm': 0.6225082874298096, 'learning_rate': 7.568915114187196e-06, 'epoch': 1.19}

 40%|███▉      | 5266/13317 [54:16<1:22:22,  1.63it/s]
 40%|███▉      | 5267/13317 [54:16<1:22:23,  1.63it/s]
                                                      
{'loss': 0.2528, 'grad_norm': 0.5288520455360413, 'learning_rate': 7.567790605551966e-06, 'epoch': 1.19}

 40%|███▉      | 5267/13317 [54:16<1:22:23,  1.63it/s]
 40%|███▉      | 5268/13317 [54:17<1:22:23,  1.63it/s]
                                                      
{'loss': 0.2092, 'grad_norm': 0.6110365390777588, 'learning_rate': 7.566665920482333e-06, 'epoch': 1.19}

 40%|███▉      | 5268/13317 [54:17<1:22:23,  1.63it/s]
 40%|███▉      | 5269/13317 [54:17<1:22:19,  1.63it/s]
                                                      
{'loss': 0.2103, 'grad_norm': 0.5079768896102905, 'learning_rate': 7.565541059055572e-06, 'epoch': 1.19}

 40%|███▉      | 5269/13317 [54:17<1:22:19,  1.63it/s]
 40%|███▉      | 5270/13317 [54:18<1:22:18,  1.63it/s]
                                                      
{'loss': 0.2578, 'grad_norm': 0.6588887572288513, 'learning_rate': 7.56441602134898e-06, 'epoch': 1.19}

 40%|███▉      | 5270/13317 [54:18<1:22:18,  1.63it/s]
 40%|███▉      | 5271/13317 [54:18<1:22:14,  1.63it/s]
                                                      
{'loss': 0.1681, 'grad_norm': 0.4674152731895447, 'learning_rate': 7.563290807439853e-06, 'epoch': 1.19}

 40%|███▉      | 5271/13317 [54:19<1:22:14,  1.63it/s]
 40%|███▉      | 5272/13317 [54:19<1:22:12,  1.63it/s]
                                                      
{'loss': 0.1669, 'grad_norm': 0.49287891387939453, 'learning_rate': 7.562165417405507e-06, 'epoch': 1.19}

 40%|███▉      | 5272/13317 [54:19<1:22:12,  1.63it/s]
 40%|███▉      | 5273/13317 [54:20<1:22:17,  1.63it/s]
                                                      
{'loss': 0.3259, 'grad_norm': 0.7233138680458069, 'learning_rate': 7.561039851323267e-06, 'epoch': 1.19}

 40%|███▉      | 5273/13317 [54:20<1:22:17,  1.63it/s]
 40%|███▉      | 5274/13317 [54:20<1:22:13,  1.63it/s]
                                                      
{'loss': 0.1892, 'grad_norm': 0.5273439884185791, 'learning_rate': 7.559914109270474e-06, 'epoch': 1.19}

 40%|███▉      | 5274/13317 [54:20<1:22:13,  1.63it/s]
 40%|███▉      | 5275/13317 [54:21<1:22:11,  1.63it/s]
                                                      
{'loss': 0.2218, 'grad_norm': 0.5579224228858948, 'learning_rate': 7.558788191324475e-06, 'epoch': 1.19}

 40%|███▉      | 5275/13317 [54:21<1:22:11,  1.63it/s]
 40%|███▉      | 5276/13317 [54:22<1:22:10,  1.63it/s]
                                                      
{'loss': 0.1755, 'grad_norm': 0.45055148005485535, 'learning_rate': 7.557662097562637e-06, 'epoch': 1.19}

 40%|███▉      | 5276/13317 [54:22<1:22:10,  1.63it/s]
 40%|███▉      | 5277/13317 [54:22<1:22:08,  1.63it/s]
                                                      
{'loss': 0.1689, 'grad_norm': 0.5298584699630737, 'learning_rate': 7.55653582806233e-06, 'epoch': 1.19}

 40%|███▉      | 5277/13317 [54:22<1:22:08,  1.63it/s]
 40%|███▉      | 5278/13317 [54:23<1:22:08,  1.63it/s]
                                                      
{'loss': 0.1745, 'grad_norm': 0.46176955103874207, 'learning_rate': 7.555409382900944e-06, 'epoch': 1.19}

 40%|███▉      | 5278/13317 [54:23<1:22:08,  1.63it/s]
 40%|███▉      | 5279/13317 [54:23<1:22:09,  1.63it/s]
                                                      
{'loss': 0.2265, 'grad_norm': 0.5159664154052734, 'learning_rate': 7.554282762155874e-06, 'epoch': 1.19}

 40%|███▉      | 5279/13317 [54:24<1:22:09,  1.63it/s]
 40%|███▉      | 5280/13317 [54:24<1:22:07,  1.63it/s]
                                                      
{'loss': 0.2343, 'grad_norm': 0.5539946556091309, 'learning_rate': 7.553155965904535e-06, 'epoch': 1.19}

 40%|███▉      | 5280/13317 [54:24<1:22:07,  1.63it/s]
 40%|███▉      | 5281/13317 [54:25<1:22:06,  1.63it/s]
                                                      
{'loss': 0.1836, 'grad_norm': 0.44516369700431824, 'learning_rate': 7.5520289942243475e-06, 'epoch': 1.19}

 40%|███▉      | 5281/13317 [54:25<1:22:06,  1.63it/s]
 40%|███▉      | 5282/13317 [54:25<1:22:10,  1.63it/s]
                                                      
{'loss': 0.1812, 'grad_norm': 0.5109217762947083, 'learning_rate': 7.550901847192748e-06, 'epoch': 1.19}

 40%|███▉      | 5282/13317 [54:25<1:22:10,  1.63it/s]
 40%|███▉      | 5283/13317 [54:26<1:22:06,  1.63it/s]
                                                      
{'loss': 0.2194, 'grad_norm': 0.5858359336853027, 'learning_rate': 7.549774524887181e-06, 'epoch': 1.19}

 40%|███▉      | 5283/13317 [54:26<1:22:06,  1.63it/s]
 40%|███▉      | 5284/13317 [54:26<1:22:05,  1.63it/s]
                                                      
{'loss': 0.1994, 'grad_norm': 0.4690369665622711, 'learning_rate': 7.548647027385106e-06, 'epoch': 1.19}

 40%|███▉      | 5284/13317 [54:27<1:22:05,  1.63it/s]
 40%|███▉      | 5285/13317 [54:27<1:22:06,  1.63it/s]
                                                      
{'loss': 0.1518, 'grad_norm': 0.4282287359237671, 'learning_rate': 7.547519354763996e-06, 'epoch': 1.19}

 40%|███▉      | 5285/13317 [54:27<1:22:06,  1.63it/s]
 40%|███▉      | 5286/13317 [54:28<1:22:04,  1.63it/s]
                                                      
{'loss': 0.234, 'grad_norm': 0.5603505969047546, 'learning_rate': 7.5463915071013315e-06, 'epoch': 1.19}

 40%|███▉      | 5286/13317 [54:28<1:22:04,  1.63it/s]
 40%|███▉      | 5287/13317 [54:28<1:22:02,  1.63it/s]
                                                      
{'loss': 0.2225, 'grad_norm': 0.5762304663658142, 'learning_rate': 7.54526348447461e-06, 'epoch': 1.19}

 40%|███▉      | 5287/13317 [54:28<1:22:02,  1.63it/s]
 40%|███▉      | 5288/13317 [54:29<1:22:03,  1.63it/s]
                                                      
{'loss': 0.1898, 'grad_norm': 0.5267202854156494, 'learning_rate': 7.544135286961336e-06, 'epoch': 1.19}

 40%|███▉      | 5288/13317 [54:29<1:22:03,  1.63it/s]
 40%|███▉      | 5289/13317 [54:30<1:22:01,  1.63it/s]
                                                      
{'loss': 0.2389, 'grad_norm': 0.5741389989852905, 'learning_rate': 7.543006914639032e-06, 'epoch': 1.19}

 40%|███▉      | 5289/13317 [54:30<1:22:01,  1.63it/s]
 40%|███▉      | 5290/13317 [54:30<1:22:00,  1.63it/s]
                                                      
{'loss': 0.1419, 'grad_norm': 0.4191892743110657, 'learning_rate': 7.541878367585225e-06, 'epoch': 1.19}

 40%|███▉      | 5290/13317 [54:30<1:22:00,  1.63it/s]
 40%|███▉      | 5291/13317 [54:31<1:22:01,  1.63it/s]
                                                      
{'loss': 0.2015, 'grad_norm': 0.5535183548927307, 'learning_rate': 7.540749645877462e-06, 'epoch': 1.19}

 40%|███▉      | 5291/13317 [54:31<1:22:01,  1.63it/s]
 40%|███▉      | 5292/13317 [54:31<1:22:09,  1.63it/s]
                                                      
{'loss': 0.2022, 'grad_norm': 0.5013806819915771, 'learning_rate': 7.539620749593297e-06, 'epoch': 1.19}

 40%|███▉      | 5292/13317 [54:32<1:22:09,  1.63it/s]
 40%|███▉      | 5293/13317 [54:32<1:22:03,  1.63it/s]
                                                      
{'loss': 0.163, 'grad_norm': 0.49326735734939575, 'learning_rate': 7.538491678810295e-06, 'epoch': 1.19}

 40%|███▉      | 5293/13317 [54:32<1:22:03,  1.63it/s]
 40%|███▉      | 5294/13317 [54:33<1:22:02,  1.63it/s]
                                                      
{'loss': 0.2439, 'grad_norm': 0.5808943510055542, 'learning_rate': 7.537362433606037e-06, 'epoch': 1.19}

 40%|███▉      | 5294/13317 [54:33<1:22:02,  1.63it/s]
 40%|███▉      | 5295/13317 [54:33<1:22:02,  1.63it/s]
                                                      
{'loss': 0.2431, 'grad_norm': 0.5693042278289795, 'learning_rate': 7.536233014058114e-06, 'epoch': 1.19}

 40%|███▉      | 5295/13317 [54:33<1:22:02,  1.63it/s]
 40%|███▉      | 5296/13317 [54:34<1:22:22,  1.62it/s]
                                                      
{'loss': 0.171, 'grad_norm': 0.4212907552719116, 'learning_rate': 7.535103420244129e-06, 'epoch': 1.19}

 40%|███▉      | 5296/13317 [54:34<1:22:22,  1.62it/s]
 40%|███▉      | 5297/13317 [54:34<1:22:14,  1.63it/s]
                                                      
{'loss': 0.2161, 'grad_norm': 0.5457378625869751, 'learning_rate': 7.5339736522416965e-06, 'epoch': 1.19}

 40%|███▉      | 5297/13317 [54:35<1:22:14,  1.63it/s]
 40%|███▉      | 5298/13317 [54:35<1:22:14,  1.63it/s]
                                                      
{'loss': 0.2131, 'grad_norm': 0.6361829042434692, 'learning_rate': 7.5328437101284446e-06, 'epoch': 1.19}

 40%|███▉      | 5298/13317 [54:35<1:22:14,  1.63it/s]
 40%|███▉      | 5299/13317 [54:36<1:22:07,  1.63it/s]
                                                      
{'loss': 0.22, 'grad_norm': 0.563101589679718, 'learning_rate': 7.531713593982011e-06, 'epoch': 1.19}

 40%|███▉      | 5299/13317 [54:36<1:22:07,  1.63it/s]
 40%|███▉      | 5300/13317 [54:36<1:22:05,  1.63it/s]
                                                      
{'loss': 0.1807, 'grad_norm': 0.5192630887031555, 'learning_rate': 7.530583303880048e-06, 'epoch': 1.19}

 40%|███▉      | 5300/13317 [54:36<1:22:05,  1.63it/s]
 40%|███▉      | 5301/13317 [54:37<1:22:01,  1.63it/s]
                                                      
{'loss': 0.2199, 'grad_norm': 0.646413266658783, 'learning_rate': 7.529452839900217e-06, 'epoch': 1.19}

 40%|███▉      | 5301/13317 [54:37<1:22:01,  1.63it/s]
 40%|███▉      | 5302/13317 [54:38<1:22:04,  1.63it/s]
                                                      
{'loss': 0.2883, 'grad_norm': 0.6359382271766663, 'learning_rate': 7.5283222021201925e-06, 'epoch': 1.19}

 40%|███▉      | 5302/13317 [54:38<1:22:04,  1.63it/s]
 40%|███▉      | 5303/13317 [54:38<1:22:01,  1.63it/s]
                                                      
{'loss': 0.1489, 'grad_norm': 0.4634985029697418, 'learning_rate': 7.527191390617665e-06, 'epoch': 1.19}

 40%|███▉      | 5303/13317 [54:38<1:22:01,  1.63it/s]
 40%|███▉      | 5304/13317 [54:39<1:21:58,  1.63it/s]
                                                      
{'loss': 0.2177, 'grad_norm': 0.5817708969116211, 'learning_rate': 7.526060405470329e-06, 'epoch': 1.19}

 40%|███▉      | 5304/13317 [54:39<1:21:58,  1.63it/s]
 40%|███▉      | 5305/13317 [54:39<1:22:00,  1.63it/s]
                                                      
{'loss': 0.1782, 'grad_norm': 0.49674808979034424, 'learning_rate': 7.5249292467558974e-06, 'epoch': 1.2}

 40%|███▉      | 5305/13317 [54:39<1:22:00,  1.63it/s]
 40%|███▉      | 5306/13317 [54:40<1:22:01,  1.63it/s]
                                                      
{'loss': 0.2541, 'grad_norm': 0.563177227973938, 'learning_rate': 7.5237979145520915e-06, 'epoch': 1.2}

 40%|███▉      | 5306/13317 [54:40<1:22:01,  1.63it/s]
 40%|███▉      | 5307/13317 [54:41<1:22:04,  1.63it/s]
                                                      
{'loss': 0.2167, 'grad_norm': 0.5168395042419434, 'learning_rate': 7.5226664089366474e-06, 'epoch': 1.2}

 40%|███▉      | 5307/13317 [54:41<1:22:04,  1.63it/s]
 40%|███▉      | 5308/13317 [54:41<1:22:01,  1.63it/s]
                                                      
{'loss': 0.1837, 'grad_norm': 0.49192366003990173, 'learning_rate': 7.52153472998731e-06, 'epoch': 1.2}

 40%|███▉      | 5308/13317 [54:41<1:22:01,  1.63it/s]
 40%|███▉      | 5309/13317 [54:42<1:22:00,  1.63it/s]
                                                      
{'loss': 0.1956, 'grad_norm': 0.4738052487373352, 'learning_rate': 7.520402877781839e-06, 'epoch': 1.2}

 40%|███▉      | 5309/13317 [54:42<1:22:00,  1.63it/s]
 40%|███▉      | 5310/13317 [54:42<1:22:07,  1.62it/s]
                                                      
{'loss': 0.2554, 'grad_norm': 0.591044008731842, 'learning_rate': 7.519270852398002e-06, 'epoch': 1.2}

 40%|███▉      | 5310/13317 [54:43<1:22:07,  1.62it/s]
 40%|███▉      | 5311/13317 [54:43<1:22:05,  1.63it/s]
                                                      
{'loss': 0.1733, 'grad_norm': 0.4833470582962036, 'learning_rate': 7.518138653913585e-06, 'epoch': 1.2}

 40%|███▉      | 5311/13317 [54:43<1:22:05,  1.63it/s]
 40%|███▉      | 5312/13317 [54:44<1:22:00,  1.63it/s]
                                                      
{'loss': 0.223, 'grad_norm': 0.5645701289176941, 'learning_rate': 7.517006282406379e-06, 'epoch': 1.2}

 40%|███▉      | 5312/13317 [54:44<1:22:00,  1.63it/s]
 40%|███▉      | 5313/13317 [54:44<1:21:57,  1.63it/s]
                                                      
{'loss': 0.2012, 'grad_norm': 0.497243732213974, 'learning_rate': 7.515873737954191e-06, 'epoch': 1.2}

 40%|███▉      | 5313/13317 [54:44<1:21:57,  1.63it/s]
 40%|███▉      | 5314/13317 [54:45<1:21:52,  1.63it/s]
                                                      
{'loss': 0.1717, 'grad_norm': 0.4604412019252777, 'learning_rate': 7.514741020634839e-06, 'epoch': 1.2}

 40%|███▉      | 5314/13317 [54:45<1:21:52,  1.63it/s]
 40%|███▉      | 5315/13317 [54:46<1:21:57,  1.63it/s]
                                                      
{'loss': 0.24, 'grad_norm': 0.543682336807251, 'learning_rate': 7.513608130526152e-06, 'epoch': 1.2}

 40%|███▉      | 5315/13317 [54:46<1:21:57,  1.63it/s]
 40%|███▉      | 5316/13317 [54:46<1:21:56,  1.63it/s]
                                                      
{'loss': 0.1847, 'grad_norm': 0.46904706954956055, 'learning_rate': 7.512475067705972e-06, 'epoch': 1.2}

 40%|███▉      | 5316/13317 [54:46<1:21:56,  1.63it/s]
 40%|███▉      | 5317/13317 [54:47<1:21:54,  1.63it/s]
                                                      
{'loss': 0.2536, 'grad_norm': 0.6587782502174377, 'learning_rate': 7.511341832252153e-06, 'epoch': 1.2}

 40%|███▉      | 5317/13317 [54:47<1:21:54,  1.63it/s]
 40%|███▉      | 5318/13317 [54:47<1:21:54,  1.63it/s]
                                                      
{'loss': 0.2059, 'grad_norm': 0.4662324786186218, 'learning_rate': 7.510208424242558e-06, 'epoch': 1.2}

 40%|███▉      | 5318/13317 [54:47<1:21:54,  1.63it/s]
 40%|███▉      | 5319/13317 [54:48<1:21:50,  1.63it/s]
                                                      
{'loss': 0.153, 'grad_norm': 0.43708670139312744, 'learning_rate': 7.509074843755069e-06, 'epoch': 1.2}

 40%|███▉      | 5319/13317 [54:48<1:21:50,  1.63it/s]
 40%|███▉      | 5320/13317 [54:49<1:21:56,  1.63it/s]
                                                      
{'loss': 0.2527, 'grad_norm': 0.672680139541626, 'learning_rate': 7.507941090867569e-06, 'epoch': 1.2}

 40%|███▉      | 5320/13317 [54:49<1:21:56,  1.63it/s]
 40%|███▉      | 5321/13317 [54:49<1:21:53,  1.63it/s]
                                                      
{'loss': 0.2151, 'grad_norm': 0.5229685306549072, 'learning_rate': 7.506807165657961e-06, 'epoch': 1.2}

 40%|███▉      | 5321/13317 [54:49<1:21:53,  1.63it/s]
 40%|███▉      | 5322/13317 [54:50<1:21:50,  1.63it/s]
                                                      
{'loss': 0.1927, 'grad_norm': 0.4825805425643921, 'learning_rate': 7.5056730682041605e-06, 'epoch': 1.2}

 40%|███▉      | 5322/13317 [54:50<1:21:50,  1.63it/s]
 40%|███▉      | 5323/13317 [54:50<1:21:50,  1.63it/s]
                                                      
{'loss': 0.1926, 'grad_norm': 0.5453116297721863, 'learning_rate': 7.504538798584087e-06, 'epoch': 1.2}

 40%|███▉      | 5323/13317 [54:51<1:21:50,  1.63it/s]
 40%|███▉      | 5324/13317 [54:51<1:21:47,  1.63it/s]
                                                      
{'loss': 0.2212, 'grad_norm': 0.5635163187980652, 'learning_rate': 7.503404356875681e-06, 'epoch': 1.2}

 40%|███▉      | 5324/13317 [54:51<1:21:47,  1.63it/s]
 40%|███▉      | 5325/13317 [54:52<1:21:53,  1.63it/s]
                                                      
{'loss': 0.2111, 'grad_norm': 0.4994923174381256, 'learning_rate': 7.502269743156889e-06, 'epoch': 1.2}

 40%|███▉      | 5325/13317 [54:52<1:21:53,  1.63it/s]
 40%|███▉      | 5326/13317 [54:52<1:21:51,  1.63it/s]
                                                      
{'loss': 0.1829, 'grad_norm': 0.4776623547077179, 'learning_rate': 7.501134957505672e-06, 'epoch': 1.2}

 40%|███▉      | 5326/13317 [54:52<1:21:51,  1.63it/s]
 40%|████      | 5327/13317 [54:53<1:21:47,  1.63it/s]
                                                      
{'loss': 0.1961, 'grad_norm': 0.5367742776870728, 'learning_rate': 7.500000000000001e-06, 'epoch': 1.2}

 40%|████      | 5327/13317 [54:53<1:21:47,  1.63it/s]
 40%|████      | 5328/13317 [54:53<1:21:45,  1.63it/s]
                                                      
{'loss': 0.1731, 'grad_norm': 0.4291606545448303, 'learning_rate': 7.498864870717859e-06, 'epoch': 1.2}

 40%|████      | 5328/13317 [54:54<1:21:45,  1.63it/s]
 40%|████      | 5329/13317 [54:54<1:21:44,  1.63it/s]
                                                      
{'loss': 0.2022, 'grad_norm': 0.49633145332336426, 'learning_rate': 7.4977295697372425e-06, 'epoch': 1.2}

 40%|████      | 5329/13317 [54:54<1:21:44,  1.63it/s]
 40%|████      | 5330/13317 [54:55<1:21:47,  1.63it/s]
                                                      
{'loss': 0.1905, 'grad_norm': 0.5189366340637207, 'learning_rate': 7.496594097136158e-06, 'epoch': 1.2}

 40%|████      | 5330/13317 [54:55<1:21:47,  1.63it/s]
 40%|████      | 5331/13317 [54:55<1:21:44,  1.63it/s]
                                                      
{'loss': 0.2206, 'grad_norm': 0.5041818022727966, 'learning_rate': 7.495458452992624e-06, 'epoch': 1.2}

 40%|████      | 5331/13317 [54:55<1:21:44,  1.63it/s]
 40%|████      | 5332/13317 [54:56<1:21:41,  1.63it/s]
                                                      
{'loss': 0.2768, 'grad_norm': 0.6055068969726562, 'learning_rate': 7.494322637384673e-06, 'epoch': 1.2}

 40%|████      | 5332/13317 [54:56<1:21:41,  1.63it/s]
 40%|████      | 5333/13317 [54:57<1:21:38,  1.63it/s]
                                                      
{'loss': 0.2061, 'grad_norm': 0.6053724884986877, 'learning_rate': 7.493186650390344e-06, 'epoch': 1.2}

 40%|████      | 5333/13317 [54:57<1:21:38,  1.63it/s]
 40%|████      | 5334/13317 [54:57<1:21:36,  1.63it/s]
                                                      
{'loss': 0.1785, 'grad_norm': 0.5526111721992493, 'learning_rate': 7.492050492087697e-06, 'epoch': 1.2}

 40%|████      | 5334/13317 [54:57<1:21:36,  1.63it/s]
 40%|████      | 5335/13317 [54:58<1:21:45,  1.63it/s]
                                                      
{'loss': 0.201, 'grad_norm': 0.5377492308616638, 'learning_rate': 7.490914162554792e-06, 'epoch': 1.2}

 40%|████      | 5335/13317 [54:58<1:21:45,  1.63it/s]
 40%|████      | 5336/13317 [54:58<1:21:43,  1.63it/s]
                                                      
{'loss': 0.2604, 'grad_norm': 0.5628995299339294, 'learning_rate': 7.4897776618697114e-06, 'epoch': 1.2}

 40%|████      | 5336/13317 [54:59<1:21:43,  1.63it/s]
 40%|████      | 5337/13317 [54:59<1:21:40,  1.63it/s]
                                                      
{'loss': 0.1751, 'grad_norm': 0.48674169182777405, 'learning_rate': 7.488640990110544e-06, 'epoch': 1.2}

 40%|████      | 5337/13317 [54:59<1:21:40,  1.63it/s]
 40%|████      | 5338/13317 [55:00<1:21:38,  1.63it/s]
                                                      
{'loss': 0.2556, 'grad_norm': 0.5576059222221375, 'learning_rate': 7.4875041473553886e-06, 'epoch': 1.2}

 40%|████      | 5338/13317 [55:00<1:21:38,  1.63it/s]
 40%|████      | 5339/13317 [55:00<1:21:36,  1.63it/s]
                                                      
{'loss': 0.2005, 'grad_norm': 0.5136443376541138, 'learning_rate': 7.486367133682363e-06, 'epoch': 1.2}

 40%|████      | 5339/13317 [55:00<1:21:36,  1.63it/s]
 40%|████      | 5340/13317 [55:01<1:21:40,  1.63it/s]
                                                      
{'loss': 0.222, 'grad_norm': 0.5085086226463318, 'learning_rate': 7.485229949169585e-06, 'epoch': 1.2}

 40%|████      | 5340/13317 [55:01<1:21:40,  1.63it/s]
 40%|████      | 5341/13317 [55:01<1:21:38,  1.63it/s]
                                                      
{'loss': 0.2051, 'grad_norm': 0.5513023734092712, 'learning_rate': 7.484092593895199e-06, 'epoch': 1.2}

 40%|████      | 5341/13317 [55:02<1:21:38,  1.63it/s]
 40%|████      | 5342/13317 [55:02<1:21:34,  1.63it/s]
                                                      
{'loss': 0.1756, 'grad_norm': 0.5017558932304382, 'learning_rate': 7.4829550679373495e-06, 'epoch': 1.2}

 40%|████      | 5342/13317 [55:02<1:21:34,  1.63it/s]
 40%|████      | 5343/13317 [55:03<1:21:34,  1.63it/s]
                                                      
{'loss': 0.1825, 'grad_norm': 0.4667315185070038, 'learning_rate': 7.481817371374196e-06, 'epoch': 1.2}

 40%|████      | 5343/13317 [55:03<1:21:34,  1.63it/s]
 40%|████      | 5344/13317 [55:03<1:21:34,  1.63it/s]
                                                      
{'loss': 0.1999, 'grad_norm': 0.4315888583660126, 'learning_rate': 7.480679504283911e-06, 'epoch': 1.2}

 40%|████      | 5344/13317 [55:03<1:21:34,  1.63it/s]
 40%|████      | 5345/13317 [55:04<1:21:51,  1.62it/s]
                                                      
{'loss': 0.2154, 'grad_norm': 0.4505422115325928, 'learning_rate': 7.479541466744678e-06, 'epoch': 1.2}

 40%|████      | 5345/13317 [55:04<1:21:51,  1.62it/s]
 40%|████      | 5346/13317 [55:05<1:21:45,  1.63it/s]
                                                      
{'loss': 0.1876, 'grad_norm': 0.4494840204715729, 'learning_rate': 7.478403258834693e-06, 'epoch': 1.2}

 40%|████      | 5346/13317 [55:05<1:21:45,  1.63it/s]
 40%|████      | 5347/13317 [55:05<1:21:39,  1.63it/s]
                                                      
{'loss': 0.1684, 'grad_norm': 0.43820124864578247, 'learning_rate': 7.477264880632163e-06, 'epoch': 1.2}

 40%|████      | 5347/13317 [55:05<1:21:39,  1.63it/s]
 40%|████      | 5348/13317 [55:06<1:21:38,  1.63it/s]
                                                      
{'loss': 0.1988, 'grad_norm': 0.4774411916732788, 'learning_rate': 7.476126332215305e-06, 'epoch': 1.2}

 40%|████      | 5348/13317 [55:06<1:21:38,  1.63it/s]
 40%|████      | 5349/13317 [55:06<1:21:34,  1.63it/s]
                                                      
{'loss': 0.2613, 'grad_norm': 0.5598804354667664, 'learning_rate': 7.47498761366235e-06, 'epoch': 1.21}

 40%|████      | 5349/13317 [55:07<1:21:34,  1.63it/s]
 40%|████      | 5350/13317 [55:07<1:21:34,  1.63it/s]
                                                      
{'loss': 0.2106, 'grad_norm': 0.5335102081298828, 'learning_rate': 7.473848725051541e-06, 'epoch': 1.21}

 40%|████      | 5350/13317 [55:07<1:21:34,  1.63it/s]
 40%|████      | 5351/13317 [55:08<1:21:34,  1.63it/s]
                                                      
{'loss': 0.2207, 'grad_norm': 0.5355555415153503, 'learning_rate': 7.472709666461131e-06, 'epoch': 1.21}

 40%|████      | 5351/13317 [55:08<1:21:34,  1.63it/s]
 40%|████      | 5352/13317 [55:08<1:22:02,  1.62it/s]
                                                      
{'loss': 0.2122, 'grad_norm': 0.5082271099090576, 'learning_rate': 7.471570437969385e-06, 'epoch': 1.21}

 40%|████      | 5352/13317 [55:08<1:22:02,  1.62it/s]
 40%|████      | 5353/13317 [55:09<1:21:58,  1.62it/s]
                                                      
{'loss': 0.2473, 'grad_norm': 0.6952736377716064, 'learning_rate': 7.470431039654581e-06, 'epoch': 1.21}

 40%|████      | 5353/13317 [55:09<1:21:58,  1.62it/s]
 40%|████      | 5354/13317 [55:09<1:21:50,  1.62it/s]
                                                      
{'loss': 0.1976, 'grad_norm': 0.5364501476287842, 'learning_rate': 7.469291471595008e-06, 'epoch': 1.21}

 40%|████      | 5354/13317 [55:10<1:21:50,  1.62it/s]
 40%|████      | 5355/13317 [55:10<1:21:45,  1.62it/s]
                                                      
{'loss': 0.2435, 'grad_norm': 0.575680136680603, 'learning_rate': 7.468151733868964e-06, 'epoch': 1.21}

 40%|████      | 5355/13317 [55:10<1:21:45,  1.62it/s]
 40%|████      | 5356/13317 [55:11<1:21:37,  1.63it/s]
                                                      
{'loss': 0.2279, 'grad_norm': 0.5791232585906982, 'learning_rate': 7.467011826554764e-06, 'epoch': 1.21}

 40%|████      | 5356/13317 [55:11<1:21:37,  1.63it/s]
 40%|████      | 5357/13317 [55:11<1:21:39,  1.62it/s]
                                                      
{'loss': 0.1994, 'grad_norm': 0.45850563049316406, 'learning_rate': 7.46587174973073e-06, 'epoch': 1.21}

 40%|████      | 5357/13317 [55:11<1:21:39,  1.62it/s]
 40%|████      | 5358/13317 [55:12<1:21:35,  1.63it/s]
                                                      
{'loss': 0.1581, 'grad_norm': 0.4799567461013794, 'learning_rate': 7.464731503475198e-06, 'epoch': 1.21}

 40%|████      | 5358/13317 [55:12<1:21:35,  1.63it/s]
 40%|████      | 5359/13317 [55:13<1:21:31,  1.63it/s]
                                                      
{'loss': 0.2233, 'grad_norm': 0.4756634831428528, 'learning_rate': 7.463591087866514e-06, 'epoch': 1.21}

 40%|████      | 5359/13317 [55:13<1:21:31,  1.63it/s]
 40%|████      | 5360/13317 [55:13<1:21:35,  1.63it/s]
                                                      
{'loss': 0.2099, 'grad_norm': 0.6196479797363281, 'learning_rate': 7.462450502983036e-06, 'epoch': 1.21}

 40%|████      | 5360/13317 [55:13<1:21:35,  1.63it/s]
 40%|████      | 5361/13317 [55:14<1:21:27,  1.63it/s]
                                                      
{'loss': 0.2274, 'grad_norm': 0.5063779354095459, 'learning_rate': 7.461309748903138e-06, 'epoch': 1.21}

 40%|████      | 5361/13317 [55:14<1:21:27,  1.63it/s]
 40%|████      | 5362/13317 [55:14<1:21:26,  1.63it/s]
                                                      
{'loss': 0.1965, 'grad_norm': 0.5280309319496155, 'learning_rate': 7.4601688257052e-06, 'epoch': 1.21}

 40%|████      | 5362/13317 [55:15<1:21:26,  1.63it/s]
 40%|████      | 5363/13317 [55:15<1:21:33,  1.63it/s]
                                                      
{'loss': 0.1965, 'grad_norm': 0.5306863784790039, 'learning_rate': 7.459027733467613e-06, 'epoch': 1.21}

 40%|████      | 5363/13317 [55:15<1:21:33,  1.63it/s]
 40%|████      | 5364/13317 [55:16<1:21:28,  1.63it/s]
                                                      
{'loss': 0.1574, 'grad_norm': 0.43923622369766235, 'learning_rate': 7.457886472268785e-06, 'epoch': 1.21}

 40%|████      | 5364/13317 [55:16<1:21:28,  1.63it/s]
 40%|████      | 5365/13317 [55:16<1:21:29,  1.63it/s]
                                                      
{'loss': 0.1871, 'grad_norm': 0.5241957306861877, 'learning_rate': 7.456745042187131e-06, 'epoch': 1.21}

 40%|████      | 5365/13317 [55:16<1:21:29,  1.63it/s]
 40%|████      | 5366/13317 [55:17<1:21:24,  1.63it/s]
                                                      
{'loss': 0.2394, 'grad_norm': 0.5574789047241211, 'learning_rate': 7.455603443301081e-06, 'epoch': 1.21}

 40%|████      | 5366/13317 [55:17<1:21:24,  1.63it/s]
 40%|████      | 5367/13317 [55:17<1:21:21,  1.63it/s]
                                                      
{'loss': 0.2291, 'grad_norm': 0.5927870869636536, 'learning_rate': 7.454461675689073e-06, 'epoch': 1.21}

 40%|████      | 5367/13317 [55:18<1:21:21,  1.63it/s]
 40%|████      | 5368/13317 [55:18<1:21:31,  1.63it/s]
                                                      
{'loss': 0.192, 'grad_norm': 0.6213877201080322, 'learning_rate': 7.453319739429562e-06, 'epoch': 1.21}

 40%|████      | 5368/13317 [55:18<1:21:31,  1.63it/s]
 40%|████      | 5369/13317 [55:19<1:21:27,  1.63it/s]
                                                      
{'loss': 0.2172, 'grad_norm': 0.5168600678443909, 'learning_rate': 7.452177634601007e-06, 'epoch': 1.21}

 40%|████      | 5369/13317 [55:19<1:21:27,  1.63it/s]
 40%|████      | 5370/13317 [55:19<1:21:30,  1.62it/s]
                                                      
{'loss': 0.2543, 'grad_norm': 0.6854785084724426, 'learning_rate': 7.451035361281884e-06, 'epoch': 1.21}

 40%|████      | 5370/13317 [55:19<1:21:30,  1.62it/s]
 40%|████      | 5371/13317 [55:20<1:21:33,  1.62it/s]
                                                      
{'loss': 0.28, 'grad_norm': 0.7481321096420288, 'learning_rate': 7.4498929195506805e-06, 'epoch': 1.21}

 40%|████      | 5371/13317 [55:20<1:21:33,  1.62it/s]
 40%|████      | 5372/13317 [55:21<1:21:30,  1.62it/s]
                                                      
{'loss': 0.186, 'grad_norm': 0.48836830258369446, 'learning_rate': 7.4487503094858935e-06, 'epoch': 1.21}

 40%|████      | 5372/13317 [55:21<1:21:30,  1.62it/s]
 40%|████      | 5373/13317 [55:21<1:21:30,  1.62it/s]
                                                      
{'loss': 0.1854, 'grad_norm': 0.46559202671051025, 'learning_rate': 7.447607531166034e-06, 'epoch': 1.21}

 40%|████      | 5373/13317 [55:21<1:21:30,  1.62it/s]
 40%|████      | 5374/13317 [55:22<1:21:30,  1.62it/s]
                                                      
{'loss': 0.2161, 'grad_norm': 0.48691830039024353, 'learning_rate': 7.4464645846696186e-06, 'epoch': 1.21}

 40%|████      | 5374/13317 [55:22<1:21:30,  1.62it/s]
 40%|████      | 5375/13317 [55:22<1:21:32,  1.62it/s]
                                                      
{'loss': 0.2191, 'grad_norm': 0.4851442873477936, 'learning_rate': 7.445321470075184e-06, 'epoch': 1.21}

 40%|████      | 5375/13317 [55:23<1:21:32,  1.62it/s]
 40%|████      | 5376/13317 [55:23<1:21:24,  1.63it/s]
                                                      
{'loss': 0.1775, 'grad_norm': 0.4961547553539276, 'learning_rate': 7.4441781874612725e-06, 'epoch': 1.21}

 40%|████      | 5376/13317 [55:23<1:21:24,  1.63it/s]
 40%|████      | 5377/13317 [55:24<1:21:27,  1.62it/s]
                                                      
{'loss': 0.1822, 'grad_norm': 0.4640215039253235, 'learning_rate': 7.443034736906441e-06, 'epoch': 1.21}

 40%|████      | 5377/13317 [55:24<1:21:27,  1.62it/s]
 40%|████      | 5378/13317 [55:24<1:21:40,  1.62it/s]
                                                      
{'loss': 0.198, 'grad_norm': 0.5355121493339539, 'learning_rate': 7.441891118489255e-06, 'epoch': 1.21}

 40%|████      | 5378/13317 [55:24<1:21:40,  1.62it/s]
 40%|████      | 5379/13317 [55:25<1:21:30,  1.62it/s]
                                                      
{'loss': 0.2068, 'grad_norm': 0.5113304257392883, 'learning_rate': 7.4407473322882935e-06, 'epoch': 1.21}

 40%|████      | 5379/13317 [55:25<1:21:30,  1.62it/s]
 40%|████      | 5380/13317 [55:25<1:21:28,  1.62it/s]
                                                      
{'loss': 0.1855, 'grad_norm': 0.5405104756355286, 'learning_rate': 7.439603378382148e-06, 'epoch': 1.21}

 40%|████      | 5380/13317 [55:26<1:21:28,  1.62it/s]
 40%|████      | 5381/13317 [55:26<1:21:23,  1.62it/s]
                                                      
{'loss': 0.2387, 'grad_norm': 0.6166291832923889, 'learning_rate': 7.43845925684942e-06, 'epoch': 1.21}

 40%|████      | 5381/13317 [55:26<1:21:23,  1.62it/s]
 40%|████      | 5382/13317 [55:27<1:21:24,  1.62it/s]
                                                      
{'loss': 0.1896, 'grad_norm': 0.49954792857170105, 'learning_rate': 7.437314967768723e-06, 'epoch': 1.21}

 40%|████      | 5382/13317 [55:27<1:21:24,  1.62it/s]
 40%|████      | 5383/13317 [55:27<1:21:21,  1.63it/s]
                                                      
{'loss': 0.1469, 'grad_norm': 0.47334742546081543, 'learning_rate': 7.436170511218679e-06, 'epoch': 1.21}

 40%|████      | 5383/13317 [55:27<1:21:21,  1.63it/s]
 40%|████      | 5384/13317 [55:28<1:21:19,  1.63it/s]
                                                      
{'loss': 0.1797, 'grad_norm': 0.47291895747184753, 'learning_rate': 7.435025887277927e-06, 'epoch': 1.21}

 40%|████      | 5384/13317 [55:28<1:21:19,  1.63it/s]
 40%|████      | 5385/13317 [55:29<1:21:23,  1.62it/s]
                                                      
{'loss': 0.2158, 'grad_norm': 0.5728932619094849, 'learning_rate': 7.433881096025114e-06, 'epoch': 1.21}

 40%|████      | 5385/13317 [55:29<1:21:23,  1.62it/s]
 40%|████      | 5386/13317 [55:29<1:21:15,  1.63it/s]
                                                      
{'loss': 0.1659, 'grad_norm': 0.4107605218887329, 'learning_rate': 7.4327361375389e-06, 'epoch': 1.21}

 40%|████      | 5386/13317 [55:29<1:21:15,  1.63it/s]
 40%|████      | 5387/13317 [55:30<1:21:11,  1.63it/s]
                                                      
{'loss': 0.1456, 'grad_norm': 0.42189210653305054, 'learning_rate': 7.431591011897954e-06, 'epoch': 1.21}

 40%|████      | 5387/13317 [55:30<1:21:11,  1.63it/s]
 40%|████      | 5388/13317 [55:30<1:21:08,  1.63it/s]
                                                      
{'loss': 0.1913, 'grad_norm': 0.4410170018672943, 'learning_rate': 7.4304457191809605e-06, 'epoch': 1.21}

 40%|████      | 5388/13317 [55:31<1:21:08,  1.63it/s]
 40%|████      | 5389/13317 [55:31<1:21:05,  1.63it/s]
                                                      
{'loss': 0.2106, 'grad_norm': 0.6148915886878967, 'learning_rate': 7.42930025946661e-06, 'epoch': 1.21}

 40%|████      | 5389/13317 [55:31<1:21:05,  1.63it/s]
 40%|████      | 5390/13317 [55:32<1:21:10,  1.63it/s]
                                                      
{'loss': 0.2096, 'grad_norm': 0.5302537083625793, 'learning_rate': 7.428154632833613e-06, 'epoch': 1.21}

 40%|████      | 5390/13317 [55:32<1:21:10,  1.63it/s]
 40%|████      | 5391/13317 [55:32<1:21:08,  1.63it/s]
                                                      
{'loss': 0.1691, 'grad_norm': 0.46292442083358765, 'learning_rate': 7.427008839360681e-06, 'epoch': 1.21}

 40%|████      | 5391/13317 [55:32<1:21:08,  1.63it/s]
 40%|████      | 5392/13317 [55:33<1:21:06,  1.63it/s]
                                                      
{'loss': 0.1646, 'grad_norm': 0.46055322885513306, 'learning_rate': 7.425862879126544e-06, 'epoch': 1.21}

 40%|████      | 5392/13317 [55:33<1:21:06,  1.63it/s]
 40%|████      | 5393/13317 [55:33<1:21:09,  1.63it/s]
                                                      
{'loss': 0.1735, 'grad_norm': 0.460683673620224, 'learning_rate': 7.424716752209944e-06, 'epoch': 1.21}

 40%|████      | 5393/13317 [55:34<1:21:09,  1.63it/s]
 41%|████      | 5394/13317 [55:34<1:21:07,  1.63it/s]
                                                      
{'loss': 0.2386, 'grad_norm': 0.5370166301727295, 'learning_rate': 7.423570458689626e-06, 'epoch': 1.22}

 41%|████      | 5394/13317 [55:34<1:21:07,  1.63it/s]
 41%|████      | 5395/13317 [55:35<1:21:08,  1.63it/s]
                                                      
{'loss': 0.2791, 'grad_norm': 0.6322448253631592, 'learning_rate': 7.4224239986443594e-06, 'epoch': 1.22}

 41%|████      | 5395/13317 [55:35<1:21:08,  1.63it/s]
 41%|████      | 5396/13317 [55:35<1:21:07,  1.63it/s]
                                                      
{'loss': 0.1691, 'grad_norm': 0.461333692073822, 'learning_rate': 7.421277372152913e-06, 'epoch': 1.22}

 41%|████      | 5396/13317 [55:35<1:21:07,  1.63it/s]
 41%|████      | 5397/13317 [55:36<1:21:04,  1.63it/s]
                                                      
{'loss': 0.2669, 'grad_norm': 0.6871747374534607, 'learning_rate': 7.420130579294076e-06, 'epoch': 1.22}

 41%|████      | 5397/13317 [55:36<1:21:04,  1.63it/s]
 41%|████      | 5398/13317 [55:37<1:21:22,  1.62it/s]
                                                      
{'loss': 0.2025, 'grad_norm': 0.5686545372009277, 'learning_rate': 7.4189836201466425e-06, 'epoch': 1.22}

 41%|████      | 5398/13317 [55:37<1:21:22,  1.62it/s]
 41%|████      | 5399/13317 [55:37<1:21:15,  1.62it/s]
                                                      
{'loss': 0.1924, 'grad_norm': 0.47339123487472534, 'learning_rate': 7.417836494789422e-06, 'epoch': 1.22}

 41%|████      | 5399/13317 [55:37<1:21:15,  1.62it/s]
 41%|████      | 5400/13317 [55:38<1:21:11,  1.63it/s]
                                                      
{'loss': 0.2636, 'grad_norm': 0.6244531869888306, 'learning_rate': 7.416689203301234e-06, 'epoch': 1.22}

 41%|████      | 5400/13317 [55:38<1:21:11,  1.63it/s]
 41%|████      | 5401/13317 [55:38<1:21:05,  1.63it/s]
                                                      
{'loss': 0.2094, 'grad_norm': 0.5316630005836487, 'learning_rate': 7.415541745760909e-06, 'epoch': 1.22}

 41%|████      | 5401/13317 [55:39<1:21:05,  1.63it/s]
 41%|████      | 5402/13317 [55:39<1:21:00,  1.63it/s]
                                                      
{'loss': 0.2285, 'grad_norm': 0.5291740298271179, 'learning_rate': 7.414394122247291e-06, 'epoch': 1.22}

 41%|████      | 5402/13317 [55:39<1:21:00,  1.63it/s]
 41%|████      | 5403/13317 [55:40<1:21:04,  1.63it/s]
                                                      
{'loss': 0.2051, 'grad_norm': 0.4869730472564697, 'learning_rate': 7.413246332839232e-06, 'epoch': 1.22}

 41%|████      | 5403/13317 [55:40<1:21:04,  1.63it/s]
 41%|████      | 5404/13317 [55:40<1:21:04,  1.63it/s]
                                                      
{'loss': 0.1976, 'grad_norm': 0.4695529639720917, 'learning_rate': 7.412098377615597e-06, 'epoch': 1.22}

 41%|████      | 5404/13317 [55:40<1:21:04,  1.63it/s]
 41%|████      | 5405/13317 [55:41<1:21:07,  1.63it/s]
                                                      
{'loss': 0.1897, 'grad_norm': 0.5009564161300659, 'learning_rate': 7.4109502566552645e-06, 'epoch': 1.22}

 41%|████      | 5405/13317 [55:41<1:21:07,  1.63it/s]
 41%|████      | 5406/13317 [55:41<1:21:05,  1.63it/s]
                                                      
{'loss': 0.1938, 'grad_norm': 0.4727855920791626, 'learning_rate': 7.409801970037123e-06, 'epoch': 1.22}

 41%|████      | 5406/13317 [55:42<1:21:05,  1.63it/s]
 41%|████      | 5407/13317 [55:42<1:21:05,  1.63it/s]
                                                      
{'loss': 0.2129, 'grad_norm': 0.5779024362564087, 'learning_rate': 7.40865351784007e-06, 'epoch': 1.22}

 41%|████      | 5407/13317 [55:42<1:21:05,  1.63it/s]
 41%|████      | 5408/13317 [55:43<1:21:22,  1.62it/s]
                                                      
{'loss': 0.1876, 'grad_norm': 0.5082996487617493, 'learning_rate': 7.407504900143017e-06, 'epoch': 1.22}

 41%|████      | 5408/13317 [55:43<1:21:22,  1.62it/s]
 41%|████      | 5409/13317 [55:43<1:21:12,  1.62it/s]
                                                      
{'loss': 0.204, 'grad_norm': 0.47684958577156067, 'learning_rate': 7.4063561170248866e-06, 'epoch': 1.22}

 41%|████      | 5409/13317 [55:43<1:21:12,  1.62it/s]
 41%|████      | 5410/13317 [55:44<1:21:08,  1.62it/s]
                                                      
{'loss': 0.1605, 'grad_norm': 0.4696855843067169, 'learning_rate': 7.4052071685646106e-06, 'epoch': 1.22}

 41%|████      | 5410/13317 [55:44<1:21:08,  1.62it/s]
 41%|████      | 5411/13317 [55:45<1:21:03,  1.63it/s]
                                                      
{'loss': 0.1707, 'grad_norm': 0.4982546865940094, 'learning_rate': 7.404058054841136e-06, 'epoch': 1.22}

 41%|████      | 5411/13317 [55:45<1:21:03,  1.63it/s]
 41%|████      | 5412/13317 [55:45<1:21:03,  1.63it/s]
                                                      
{'loss': 0.1842, 'grad_norm': 0.44270145893096924, 'learning_rate': 7.402908775933419e-06, 'epoch': 1.22}

 41%|████      | 5412/13317 [55:45<1:21:03,  1.63it/s]
 41%|████      | 5413/13317 [55:46<1:21:00,  1.63it/s]
                                                      
{'loss': 0.2039, 'grad_norm': 0.5317153930664062, 'learning_rate': 7.4017593319204275e-06, 'epoch': 1.22}

 41%|████      | 5413/13317 [55:46<1:21:00,  1.63it/s]
 41%|████      | 5414/13317 [55:46<1:20:57,  1.63it/s]
                                                      
{'loss': 0.2412, 'grad_norm': 0.5790808796882629, 'learning_rate': 7.400609722881138e-06, 'epoch': 1.22}

 41%|████      | 5414/13317 [55:47<1:20:57,  1.63it/s]
 41%|████      | 5415/13317 [55:47<1:21:18,  1.62it/s]
                                                      
{'loss': 0.2232, 'grad_norm': 0.5407944917678833, 'learning_rate': 7.399459948894544e-06, 'epoch': 1.22}

 41%|████      | 5415/13317 [55:47<1:21:18,  1.62it/s]
 41%|████      | 5416/13317 [55:48<1:21:08,  1.62it/s]
                                                      
{'loss': 0.2097, 'grad_norm': 0.52322918176651, 'learning_rate': 7.398310010039644e-06, 'epoch': 1.22}

 41%|████      | 5416/13317 [55:48<1:21:08,  1.62it/s]
 41%|████      | 5417/13317 [55:48<1:21:01,  1.62it/s]
                                                      
{'loss': 0.1733, 'grad_norm': 0.4788172245025635, 'learning_rate': 7.397159906395454e-06, 'epoch': 1.22}

 41%|████      | 5417/13317 [55:48<1:21:01,  1.62it/s]
 41%|████      | 5418/13317 [55:49<1:21:02,  1.62it/s]
                                                      
{'loss': 0.1965, 'grad_norm': 0.4912225008010864, 'learning_rate': 7.396009638040996e-06, 'epoch': 1.22}

 41%|████      | 5418/13317 [55:49<1:21:02,  1.62it/s]
 41%|████      | 5419/13317 [55:49<1:20:57,  1.63it/s]
                                                      
{'loss': 0.2025, 'grad_norm': 0.5057523250579834, 'learning_rate': 7.394859205055307e-06, 'epoch': 1.22}

 41%|████      | 5419/13317 [55:50<1:20:57,  1.63it/s]
 41%|████      | 5420/13317 [55:50<1:21:02,  1.62it/s]
                                                      
{'loss': 0.2248, 'grad_norm': 0.598592221736908, 'learning_rate': 7.393708607517435e-06, 'epoch': 1.22}

 41%|████      | 5420/13317 [55:50<1:21:02,  1.62it/s]
 41%|████      | 5421/13317 [55:51<1:20:55,  1.63it/s]
                                                      
{'loss': 0.1968, 'grad_norm': 0.4954659044742584, 'learning_rate': 7.392557845506433e-06, 'epoch': 1.22}

 41%|████      | 5421/13317 [55:51<1:20:55,  1.63it/s]
 41%|████      | 5422/13317 [55:51<1:20:55,  1.63it/s]
                                                      
{'loss': 0.2097, 'grad_norm': 0.5408459901809692, 'learning_rate': 7.3914069191013774e-06, 'epoch': 1.22}

 41%|████      | 5422/13317 [55:51<1:20:55,  1.63it/s]
 41%|████      | 5423/13317 [55:52<1:20:54,  1.63it/s]
                                                      
{'loss': 0.149, 'grad_norm': 0.4187469780445099, 'learning_rate': 7.390255828381343e-06, 'epoch': 1.22}

 41%|████      | 5423/13317 [55:52<1:20:54,  1.63it/s]
 41%|████      | 5424/13317 [55:53<1:20:51,  1.63it/s]
                                                      
{'loss': 0.2008, 'grad_norm': 0.5151179432868958, 'learning_rate': 7.389104573425427e-06, 'epoch': 1.22}

 41%|████      | 5424/13317 [55:53<1:20:51,  1.63it/s]
 41%|████      | 5425/13317 [55:53<1:20:57,  1.62it/s]
                                                      
{'loss': 0.1675, 'grad_norm': 0.4561629593372345, 'learning_rate': 7.38795315431273e-06, 'epoch': 1.22}

 41%|████      | 5425/13317 [55:53<1:20:57,  1.62it/s]
 41%|████      | 5426/13317 [55:54<1:20:53,  1.63it/s]
                                                      
{'loss': 0.2876, 'grad_norm': 0.5528080463409424, 'learning_rate': 7.3868015711223675e-06, 'epoch': 1.22}

 41%|████      | 5426/13317 [55:54<1:20:53,  1.63it/s]
 41%|████      | 5427/13317 [55:54<1:20:48,  1.63it/s]
                                                      
{'loss': 0.1616, 'grad_norm': 0.4959191083908081, 'learning_rate': 7.385649823933465e-06, 'epoch': 1.22}

 41%|████      | 5427/13317 [55:55<1:20:48,  1.63it/s]
 41%|████      | 5428/13317 [55:55<1:20:47,  1.63it/s]
                                                      
{'loss': 0.1346, 'grad_norm': 0.43361762166023254, 'learning_rate': 7.384497912825159e-06, 'epoch': 1.22}

 41%|████      | 5428/13317 [55:55<1:20:47,  1.63it/s]
 41%|████      | 5429/13317 [55:56<1:20:44,  1.63it/s]
                                                      
{'loss': 0.1764, 'grad_norm': 0.4606938362121582, 'learning_rate': 7.383345837876601e-06, 'epoch': 1.22}

 41%|████      | 5429/13317 [55:56<1:20:44,  1.63it/s]
 41%|████      | 5430/13317 [55:56<1:20:48,  1.63it/s]
                                                      
{'loss': 0.2044, 'grad_norm': 0.5725260376930237, 'learning_rate': 7.3821935991669455e-06, 'epoch': 1.22}

 41%|████      | 5430/13317 [55:56<1:20:48,  1.63it/s]
 41%|████      | 5431/13317 [55:57<1:20:44,  1.63it/s]
                                                      
{'loss': 0.1682, 'grad_norm': 0.45807361602783203, 'learning_rate': 7.381041196775369e-06, 'epoch': 1.22}

 41%|████      | 5431/13317 [55:57<1:20:44,  1.63it/s]
 41%|████      | 5432/13317 [55:57<1:20:47,  1.63it/s]
                                                      
{'loss': 0.1719, 'grad_norm': 0.4479042887687683, 'learning_rate': 7.3798886307810515e-06, 'epoch': 1.22}

 41%|████      | 5432/13317 [55:58<1:20:47,  1.63it/s]
 41%|████      | 5433/13317 [55:58<1:20:50,  1.63it/s]
                                                      
{'loss': 0.2105, 'grad_norm': 0.5160598158836365, 'learning_rate': 7.378735901263186e-06, 'epoch': 1.22}

 41%|████      | 5433/13317 [55:58<1:20:50,  1.63it/s]
 41%|████      | 5434/13317 [55:59<1:20:46,  1.63it/s]
                                                      
{'loss': 0.1878, 'grad_norm': 0.5149244070053101, 'learning_rate': 7.377583008300977e-06, 'epoch': 1.22}

 41%|████      | 5434/13317 [55:59<1:20:46,  1.63it/s]
 41%|████      | 5435/13317 [55:59<1:21:11,  1.62it/s]
                                                      
{'loss': 0.2057, 'grad_norm': 0.5326506495475769, 'learning_rate': 7.376429951973642e-06, 'epoch': 1.22}

 41%|████      | 5435/13317 [55:59<1:21:11,  1.62it/s]
 41%|████      | 5436/13317 [56:00<1:21:01,  1.62it/s]
                                                      
{'loss': 0.2318, 'grad_norm': 0.5774393081665039, 'learning_rate': 7.375276732360408e-06, 'epoch': 1.22}

 41%|████      | 5436/13317 [56:00<1:21:01,  1.62it/s]
 41%|████      | 5437/13317 [56:01<1:20:58,  1.62it/s]
                                                      
{'loss': 0.1751, 'grad_norm': 0.44467613101005554, 'learning_rate': 7.3741233495405105e-06, 'epoch': 1.22}

 41%|████      | 5437/13317 [56:01<1:20:58,  1.62it/s]
 41%|████      | 5438/13317 [56:01<1:20:55,  1.62it/s]
                                                      
{'loss': 0.1845, 'grad_norm': 0.46434295177459717, 'learning_rate': 7.372969803593202e-06, 'epoch': 1.23}

 41%|████      | 5438/13317 [56:01<1:20:55,  1.62it/s]
 41%|████      | 5439/13317 [56:02<1:20:49,  1.62it/s]
                                                      
{'loss': 0.1645, 'grad_norm': 0.4498651623725891, 'learning_rate': 7.371816094597743e-06, 'epoch': 1.23}

 41%|████      | 5439/13317 [56:02<1:20:49,  1.62it/s]
 41%|████      | 5440/13317 [56:02<1:20:53,  1.62it/s]
                                                      
{'loss': 0.1931, 'grad_norm': 0.5200582146644592, 'learning_rate': 7.3706622226334044e-06, 'epoch': 1.23}

 41%|████      | 5440/13317 [56:03<1:20:53,  1.62it/s]
 41%|████      | 5441/13317 [56:03<1:20:46,  1.63it/s]
                                                      
{'loss': 0.1755, 'grad_norm': 0.43385088443756104, 'learning_rate': 7.369508187779471e-06, 'epoch': 1.23}

 41%|████      | 5441/13317 [56:03<1:20:46,  1.63it/s]
 41%|████      | 5442/13317 [56:04<1:20:45,  1.63it/s]
                                                      
{'loss': 0.1985, 'grad_norm': 0.5250884294509888, 'learning_rate': 7.368353990115237e-06, 'epoch': 1.23}

 41%|████      | 5442/13317 [56:04<1:20:45,  1.63it/s]
 41%|████      | 5443/13317 [56:04<1:20:44,  1.63it/s]
                                                      
{'loss': 0.1544, 'grad_norm': 0.42462125420570374, 'learning_rate': 7.367199629720006e-06, 'epoch': 1.23}

 41%|████      | 5443/13317 [56:04<1:20:44,  1.63it/s]
 41%|████      | 5444/13317 [56:05<1:20:39,  1.63it/s]
                                                      
{'loss': 0.2361, 'grad_norm': 0.5963366031646729, 'learning_rate': 7.366045106673097e-06, 'epoch': 1.23}

 41%|████      | 5444/13317 [56:05<1:20:39,  1.63it/s]
 41%|████      | 5445/13317 [56:05<1:20:43,  1.63it/s]
                                                      
{'loss': 0.2037, 'grad_norm': 0.524165689945221, 'learning_rate': 7.364890421053836e-06, 'epoch': 1.23}

 41%|████      | 5445/13317 [56:06<1:20:43,  1.63it/s]
 41%|████      | 5446/13317 [56:06<1:20:37,  1.63it/s]
                                                      
{'loss': 0.1718, 'grad_norm': 0.48167216777801514, 'learning_rate': 7.363735572941564e-06, 'epoch': 1.23}

 41%|████      | 5446/13317 [56:06<1:20:37,  1.63it/s]
 41%|████      | 5447/13317 [56:07<1:20:42,  1.63it/s]
                                                      
{'loss': 0.2055, 'grad_norm': 0.49709656834602356, 'learning_rate': 7.362580562415631e-06, 'epoch': 1.23}

 41%|████      | 5447/13317 [56:07<1:20:42,  1.63it/s]
 41%|████      | 5448/13317 [56:07<1:20:41,  1.63it/s]
                                                      
{'loss': 0.2776, 'grad_norm': 0.7219027280807495, 'learning_rate': 7.361425389555395e-06, 'epoch': 1.23}

 41%|████      | 5448/13317 [56:07<1:20:41,  1.63it/s]
 41%|████      | 5449/13317 [56:08<1:20:35,  1.63it/s]
                                                      
{'loss': 0.1609, 'grad_norm': 0.463863343000412, 'learning_rate': 7.360270054440235e-06, 'epoch': 1.23}

 41%|████      | 5449/13317 [56:08<1:20:35,  1.63it/s]
 41%|████      | 5450/13317 [56:09<1:20:39,  1.63it/s]
                                                      
{'loss': 0.207, 'grad_norm': 0.5155826210975647, 'learning_rate': 7.359114557149529e-06, 'epoch': 1.23}

 41%|████      | 5450/13317 [56:09<1:20:39,  1.63it/s]
 41%|████      | 5451/13317 [56:09<1:20:34,  1.63it/s]
                                                      
{'loss': 0.2278, 'grad_norm': 0.554058849811554, 'learning_rate': 7.357958897762676e-06, 'epoch': 1.23}

 41%|████      | 5451/13317 [56:09<1:20:34,  1.63it/s]
 41%|████      | 5452/13317 [56:10<1:21:02,  1.62it/s]
                                                      
{'loss': 0.1808, 'grad_norm': 0.46746042370796204, 'learning_rate': 7.3568030763590795e-06, 'epoch': 1.23}

 41%|████      | 5452/13317 [56:10<1:21:02,  1.62it/s]
 41%|████      | 5453/13317 [56:10<1:21:22,  1.61it/s]
                                                      
{'loss': 0.1797, 'grad_norm': 0.42961034178733826, 'learning_rate': 7.355647093018158e-06, 'epoch': 1.23}

 41%|████      | 5453/13317 [56:11<1:21:22,  1.61it/s]
 41%|████      | 5454/13317 [56:11<1:21:10,  1.61it/s]
                                                      
{'loss': 0.1577, 'grad_norm': 0.4278031289577484, 'learning_rate': 7.354490947819338e-06, 'epoch': 1.23}

 41%|████      | 5454/13317 [56:11<1:21:10,  1.61it/s]
 41%|████      | 5455/13317 [56:12<1:21:03,  1.62it/s]
                                                      
{'loss': 0.2445, 'grad_norm': 0.6184197068214417, 'learning_rate': 7.3533346408420616e-06, 'epoch': 1.23}

 41%|████      | 5455/13317 [56:12<1:21:03,  1.62it/s]
 41%|████      | 5456/13317 [56:12<1:20:50,  1.62it/s]
                                                      
{'loss': 0.1838, 'grad_norm': 0.5277581214904785, 'learning_rate': 7.3521781721657784e-06, 'epoch': 1.23}

 41%|████      | 5456/13317 [56:12<1:20:50,  1.62it/s]
 41%|████      | 5457/13317 [56:13<1:20:42,  1.62it/s]
                                                      
{'loss': 0.2361, 'grad_norm': 0.5809345245361328, 'learning_rate': 7.351021541869946e-06, 'epoch': 1.23}

 41%|████      | 5457/13317 [56:13<1:20:42,  1.62it/s]
 41%|████      | 5458/13317 [56:13<1:20:43,  1.62it/s]
                                                      
{'loss': 0.2021, 'grad_norm': 0.573833167552948, 'learning_rate': 7.349864750034044e-06, 'epoch': 1.23}

 41%|████      | 5458/13317 [56:14<1:20:43,  1.62it/s]
 41%|████      | 5459/13317 [56:14<1:20:32,  1.63it/s]
                                                      
{'loss': 0.1947, 'grad_norm': 0.4775516986846924, 'learning_rate': 7.348707796737552e-06, 'epoch': 1.23}

 41%|████      | 5459/13317 [56:14<1:20:32,  1.63it/s]
 41%|████      | 5460/13317 [56:15<1:20:34,  1.63it/s]
                                                      
{'loss': 0.1698, 'grad_norm': 0.45742520689964294, 'learning_rate': 7.347550682059965e-06, 'epoch': 1.23}

 41%|████      | 5460/13317 [56:15<1:20:34,  1.63it/s]
 41%|████      | 5461/13317 [56:15<1:20:29,  1.63it/s]
                                                      
{'loss': 0.2032, 'grad_norm': 0.5766906142234802, 'learning_rate': 7.3463934060807905e-06, 'epoch': 1.23}

 41%|████      | 5461/13317 [56:15<1:20:29,  1.63it/s]
 41%|████      | 5462/13317 [56:16<1:20:32,  1.63it/s]
                                                      
{'loss': 0.182, 'grad_norm': 0.5034492611885071, 'learning_rate': 7.345235968879545e-06, 'epoch': 1.23}

 41%|████      | 5462/13317 [56:16<1:20:32,  1.63it/s]
 41%|████      | 5463/13317 [56:17<1:20:46,  1.62it/s]
                                                      
{'loss': 0.195, 'grad_norm': 0.507806658744812, 'learning_rate': 7.344078370535757e-06, 'epoch': 1.23}

 41%|████      | 5463/13317 [56:17<1:20:46,  1.62it/s]
 41%|████      | 5464/13317 [56:17<1:20:39,  1.62it/s]
                                                      
{'loss': 0.247, 'grad_norm': 0.5764020085334778, 'learning_rate': 7.342920611128965e-06, 'epoch': 1.23}

 41%|████      | 5464/13317 [56:17<1:20:39,  1.62it/s]
 41%|████      | 5465/13317 [56:18<1:20:38,  1.62it/s]
                                                      
{'loss': 0.1725, 'grad_norm': 0.43110644817352295, 'learning_rate': 7.341762690738721e-06, 'epoch': 1.23}

 41%|████      | 5465/13317 [56:18<1:20:38,  1.62it/s]
 41%|████      | 5466/13317 [56:18<1:20:32,  1.62it/s]
                                                      
{'loss': 0.2744, 'grad_norm': 0.6438152194023132, 'learning_rate': 7.340604609444584e-06, 'epoch': 1.23}

 41%|████      | 5466/13317 [56:19<1:20:32,  1.62it/s]
 41%|████      | 5467/13317 [56:19<1:20:26,  1.63it/s]
                                                      
{'loss': 0.2299, 'grad_norm': 0.5722547173500061, 'learning_rate': 7.339446367326126e-06, 'epoch': 1.23}

 41%|████      | 5467/13317 [56:19<1:20:26,  1.63it/s]
 41%|████      | 5468/13317 [56:20<1:20:54,  1.62it/s]
                                                      
{'loss': 0.1596, 'grad_norm': 0.44278305768966675, 'learning_rate': 7.3382879644629345e-06, 'epoch': 1.23}

 41%|████      | 5468/13317 [56:20<1:20:54,  1.62it/s]
 41%|████      | 5469/13317 [56:20<1:20:42,  1.62it/s]
                                                      
{'loss': 0.2318, 'grad_norm': 0.5558099746704102, 'learning_rate': 7.337129400934601e-06, 'epoch': 1.23}

 41%|████      | 5469/13317 [56:20<1:20:42,  1.62it/s]
 41%|████      | 5470/13317 [56:21<1:20:38,  1.62it/s]
                                                      
{'loss': 0.2258, 'grad_norm': 0.6205344200134277, 'learning_rate': 7.335970676820731e-06, 'epoch': 1.23}

 41%|████      | 5470/13317 [56:21<1:20:38,  1.62it/s]
 41%|████      | 5471/13317 [56:21<1:20:30,  1.62it/s]
                                                      
{'loss': 0.1775, 'grad_norm': 0.4886016249656677, 'learning_rate': 7.334811792200944e-06, 'epoch': 1.23}

 41%|████      | 5471/13317 [56:22<1:20:30,  1.62it/s]
 41%|████      | 5472/13317 [56:22<1:20:32,  1.62it/s]
                                                      
{'loss': 0.1931, 'grad_norm': 0.4812786877155304, 'learning_rate': 7.3336527471548634e-06, 'epoch': 1.23}

 41%|████      | 5472/13317 [56:22<1:20:32,  1.62it/s]
 41%|████      | 5473/13317 [56:23<1:20:32,  1.62it/s]
                                                      
{'loss': 0.2543, 'grad_norm': 0.6079235672950745, 'learning_rate': 7.332493541762131e-06, 'epoch': 1.23}

 41%|████      | 5473/13317 [56:23<1:20:32,  1.62it/s]
 41%|████      | 5474/13317 [56:23<1:20:23,  1.63it/s]
                                                      
{'loss': 0.2319, 'grad_norm': 0.4830673038959503, 'learning_rate': 7.331334176102395e-06, 'epoch': 1.23}

 41%|████      | 5474/13317 [56:23<1:20:23,  1.63it/s]
 41%|████      | 5475/13317 [56:24<1:20:23,  1.63it/s]
                                                      
{'loss': 0.2191, 'grad_norm': 0.5482299327850342, 'learning_rate': 7.330174650255316e-06, 'epoch': 1.23}

 41%|████      | 5475/13317 [56:24<1:20:23,  1.63it/s]
 41%|████      | 5476/13317 [56:25<1:20:17,  1.63it/s]
                                                      
{'loss': 0.1994, 'grad_norm': 0.5422577261924744, 'learning_rate': 7.329014964300568e-06, 'epoch': 1.23}

 41%|████      | 5476/13317 [56:25<1:20:17,  1.63it/s]
 41%|████      | 5477/13317 [56:25<1:20:22,  1.63it/s]
                                                      
{'loss': 0.2246, 'grad_norm': 0.5225368142127991, 'learning_rate': 7.327855118317829e-06, 'epoch': 1.23}

 41%|████      | 5477/13317 [56:25<1:20:22,  1.63it/s]
 41%|████      | 5478/13317 [56:26<1:20:17,  1.63it/s]
                                                      
{'loss': 0.2301, 'grad_norm': 0.5768547058105469, 'learning_rate': 7.326695112386799e-06, 'epoch': 1.23}

 41%|████      | 5478/13317 [56:26<1:20:17,  1.63it/s]
 41%|████      | 5479/13317 [56:26<1:20:13,  1.63it/s]
                                                      
{'loss': 0.2307, 'grad_norm': 0.6416845917701721, 'learning_rate': 7.325534946587176e-06, 'epoch': 1.23}

 41%|████      | 5479/13317 [56:27<1:20:13,  1.63it/s]
 41%|████      | 5480/13317 [56:27<1:20:26,  1.62it/s]
                                                      
{'loss': 0.2164, 'grad_norm': 0.5827131867408752, 'learning_rate': 7.324374620998682e-06, 'epoch': 1.23}

 41%|████      | 5480/13317 [56:27<1:20:26,  1.62it/s]
 41%|████      | 5481/13317 [56:28<1:20:20,  1.63it/s]
                                                      
{'loss': 0.207, 'grad_norm': 0.5689083933830261, 'learning_rate': 7.32321413570104e-06, 'epoch': 1.23}

 41%|████      | 5481/13317 [56:28<1:20:20,  1.63it/s]
 41%|████      | 5482/13317 [56:28<1:20:22,  1.62it/s]
                                                      
{'loss': 0.1804, 'grad_norm': 0.49248364567756653, 'learning_rate': 7.322053490773987e-06, 'epoch': 1.23}

 41%|████      | 5482/13317 [56:28<1:20:22,  1.62it/s]
 41%|████      | 5483/13317 [56:29<1:20:18,  1.63it/s]
                                                      
{'loss': 0.2533, 'grad_norm': 0.5297258496284485, 'learning_rate': 7.320892686297274e-06, 'epoch': 1.24}

 41%|████      | 5483/13317 [56:29<1:20:18,  1.63it/s]
 41%|████      | 5484/13317 [56:29<1:20:13,  1.63it/s]
                                                      
{'loss': 0.1927, 'grad_norm': 0.5205669403076172, 'learning_rate': 7.319731722350658e-06, 'epoch': 1.24}

 41%|████      | 5484/13317 [56:30<1:20:13,  1.63it/s]
 41%|████      | 5485/13317 [56:30<1:20:14,  1.63it/s]
                                                      
{'loss': 0.2005, 'grad_norm': 0.5146926641464233, 'learning_rate': 7.318570599013914e-06, 'epoch': 1.24}

 41%|████      | 5485/13317 [56:30<1:20:14,  1.63it/s]
 41%|████      | 5486/13317 [56:31<1:20:12,  1.63it/s]
                                                      
{'loss': 0.179, 'grad_norm': 0.44519785046577454, 'learning_rate': 7.317409316366817e-06, 'epoch': 1.24}

 41%|████      | 5486/13317 [56:31<1:20:12,  1.63it/s]
 41%|████      | 5487/13317 [56:31<1:20:18,  1.62it/s]
                                                      
{'loss': 0.2019, 'grad_norm': 0.538680911064148, 'learning_rate': 7.316247874489165e-06, 'epoch': 1.24}

 41%|████      | 5487/13317 [56:31<1:20:18,  1.62it/s]
 41%|████      | 5488/13317 [56:32<1:20:19,  1.62it/s]
                                                      
{'loss': 0.1959, 'grad_norm': 0.4845554828643799, 'learning_rate': 7.315086273460759e-06, 'epoch': 1.24}

 41%|████      | 5488/13317 [56:32<1:20:19,  1.62it/s]
 41%|████      | 5489/13317 [56:33<1:20:14,  1.63it/s]
                                                      
{'loss': 0.1812, 'grad_norm': 0.47449132800102234, 'learning_rate': 7.313924513361413e-06, 'epoch': 1.24}

 41%|████      | 5489/13317 [56:33<1:20:14,  1.63it/s]
 41%|████      | 5490/13317 [56:33<1:20:15,  1.63it/s]
                                                      
{'loss': 0.2003, 'grad_norm': 0.5559443235397339, 'learning_rate': 7.312762594270952e-06, 'epoch': 1.24}

 41%|████      | 5490/13317 [56:33<1:20:15,  1.63it/s]
 41%|████      | 5491/13317 [56:34<1:20:12,  1.63it/s]
                                                      
{'loss': 0.2274, 'grad_norm': 0.6122848987579346, 'learning_rate': 7.311600516269213e-06, 'epoch': 1.24}

 41%|████      | 5491/13317 [56:34<1:20:12,  1.63it/s]
 41%|████      | 5492/13317 [56:34<1:20:15,  1.63it/s]
                                                      
{'loss': 0.2895, 'grad_norm': 0.6055362820625305, 'learning_rate': 7.310438279436044e-06, 'epoch': 1.24}

 41%|████      | 5492/13317 [56:35<1:20:15,  1.63it/s]
 41%|████      | 5493/13317 [56:35<1:20:18,  1.62it/s]
                                                      
{'loss': 0.1508, 'grad_norm': 0.39373108744621277, 'learning_rate': 7.3092758838513015e-06, 'epoch': 1.24}

 41%|████      | 5493/13317 [56:35<1:20:18,  1.62it/s]
 41%|████▏     | 5494/13317 [56:36<1:20:15,  1.62it/s]
                                                      
{'loss': 0.248, 'grad_norm': 0.5814011693000793, 'learning_rate': 7.308113329594855e-06, 'epoch': 1.24}

 41%|████▏     | 5494/13317 [56:36<1:20:15,  1.62it/s]
 41%|████▏     | 5495/13317 [56:36<1:20:14,  1.62it/s]
                                                      
{'loss': 0.2004, 'grad_norm': 0.5108599662780762, 'learning_rate': 7.306950616746583e-06, 'epoch': 1.24}

 41%|████▏     | 5495/13317 [56:36<1:20:14,  1.62it/s]
 41%|████▏     | 5496/13317 [56:37<1:20:10,  1.63it/s]
                                                      
{'loss': 0.2409, 'grad_norm': 0.5601977705955505, 'learning_rate': 7.305787745386378e-06, 'epoch': 1.24}

 41%|████▏     | 5496/13317 [56:37<1:20:10,  1.63it/s]
 41%|████▏     | 5497/13317 [56:37<1:20:10,  1.63it/s]
                                                      
{'loss': 0.1957, 'grad_norm': 0.4791105389595032, 'learning_rate': 7.30462471559414e-06, 'epoch': 1.24}

 41%|████▏     | 5497/13317 [56:38<1:20:10,  1.63it/s]
 41%|████▏     | 5498/13317 [56:38<1:20:05,  1.63it/s]
                                                      
{'loss': 0.1699, 'grad_norm': 0.5623233318328857, 'learning_rate': 7.303461527449784e-06, 'epoch': 1.24}

 41%|████▏     | 5498/13317 [56:38<1:20:05,  1.63it/s]
 41%|████▏     | 5499/13317 [56:39<1:20:03,  1.63it/s]
                                                      
{'loss': 0.1851, 'grad_norm': 0.5045889019966125, 'learning_rate': 7.30229818103323e-06, 'epoch': 1.24}

 41%|████▏     | 5499/13317 [56:39<1:20:03,  1.63it/s]
 41%|████▏     | 5500/13317 [56:39<1:20:11,  1.62it/s]
                                                      
{'loss': 0.2196, 'grad_norm': 0.5344473123550415, 'learning_rate': 7.3011346764244155e-06, 'epoch': 1.24}

 41%|████▏     | 5500/13317 [56:39<1:20:11,  1.62it/s]

  0%|          | 0/13 [00:00<?, ?it/s]

 15%|█▌        | 2/13 [00:00<00:00, 13.38it/s]

 31%|███       | 4/13 [00:00<00:01,  8.39it/s]

 38%|███▊      | 5/13 [00:00<00:01,  7.79it/s]

 46%|████▌     | 6/13 [00:00<00:00,  7.42it/s]

 54%|█████▍    | 7/13 [00:00<00:00,  7.18it/s]

 62%|██████▏   | 8/13 [00:01<00:00,  7.02it/s]

 69%|██████▉   | 9/13 [00:01<00:00,  6.91it/s]

 77%|███████▋  | 10/13 [00:01<00:00,  6.84it/s]

 85%|████████▍ | 11/13 [00:01<00:00,  6.79it/s]

 92%|█████████▏| 12/13 [00:01<00:00,  6.75it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]
                                                      

                                               
{'eval_loss': 0.5995723009109497, 'eval_runtime': 1.9526, 'eval_samples_per_second': 102.429, 'eval_steps_per_second': 6.658, 'epoch': 1.24}

 41%|████▏     | 5500/13317 [56:41<1:20:11,  1.62it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]

                                               
 41%|████▏     | 5501/13317 [56:42<2:36:41,  1.20s/it]
                                                      
{'loss': 0.208, 'grad_norm': 0.5332439541816711, 'learning_rate': 7.299971013703282e-06, 'epoch': 1.24}

 41%|████▏     | 5501/13317 [56:42<2:36:41,  1.20s/it]
 41%|████▏     | 5502/13317 [56:43<2:13:41,  1.03s/it]
                                                      
{'loss': 0.2155, 'grad_norm': 0.5465503931045532, 'learning_rate': 7.298807192949788e-06, 'epoch': 1.24}

 41%|████▏     | 5502/13317 [56:43<2:13:41,  1.03s/it]
 41%|████▏     | 5503/13317 [56:43<1:57:31,  1.11it/s]
                                                      
{'loss': 0.2988, 'grad_norm': 0.7121220231056213, 'learning_rate': 7.297643214243901e-06, 'epoch': 1.24}

 41%|████▏     | 5503/13317 [56:43<1:57:31,  1.11it/s]
 41%|████▏     | 5504/13317 [56:44<1:46:13,  1.23it/s]
                                                      
{'loss': 0.2239, 'grad_norm': 0.480521023273468, 'learning_rate': 7.296479077665596e-06, 'epoch': 1.24}

 41%|████▏     | 5504/13317 [56:44<1:46:13,  1.23it/s]
 41%|████▏     | 5505/13317 [56:44<1:38:19,  1.32it/s]
                                                      
{'loss': 0.2002, 'grad_norm': 0.5346475839614868, 'learning_rate': 7.295314783294863e-06, 'epoch': 1.24}

 41%|████▏     | 5505/13317 [56:44<1:38:19,  1.32it/s]
 41%|████▏     | 5506/13317 [56:45<1:32:43,  1.40it/s]
                                                      
{'loss': 0.2106, 'grad_norm': 0.5093719959259033, 'learning_rate': 7.2941503312117005e-06, 'epoch': 1.24}

 41%|████▏     | 5506/13317 [56:45<1:32:43,  1.40it/s]
 41%|████▏     | 5507/13317 [56:46<1:29:02,  1.46it/s]
                                                      
{'loss': 0.1853, 'grad_norm': 0.46514225006103516, 'learning_rate': 7.292985721496122e-06, 'epoch': 1.24}

 41%|████▏     | 5507/13317 [56:46<1:29:02,  1.46it/s]
 41%|████▏     | 5508/13317 [56:46<1:26:17,  1.51it/s]
                                                      
{'loss': 0.2468, 'grad_norm': 0.6027478575706482, 'learning_rate': 7.291820954228144e-06, 'epoch': 1.24}

 41%|████▏     | 5508/13317 [56:46<1:26:17,  1.51it/s]
 41%|████▏     | 5509/13317 [56:47<1:24:22,  1.54it/s]
                                                      
{'loss': 0.1665, 'grad_norm': 0.4785039722919464, 'learning_rate': 7.2906560294877995e-06, 'epoch': 1.24}

 41%|████▏     | 5509/13317 [56:47<1:24:22,  1.54it/s]
 41%|████▏     | 5510/13317 [56:47<1:22:57,  1.57it/s]
                                                      
{'loss': 0.2164, 'grad_norm': 0.5094742178916931, 'learning_rate': 7.2894909473551335e-06, 'epoch': 1.24}

 41%|████▏     | 5510/13317 [56:48<1:22:57,  1.57it/s]
 41%|████▏     | 5511/13317 [56:48<1:21:59,  1.59it/s]
                                                      
{'loss': 0.2355, 'grad_norm': 0.6507981419563293, 'learning_rate': 7.2883257079101965e-06, 'epoch': 1.24}

 41%|████▏     | 5511/13317 [56:48<1:21:59,  1.59it/s]
 41%|████▏     | 5512/13317 [56:49<1:21:22,  1.60it/s]
                                                      
{'loss': 0.2106, 'grad_norm': 0.5546128749847412, 'learning_rate': 7.287160311233056e-06, 'epoch': 1.24}

 41%|████▏     | 5512/13317 [56:49<1:21:22,  1.60it/s]
 41%|████▏     | 5513/13317 [56:49<1:20:55,  1.61it/s]
                                                      
{'loss': 0.2, 'grad_norm': 0.5482136011123657, 'learning_rate': 7.285994757403783e-06, 'epoch': 1.24}

 41%|████▏     | 5513/13317 [56:49<1:20:55,  1.61it/s]
 41%|████▏     | 5514/13317 [56:50<1:20:33,  1.61it/s]
                                                      
{'loss': 0.2132, 'grad_norm': 0.5510337352752686, 'learning_rate': 7.284829046502467e-06, 'epoch': 1.24}

 41%|████▏     | 5514/13317 [56:50<1:20:33,  1.61it/s]
 41%|████▏     | 5515/13317 [56:50<1:20:18,  1.62it/s]
                                                      
{'loss': 0.1757, 'grad_norm': 0.5449567437171936, 'learning_rate': 7.283663178609204e-06, 'epoch': 1.24}

 41%|████▏     | 5515/13317 [56:51<1:20:18,  1.62it/s]
 41%|████▏     | 5516/13317 [56:51<1:20:05,  1.62it/s]
                                                      
{'loss': 0.2423, 'grad_norm': 0.5792334079742432, 'learning_rate': 7.2824971538041e-06, 'epoch': 1.24}

 41%|████▏     | 5516/13317 [56:51<1:20:05,  1.62it/s]
 41%|████▏     | 5517/13317 [56:52<1:20:03,  1.62it/s]
                                                      
{'loss': 0.2459, 'grad_norm': 0.6528760194778442, 'learning_rate': 7.281330972167273e-06, 'epoch': 1.24}

 41%|████▏     | 5517/13317 [56:52<1:20:03,  1.62it/s]
 41%|████▏     | 5518/13317 [56:52<1:19:55,  1.63it/s]
                                                      
{'loss': 0.1962, 'grad_norm': 0.48418495059013367, 'learning_rate': 7.2801646337788545e-06, 'epoch': 1.24}

 41%|████▏     | 5518/13317 [56:52<1:19:55,  1.63it/s]
 41%|████▏     | 5519/13317 [56:53<1:19:52,  1.63it/s]
                                                      
{'loss': 0.233, 'grad_norm': 0.591803252696991, 'learning_rate': 7.2789981387189835e-06, 'epoch': 1.24}

 41%|████▏     | 5519/13317 [56:53<1:19:52,  1.63it/s]
 41%|████▏     | 5520/13317 [56:54<1:19:49,  1.63it/s]
                                                      
{'loss': 0.253, 'grad_norm': 0.6051661372184753, 'learning_rate': 7.277831487067809e-06, 'epoch': 1.24}

 41%|████▏     | 5520/13317 [56:54<1:19:49,  1.63it/s]
 41%|████▏     | 5521/13317 [56:54<1:19:50,  1.63it/s]
                                                      
{'loss': 0.1934, 'grad_norm': 0.48544079065322876, 'learning_rate': 7.276664678905493e-06, 'epoch': 1.24}

 41%|████▏     | 5521/13317 [56:54<1:19:50,  1.63it/s]
 41%|████▏     | 5522/13317 [56:55<1:19:59,  1.62it/s]
                                                      
{'loss': 0.2027, 'grad_norm': 0.5134294033050537, 'learning_rate': 7.275497714312209e-06, 'epoch': 1.24}

 41%|████▏     | 5522/13317 [56:55<1:19:59,  1.62it/s]
 41%|████▏     | 5523/13317 [56:55<1:19:52,  1.63it/s]
                                                      
{'loss': 0.2197, 'grad_norm': 0.5456287860870361, 'learning_rate': 7.274330593368138e-06, 'epoch': 1.24}

 41%|████▏     | 5523/13317 [56:56<1:19:52,  1.63it/s]
 41%|████▏     | 5524/13317 [56:56<1:19:48,  1.63it/s]
                                                      
{'loss': 0.1974, 'grad_norm': 0.538367748260498, 'learning_rate': 7.2731633161534735e-06, 'epoch': 1.24}

 41%|████▏     | 5524/13317 [56:56<1:19:48,  1.63it/s]
 41%|████▏     | 5525/13317 [56:57<1:19:44,  1.63it/s]
                                                      
{'loss': 0.1636, 'grad_norm': 0.5142098069190979, 'learning_rate': 7.2719958827484215e-06, 'epoch': 1.24}

 41%|████▏     | 5525/13317 [56:57<1:19:44,  1.63it/s]
 41%|████▏     | 5526/13317 [56:57<1:19:42,  1.63it/s]
                                                      
{'loss': 0.2036, 'grad_norm': 0.4980158507823944, 'learning_rate': 7.270828293233195e-06, 'epoch': 1.24}

 41%|████▏     | 5526/13317 [56:57<1:19:42,  1.63it/s]
 42%|████▏     | 5527/13317 [56:58<1:19:49,  1.63it/s]
                                                      
{'loss': 0.1616, 'grad_norm': 0.3845101594924927, 'learning_rate': 7.269660547688022e-06, 'epoch': 1.25}

 42%|████▏     | 5527/13317 [56:58<1:19:49,  1.63it/s]
 42%|████▏     | 5528/13317 [56:58<1:19:46,  1.63it/s]
                                                      
{'loss': 0.196, 'grad_norm': 0.5514487028121948, 'learning_rate': 7.268492646193137e-06, 'epoch': 1.25}

 42%|████▏     | 5528/13317 [56:59<1:19:46,  1.63it/s]
 42%|████▏     | 5529/13317 [56:59<1:19:43,  1.63it/s]
                                                      
{'loss': 0.1617, 'grad_norm': 0.4322464168071747, 'learning_rate': 7.267324588828788e-06, 'epoch': 1.25}

 42%|████▏     | 5529/13317 [56:59<1:19:43,  1.63it/s]
 42%|████▏     | 5530/13317 [57:00<1:19:50,  1.63it/s]
                                                      
{'loss': 0.2124, 'grad_norm': 0.5677034258842468, 'learning_rate': 7.266156375675234e-06, 'epoch': 1.25}

 42%|████▏     | 5530/13317 [57:00<1:19:50,  1.63it/s]
 42%|████▏     | 5531/13317 [57:00<1:19:45,  1.63it/s]
                                                      
{'loss': 0.2459, 'grad_norm': 0.5953633189201355, 'learning_rate': 7.264988006812741e-06, 'epoch': 1.25}

 42%|████▏     | 5531/13317 [57:00<1:19:45,  1.63it/s]
 42%|████▏     | 5532/13317 [57:01<1:19:49,  1.63it/s]
                                                      
{'loss': 0.1739, 'grad_norm': 0.42353585362434387, 'learning_rate': 7.263819482321591e-06, 'epoch': 1.25}

 42%|████▏     | 5532/13317 [57:01<1:19:49,  1.63it/s]
 42%|████▏     | 5533/13317 [57:02<1:19:46,  1.63it/s]
                                                      
{'loss': 0.3307, 'grad_norm': 0.6663879156112671, 'learning_rate': 7.262650802282072e-06, 'epoch': 1.25}

 42%|████▏     | 5533/13317 [57:02<1:19:46,  1.63it/s]
 42%|████▏     | 5534/13317 [57:02<1:19:41,  1.63it/s]
                                                      
{'loss': 0.1857, 'grad_norm': 0.5142697691917419, 'learning_rate': 7.261481966774487e-06, 'epoch': 1.25}

 42%|████▏     | 5534/13317 [57:02<1:19:41,  1.63it/s]
 42%|████▏     | 5535/13317 [57:03<1:19:47,  1.63it/s]
                                                      
{'loss': 0.1937, 'grad_norm': 0.4344891905784607, 'learning_rate': 7.260312975879145e-06, 'epoch': 1.25}

 42%|████▏     | 5535/13317 [57:03<1:19:47,  1.63it/s]
 42%|████▏     | 5536/13317 [57:03<1:19:43,  1.63it/s]
                                                      
{'loss': 0.1742, 'grad_norm': 0.4680497944355011, 'learning_rate': 7.2591438296763695e-06, 'epoch': 1.25}

 42%|████▏     | 5536/13317 [57:04<1:19:43,  1.63it/s]
 42%|████▏     | 5537/13317 [57:04<1:19:43,  1.63it/s]
                                                      
{'loss': 0.2909, 'grad_norm': 0.7232093214988708, 'learning_rate': 7.257974528246492e-06, 'epoch': 1.25}

 42%|████▏     | 5537/13317 [57:04<1:19:43,  1.63it/s]
 42%|████▏     | 5538/13317 [57:05<1:19:39,  1.63it/s]
                                                      
{'loss': 0.2245, 'grad_norm': 0.5348172187805176, 'learning_rate': 7.256805071669856e-06, 'epoch': 1.25}

 42%|████▏     | 5538/13317 [57:05<1:19:39,  1.63it/s]
 42%|████▏     | 5539/13317 [57:05<1:19:37,  1.63it/s]
                                                      
{'loss': 0.1882, 'grad_norm': 0.5196880102157593, 'learning_rate': 7.255635460026818e-06, 'epoch': 1.25}

 42%|████▏     | 5539/13317 [57:05<1:19:37,  1.63it/s]
 42%|████▏     | 5540/13317 [57:06<1:23:39,  1.55it/s]
                                                      
{'loss': 0.1766, 'grad_norm': 0.4621390700340271, 'learning_rate': 7.2544656933977384e-06, 'epoch': 1.25}

 42%|████▏     | 5540/13317 [57:06<1:23:39,  1.55it/s]
 42%|████▏     | 5541/13317 [57:07<1:22:22,  1.57it/s]
                                                      
{'loss': 0.1894, 'grad_norm': 0.6668553948402405, 'learning_rate': 7.253295771862997e-06, 'epoch': 1.25}

 42%|████▏     | 5541/13317 [57:07<1:22:22,  1.57it/s]
 42%|████▏     | 5542/13317 [57:07<1:21:30,  1.59it/s]
                                                      
{'loss': 0.1906, 'grad_norm': 0.4550929367542267, 'learning_rate': 7.252125695502978e-06, 'epoch': 1.25}

 42%|████▏     | 5542/13317 [57:07<1:21:30,  1.59it/s]
 42%|████▏     | 5543/13317 [57:08<1:20:53,  1.60it/s]
                                                      
{'loss': 0.165, 'grad_norm': 0.4468742311000824, 'learning_rate': 7.250955464398076e-06, 'epoch': 1.25}

 42%|████▏     | 5543/13317 [57:08<1:20:53,  1.60it/s]
 42%|████▏     | 5544/13317 [57:08<1:20:27,  1.61it/s]
                                                      
{'loss': 0.1741, 'grad_norm': 0.42325782775878906, 'learning_rate': 7.249785078628702e-06, 'epoch': 1.25}

 42%|████▏     | 5544/13317 [57:09<1:20:27,  1.61it/s]
 42%|████▏     | 5545/13317 [57:09<1:20:15,  1.61it/s]
                                                      
{'loss': 0.1646, 'grad_norm': 0.4355441927909851, 'learning_rate': 7.24861453827527e-06, 'epoch': 1.25}

 42%|████▏     | 5545/13317 [57:09<1:20:15,  1.61it/s]
 42%|████▏     | 5546/13317 [57:10<1:24:19,  1.54it/s]
                                                      
{'loss': 0.2413, 'grad_norm': 0.5182795524597168, 'learning_rate': 7.2474438434182125e-06, 'epoch': 1.25}

 42%|████▏     | 5546/13317 [57:10<1:24:19,  1.54it/s]
 42%|████▏     | 5547/13317 [57:10<1:22:51,  1.56it/s]
                                                      
{'loss': 0.1921, 'grad_norm': 0.5620766282081604, 'learning_rate': 7.246272994137966e-06, 'epoch': 1.25}

 42%|████▏     | 5547/13317 [57:10<1:22:51,  1.56it/s]
 42%|████▏     | 5548/13317 [57:11<1:21:48,  1.58it/s]
                                                      
{'loss': 0.2003, 'grad_norm': 0.5469702482223511, 'learning_rate': 7.24510199051498e-06, 'epoch': 1.25}

 42%|████▏     | 5548/13317 [57:11<1:21:48,  1.58it/s]
 42%|████▏     | 5549/13317 [57:12<1:21:06,  1.60it/s]
                                                      
{'loss': 0.2278, 'grad_norm': 0.5595508813858032, 'learning_rate': 7.243930832629717e-06, 'epoch': 1.25}

 42%|████▏     | 5549/13317 [57:12<1:21:06,  1.60it/s]
 42%|████▏     | 5550/13317 [57:12<1:20:33,  1.61it/s]
                                                      
{'loss': 0.1694, 'grad_norm': 0.4346829652786255, 'learning_rate': 7.242759520562646e-06, 'epoch': 1.25}

 42%|████▏     | 5550/13317 [57:12<1:20:33,  1.61it/s]
 42%|████▏     | 5551/13317 [57:13<1:20:18,  1.61it/s]
                                                      
{'loss': 0.1709, 'grad_norm': 0.48799076676368713, 'learning_rate': 7.2415880543942484e-06, 'epoch': 1.25}

 42%|████▏     | 5551/13317 [57:13<1:20:18,  1.61it/s]
 42%|████▏     | 5552/13317 [57:14<1:24:02,  1.54it/s]
                                                      
{'loss': 0.1979, 'grad_norm': 0.525801420211792, 'learning_rate': 7.2404164342050195e-06, 'epoch': 1.25}

 42%|████▏     | 5552/13317 [57:14<1:24:02,  1.54it/s]
 42%|████▏     | 5553/13317 [57:14<1:22:48,  1.56it/s]
                                                      
{'loss': 0.2002, 'grad_norm': 0.5620238780975342, 'learning_rate': 7.239244660075457e-06, 'epoch': 1.25}

 42%|████▏     | 5553/13317 [57:14<1:22:48,  1.56it/s]
 42%|████▏     | 5554/13317 [57:15<1:21:47,  1.58it/s]
                                                      
{'loss': 0.1818, 'grad_norm': 0.4674803614616394, 'learning_rate': 7.238072732086079e-06, 'epoch': 1.25}

 42%|████▏     | 5554/13317 [57:15<1:21:47,  1.58it/s]
 42%|████▏     | 5555/13317 [57:15<1:21:04,  1.60it/s]
                                                      
{'loss': 0.1912, 'grad_norm': 0.4388124346733093, 'learning_rate': 7.236900650317406e-06, 'epoch': 1.25}

 42%|████▏     | 5555/13317 [57:16<1:21:04,  1.60it/s]
 42%|████▏     | 5556/13317 [57:16<1:20:34,  1.61it/s]
                                                      
{'loss': 0.2759, 'grad_norm': 0.5797873735427856, 'learning_rate': 7.235728414849974e-06, 'epoch': 1.25}

 42%|████▏     | 5556/13317 [57:16<1:20:34,  1.61it/s]
 42%|████▏     | 5557/13317 [57:17<1:20:12,  1.61it/s]
                                                      
{'loss': 0.2249, 'grad_norm': 0.5683408379554749, 'learning_rate': 7.234556025764326e-06, 'epoch': 1.25}

 42%|████▏     | 5557/13317 [57:17<1:20:12,  1.61it/s]
 42%|████▏     | 5558/13317 [57:17<1:20:04,  1.61it/s]
                                                      
{'loss': 0.1831, 'grad_norm': 0.48294803500175476, 'learning_rate': 7.233383483141021e-06, 'epoch': 1.25}

 42%|████▏     | 5558/13317 [57:17<1:20:04,  1.61it/s]
 42%|████▏     | 5559/13317 [57:18<1:19:50,  1.62it/s]
                                                      
{'loss': 0.1776, 'grad_norm': 0.46612611413002014, 'learning_rate': 7.232210787060624e-06, 'epoch': 1.25}

 42%|████▏     | 5559/13317 [57:18<1:19:50,  1.62it/s]
 42%|████▏     | 5560/13317 [57:18<1:19:37,  1.62it/s]
                                                      
{'loss': 0.241, 'grad_norm': 0.46475762128829956, 'learning_rate': 7.231037937603709e-06, 'epoch': 1.25}

 42%|████▏     | 5560/13317 [57:19<1:19:37,  1.62it/s]
 42%|████▏     | 5561/13317 [57:19<1:19:31,  1.63it/s]
                                                      
{'loss': 0.2578, 'grad_norm': 0.6640816926956177, 'learning_rate': 7.229864934850867e-06, 'epoch': 1.25}

 42%|████▏     | 5561/13317 [57:19<1:19:31,  1.63it/s]
 42%|████▏     | 5562/13317 [57:20<1:19:39,  1.62it/s]
                                                      
{'loss': 0.1911, 'grad_norm': 0.5249108672142029, 'learning_rate': 7.2286917788826926e-06, 'epoch': 1.25}

 42%|████▏     | 5562/13317 [57:20<1:19:39,  1.62it/s]
 42%|████▏     | 5563/13317 [57:20<1:19:39,  1.62it/s]
                                                      
{'loss': 0.2181, 'grad_norm': 0.5447158813476562, 'learning_rate': 7.227518469779795e-06, 'epoch': 1.25}

 42%|████▏     | 5563/13317 [57:20<1:19:39,  1.62it/s]
 42%|████▏     | 5564/13317 [57:21<1:19:33,  1.62it/s]
                                                      
{'loss': 0.2066, 'grad_norm': 0.564319908618927, 'learning_rate': 7.226345007622793e-06, 'epoch': 1.25}

 42%|████▏     | 5564/13317 [57:21<1:19:33,  1.62it/s]
 42%|████▏     | 5565/13317 [57:22<1:19:26,  1.63it/s]
                                                      
{'loss': 0.2003, 'grad_norm': 0.5384643077850342, 'learning_rate': 7.225171392492316e-06, 'epoch': 1.25}

 42%|████▏     | 5565/13317 [57:22<1:19:26,  1.63it/s]
 42%|████▏     | 5566/13317 [57:22<1:19:24,  1.63it/s]
                                                      
{'loss': 0.2139, 'grad_norm': 0.5467650890350342, 'learning_rate': 7.223997624469006e-06, 'epoch': 1.25}

 42%|████▏     | 5566/13317 [57:22<1:19:24,  1.63it/s]
 42%|████▏     | 5567/13317 [57:23<1:19:20,  1.63it/s]
                                                      
{'loss': 0.1779, 'grad_norm': 0.4762992560863495, 'learning_rate': 7.2228237036335085e-06, 'epoch': 1.25}

 42%|████▏     | 5567/13317 [57:23<1:19:20,  1.63it/s]
 42%|████▏     | 5568/13317 [57:23<1:19:26,  1.63it/s]
                                                      
{'loss': 0.1596, 'grad_norm': 0.45289310812950134, 'learning_rate': 7.221649630066488e-06, 'epoch': 1.25}

 42%|████▏     | 5568/13317 [57:24<1:19:26,  1.63it/s]
 42%|████▏     | 5569/13317 [57:24<1:19:25,  1.63it/s]
                                                      
{'loss': 0.2281, 'grad_norm': 0.498089075088501, 'learning_rate': 7.220475403848615e-06, 'epoch': 1.25}

 42%|████▏     | 5569/13317 [57:24<1:19:25,  1.63it/s]
 42%|████▏     | 5570/13317 [57:25<1:19:20,  1.63it/s]
                                                      
{'loss': 0.2754, 'grad_norm': 0.5271407961845398, 'learning_rate': 7.219301025060571e-06, 'epoch': 1.25}

 42%|████▏     | 5570/13317 [57:25<1:19:20,  1.63it/s]
 42%|████▏     | 5571/13317 [57:25<1:19:15,  1.63it/s]
                                                      
{'loss': 0.2377, 'grad_norm': 0.5635782480239868, 'learning_rate': 7.218126493783048e-06, 'epoch': 1.26}

 42%|████▏     | 5571/13317 [57:25<1:19:15,  1.63it/s]
 42%|████▏     | 5572/13317 [57:26<1:19:14,  1.63it/s]
                                                      
{'loss': 0.1692, 'grad_norm': 0.43977975845336914, 'learning_rate': 7.216951810096748e-06, 'epoch': 1.26}

 42%|████▏     | 5572/13317 [57:26<1:19:14,  1.63it/s]
 42%|████▏     | 5573/13317 [57:26<1:19:20,  1.63it/s]
                                                      
{'loss': 0.21, 'grad_norm': 0.5015772581100464, 'learning_rate': 7.215776974082387e-06, 'epoch': 1.26}

 42%|████▏     | 5573/13317 [57:27<1:19:20,  1.63it/s]
 42%|████▏     | 5574/13317 [57:27<1:19:18,  1.63it/s]
                                                      
{'loss': 0.2466, 'grad_norm': 0.573887050151825, 'learning_rate': 7.214601985820685e-06, 'epoch': 1.26}

 42%|████▏     | 5574/13317 [57:27<1:19:18,  1.63it/s]
 42%|████▏     | 5575/13317 [57:28<1:19:17,  1.63it/s]
                                                      
{'loss': 0.1739, 'grad_norm': 0.4588243067264557, 'learning_rate': 7.21342684539238e-06, 'epoch': 1.26}

 42%|████▏     | 5575/13317 [57:28<1:19:17,  1.63it/s]
 42%|████▏     | 5576/13317 [57:28<1:19:14,  1.63it/s]
                                                      
{'loss': 0.1799, 'grad_norm': 0.4997675120830536, 'learning_rate': 7.212251552878213e-06, 'epoch': 1.26}

 42%|████▏     | 5576/13317 [57:28<1:19:14,  1.63it/s]
 42%|████▏     | 5577/13317 [57:29<1:19:10,  1.63it/s]
                                                      
{'loss': 0.176, 'grad_norm': 0.5211914777755737, 'learning_rate': 7.211076108358941e-06, 'epoch': 1.26}

 42%|████▏     | 5577/13317 [57:29<1:19:10,  1.63it/s]
 42%|████▏     | 5578/13317 [57:30<1:19:36,  1.62it/s]
                                                      
{'loss': 0.1864, 'grad_norm': 0.49628573656082153, 'learning_rate': 7.209900511915328e-06, 'epoch': 1.26}

 42%|████▏     | 5578/13317 [57:30<1:19:36,  1.62it/s]
 42%|████▏     | 5579/13317 [57:30<1:19:27,  1.62it/s]
                                                      
{'loss': 0.2689, 'grad_norm': 0.6715751886367798, 'learning_rate': 7.208724763628153e-06, 'epoch': 1.26}

 42%|████▏     | 5579/13317 [57:30<1:19:27,  1.62it/s]
 42%|████▏     | 5580/13317 [57:31<1:19:21,  1.62it/s]
                                                      
{'loss': 0.2071, 'grad_norm': 0.5575264096260071, 'learning_rate': 7.2075488635782e-06, 'epoch': 1.26}

 42%|████▏     | 5580/13317 [57:31<1:19:21,  1.62it/s]
 42%|████▏     | 5581/13317 [57:31<1:19:15,  1.63it/s]
                                                      
{'loss': 0.1848, 'grad_norm': 0.49134936928749084, 'learning_rate': 7.206372811846267e-06, 'epoch': 1.26}

 42%|████▏     | 5581/13317 [57:31<1:19:15,  1.63it/s]
 42%|████▏     | 5582/13317 [57:32<1:19:14,  1.63it/s]
                                                      
{'loss': 0.2029, 'grad_norm': 0.5254290103912354, 'learning_rate': 7.2051966085131584e-06, 'epoch': 1.26}

 42%|████▏     | 5582/13317 [57:32<1:19:14,  1.63it/s]
 42%|████▏     | 5583/13317 [57:33<1:19:36,  1.62it/s]
                                                      
{'loss': 0.1958, 'grad_norm': 0.5080187320709229, 'learning_rate': 7.204020253659695e-06, 'epoch': 1.26}

 42%|████▏     | 5583/13317 [57:33<1:19:36,  1.62it/s]
 42%|████▏     | 5584/13317 [57:33<1:19:29,  1.62it/s]
                                                      
{'loss': 0.1828, 'grad_norm': 0.49849313497543335, 'learning_rate': 7.202843747366704e-06, 'epoch': 1.26}

 42%|████▏     | 5584/13317 [57:33<1:19:29,  1.62it/s]
 42%|████▏     | 5585/13317 [57:34<1:19:22,  1.62it/s]
                                                      
{'loss': 0.2005, 'grad_norm': 0.4881468117237091, 'learning_rate': 7.201667089715021e-06, 'epoch': 1.26}

 42%|████▏     | 5585/13317 [57:34<1:19:22,  1.62it/s]
 42%|████▏     | 5586/13317 [57:34<1:19:14,  1.63it/s]
                                                      
{'loss': 0.2029, 'grad_norm': 0.4706321358680725, 'learning_rate': 7.2004902807855005e-06, 'epoch': 1.26}

 42%|████▏     | 5586/13317 [57:35<1:19:14,  1.63it/s]
 42%|████▏     | 5587/13317 [57:35<1:19:10,  1.63it/s]
                                                      
{'loss': 0.2047, 'grad_norm': 0.541584849357605, 'learning_rate': 7.199313320658996e-06, 'epoch': 1.26}

 42%|████▏     | 5587/13317 [57:35<1:19:10,  1.63it/s]
 42%|████▏     | 5588/13317 [57:36<1:19:12,  1.63it/s]
                                                      
{'loss': 0.192, 'grad_norm': 0.49466368556022644, 'learning_rate': 7.198136209416382e-06, 'epoch': 1.26}

 42%|████▏     | 5588/13317 [57:36<1:19:12,  1.63it/s]
 42%|████▏     | 5589/13317 [57:36<1:19:31,  1.62it/s]
                                                      
{'loss': 0.2088, 'grad_norm': 0.5101885795593262, 'learning_rate': 7.196958947138535e-06, 'epoch': 1.26}

 42%|████▏     | 5589/13317 [57:36<1:19:31,  1.62it/s]
 42%|████▏     | 5590/13317 [57:37<1:19:18,  1.62it/s]
                                                      
{'loss': 0.26, 'grad_norm': 0.6368480324745178, 'learning_rate': 7.195781533906345e-06, 'epoch': 1.26}

 42%|████▏     | 5590/13317 [57:37<1:19:18,  1.62it/s]
 42%|████▏     | 5591/13317 [57:38<1:19:14,  1.63it/s]
                                                      
{'loss': 0.2449, 'grad_norm': 0.6507318615913391, 'learning_rate': 7.194603969800716e-06, 'epoch': 1.26}

 42%|████▏     | 5591/13317 [57:38<1:19:14,  1.63it/s]
 42%|████▏     | 5592/13317 [57:38<1:19:09,  1.63it/s]
                                                      
{'loss': 0.2126, 'grad_norm': 0.546613872051239, 'learning_rate': 7.193426254902558e-06, 'epoch': 1.26}

 42%|████▏     | 5592/13317 [57:38<1:19:09,  1.63it/s]
 42%|████▏     | 5593/13317 [57:39<1:19:12,  1.63it/s]
                                                      
{'loss': 0.2081, 'grad_norm': 0.5180566310882568, 'learning_rate': 7.192248389292791e-06, 'epoch': 1.26}

 42%|████▏     | 5593/13317 [57:39<1:19:12,  1.63it/s]
 42%|████▏     | 5594/13317 [57:39<1:19:14,  1.62it/s]
                                                      
{'loss': 0.2619, 'grad_norm': 0.6700891256332397, 'learning_rate': 7.191070373052347e-06, 'epoch': 1.26}

 42%|████▏     | 5594/13317 [57:40<1:19:14,  1.62it/s]
 42%|████▏     | 5595/13317 [57:40<1:19:11,  1.63it/s]
                                                      
{'loss': 0.2404, 'grad_norm': 0.5427171587944031, 'learning_rate': 7.18989220626217e-06, 'epoch': 1.26}

 42%|████▏     | 5595/13317 [57:40<1:19:11,  1.63it/s]
 42%|████▏     | 5596/13317 [57:41<1:19:05,  1.63it/s]
                                                      
{'loss': 0.186, 'grad_norm': 0.527807354927063, 'learning_rate': 7.188713889003211e-06, 'epoch': 1.26}

 42%|████▏     | 5596/13317 [57:41<1:19:05,  1.63it/s]
 42%|████▏     | 5597/13317 [57:41<1:19:04,  1.63it/s]
                                                      
{'loss': 0.1696, 'grad_norm': 0.47606632113456726, 'learning_rate': 7.187535421356434e-06, 'epoch': 1.26}

 42%|████▏     | 5597/13317 [57:41<1:19:04,  1.63it/s]
 42%|████▏     | 5598/13317 [57:42<1:19:09,  1.63it/s]
                                                      
{'loss': 0.2176, 'grad_norm': 0.6119983792304993, 'learning_rate': 7.186356803402812e-06, 'epoch': 1.26}

 42%|████▏     | 5598/13317 [57:42<1:19:09,  1.63it/s]
 42%|████▏     | 5599/13317 [57:42<1:19:13,  1.62it/s]
                                                      
{'loss': 0.201, 'grad_norm': 0.5082367658615112, 'learning_rate': 7.185178035223327e-06, 'epoch': 1.26}

 42%|████▏     | 5599/13317 [57:43<1:19:13,  1.62it/s]
 42%|████▏     | 5600/13317 [57:43<1:19:06,  1.63it/s]
                                                      
{'loss': 0.1741, 'grad_norm': 0.49046337604522705, 'learning_rate': 7.183999116898975e-06, 'epoch': 1.26}

 42%|████▏     | 5600/13317 [57:43<1:19:06,  1.63it/s]
 42%|████▏     | 5601/13317 [57:44<1:19:05,  1.63it/s]
                                                      
{'loss': 0.1865, 'grad_norm': 0.5547100305557251, 'learning_rate': 7.182820048510757e-06, 'epoch': 1.26}

 42%|████▏     | 5601/13317 [57:44<1:19:05,  1.63it/s]
 42%|████▏     | 5602/13317 [57:44<1:18:57,  1.63it/s]
                                                      
{'loss': 0.2117, 'grad_norm': 0.5340319275856018, 'learning_rate': 7.181640830139692e-06, 'epoch': 1.26}

 42%|████▏     | 5602/13317 [57:44<1:18:57,  1.63it/s]
 42%|████▏     | 5603/13317 [57:45<1:19:20,  1.62it/s]
                                                      
{'loss': 0.2212, 'grad_norm': 0.7250585556030273, 'learning_rate': 7.180461461866801e-06, 'epoch': 1.26}

 42%|████▏     | 5603/13317 [57:45<1:19:20,  1.62it/s]
 42%|████▏     | 5604/13317 [57:46<1:19:17,  1.62it/s]
                                                      
{'loss': 0.2369, 'grad_norm': 0.6020092368125916, 'learning_rate': 7.179281943773121e-06, 'epoch': 1.26}

 42%|████▏     | 5604/13317 [57:46<1:19:17,  1.62it/s]
 42%|████▏     | 5605/13317 [57:46<1:19:10,  1.62it/s]
                                                      
{'loss': 0.1552, 'grad_norm': 0.4499562084674835, 'learning_rate': 7.1781022759396965e-06, 'epoch': 1.26}

 42%|████▏     | 5605/13317 [57:46<1:19:10,  1.62it/s]
 42%|████▏     | 5606/13317 [57:47<1:19:07,  1.62it/s]
                                                      
{'loss': 0.2001, 'grad_norm': 0.5838169455528259, 'learning_rate': 7.176922458447584e-06, 'epoch': 1.26}

 42%|████▏     | 5606/13317 [57:47<1:19:07,  1.62it/s]
 42%|████▏     | 5607/13317 [57:47<1:19:03,  1.63it/s]
                                                      
{'loss': 0.1915, 'grad_norm': 0.5493305921554565, 'learning_rate': 7.1757424913778485e-06, 'epoch': 1.26}

 42%|████▏     | 5607/13317 [57:48<1:19:03,  1.63it/s]
 42%|████▏     | 5608/13317 [57:48<1:19:03,  1.63it/s]
                                                      
{'loss': 0.1749, 'grad_norm': 0.44110724329948425, 'learning_rate': 7.174562374811567e-06, 'epoch': 1.26}

 42%|████▏     | 5608/13317 [57:48<1:19:03,  1.63it/s]
 42%|████▏     | 5609/13317 [57:49<1:19:00,  1.63it/s]
                                                      
{'loss': 0.1758, 'grad_norm': 0.5001733899116516, 'learning_rate': 7.173382108829826e-06, 'epoch': 1.26}

 42%|████▏     | 5609/13317 [57:49<1:19:00,  1.63it/s]
 42%|████▏     | 5610/13317 [57:49<1:18:56,  1.63it/s]
                                                      
{'loss': 0.2534, 'grad_norm': 0.6076960563659668, 'learning_rate': 7.172201693513721e-06, 'epoch': 1.26}

 42%|████▏     | 5610/13317 [57:49<1:18:56,  1.63it/s]
 42%|████▏     | 5611/13317 [57:50<1:18:51,  1.63it/s]
                                                      
{'loss': 0.2547, 'grad_norm': 0.5921885967254639, 'learning_rate': 7.171021128944361e-06, 'epoch': 1.26}

 42%|████▏     | 5611/13317 [57:50<1:18:51,  1.63it/s]
 42%|████▏     | 5612/13317 [57:50<1:18:50,  1.63it/s]
                                                      
{'loss': 0.1854, 'grad_norm': 0.5158466100692749, 'learning_rate': 7.169840415202862e-06, 'epoch': 1.26}

 42%|████▏     | 5612/13317 [57:51<1:18:50,  1.63it/s]
 42%|████▏     | 5613/13317 [57:51<1:18:56,  1.63it/s]
                                                      
{'loss': 0.2553, 'grad_norm': 0.5667637586593628, 'learning_rate': 7.168659552370352e-06, 'epoch': 1.26}

 42%|████▏     | 5613/13317 [57:51<1:18:56,  1.63it/s]
 42%|████▏     | 5614/13317 [57:52<1:18:52,  1.63it/s]
                                                      
{'loss': 0.1684, 'grad_norm': 0.4963270425796509, 'learning_rate': 7.1674785405279676e-06, 'epoch': 1.26}

 42%|████▏     | 5614/13317 [57:52<1:18:52,  1.63it/s]
 42%|████▏     | 5615/13317 [57:52<1:18:49,  1.63it/s]
                                                      
{'loss': 0.2403, 'grad_norm': 0.5563718676567078, 'learning_rate': 7.166297379756859e-06, 'epoch': 1.26}

 42%|████▏     | 5615/13317 [57:52<1:18:49,  1.63it/s]
 42%|████▏     | 5616/13317 [57:53<1:18:46,  1.63it/s]
                                                      
{'loss': 0.167, 'grad_norm': 0.49718523025512695, 'learning_rate': 7.165116070138183e-06, 'epoch': 1.27}

 42%|████▏     | 5616/13317 [57:53<1:18:46,  1.63it/s]
 42%|████▏     | 5617/13317 [57:54<1:18:46,  1.63it/s]
                                                      
{'loss': 0.2256, 'grad_norm': 0.5302925705909729, 'learning_rate': 7.1639346117531084e-06, 'epoch': 1.27}

 42%|████▏     | 5617/13317 [57:54<1:18:46,  1.63it/s]
 42%|████▏     | 5618/13317 [57:54<1:19:11,  1.62it/s]
                                                      
{'loss': 0.2201, 'grad_norm': 0.8408058285713196, 'learning_rate': 7.162753004682813e-06, 'epoch': 1.27}

 42%|████▏     | 5618/13317 [57:54<1:19:11,  1.62it/s]
 42%|████▏     | 5619/13317 [57:55<1:19:03,  1.62it/s]
                                                      
{'loss': 0.2828, 'grad_norm': 0.550265371799469, 'learning_rate': 7.161571249008488e-06, 'epoch': 1.27}

 42%|████▏     | 5619/13317 [57:55<1:19:03,  1.62it/s]
 42%|████▏     | 5620/13317 [57:55<1:18:56,  1.62it/s]
                                                      
{'loss': 0.1895, 'grad_norm': 0.42106884717941284, 'learning_rate': 7.160389344811332e-06, 'epoch': 1.27}

 42%|████▏     | 5620/13317 [57:55<1:18:56,  1.62it/s]
 42%|████▏     | 5621/13317 [57:56<1:18:51,  1.63it/s]
                                                      
{'loss': 0.1889, 'grad_norm': 0.501895010471344, 'learning_rate': 7.15920729217255e-06, 'epoch': 1.27}

 42%|████▏     | 5621/13317 [57:56<1:18:51,  1.63it/s]
 42%|████▏     | 5622/13317 [57:57<1:18:50,  1.63it/s]
                                                      
{'loss': 0.1804, 'grad_norm': 0.49560847878456116, 'learning_rate': 7.158025091173368e-06, 'epoch': 1.27}

 42%|████▏     | 5622/13317 [57:57<1:18:50,  1.63it/s]
 42%|████▏     | 5623/13317 [57:57<1:18:55,  1.62it/s]
                                                      
{'loss': 0.1648, 'grad_norm': 0.4685954749584198, 'learning_rate': 7.1568427418950095e-06, 'epoch': 1.27}

 42%|████▏     | 5623/13317 [57:57<1:18:55,  1.62it/s]
 42%|████▏     | 5624/13317 [57:58<1:18:51,  1.63it/s]
                                                      
{'loss': 0.1658, 'grad_norm': 0.4403690993785858, 'learning_rate': 7.155660244418721e-06, 'epoch': 1.27}

 42%|████▏     | 5624/13317 [57:58<1:18:51,  1.63it/s]
 42%|████▏     | 5625/13317 [57:58<1:18:43,  1.63it/s]
                                                      
{'loss': 0.1973, 'grad_norm': 0.5378357768058777, 'learning_rate': 7.154477598825747e-06, 'epoch': 1.27}

 42%|████▏     | 5625/13317 [57:59<1:18:43,  1.63it/s]
 42%|████▏     | 5626/13317 [57:59<1:18:43,  1.63it/s]
                                                      
{'loss': 0.1937, 'grad_norm': 0.4861106276512146, 'learning_rate': 7.1532948051973505e-06, 'epoch': 1.27}

 42%|████▏     | 5626/13317 [57:59<1:18:43,  1.63it/s]
 42%|████▏     | 5627/13317 [58:00<1:18:40,  1.63it/s]
                                                      
{'loss': 0.1754, 'grad_norm': 0.4813997447490692, 'learning_rate': 7.152111863614804e-06, 'epoch': 1.27}

 42%|████▏     | 5627/13317 [58:00<1:18:40,  1.63it/s]
 42%|████▏     | 5628/13317 [58:00<1:18:45,  1.63it/s]
                                                      
{'loss': 0.222, 'grad_norm': 0.5627270936965942, 'learning_rate': 7.150928774159381e-06, 'epoch': 1.27}

 42%|████▏     | 5628/13317 [58:00<1:18:45,  1.63it/s]
 42%|████▏     | 5629/13317 [58:01<1:18:41,  1.63it/s]
                                                      
{'loss': 0.2045, 'grad_norm': 0.5346542000770569, 'learning_rate': 7.14974553691238e-06, 'epoch': 1.27}

 42%|████▏     | 5629/13317 [58:01<1:18:41,  1.63it/s]
 42%|████▏     | 5630/13317 [58:02<1:18:37,  1.63it/s]
                                                      
{'loss': 0.2162, 'grad_norm': 0.5150117874145508, 'learning_rate': 7.148562151955098e-06, 'epoch': 1.27}

 42%|████▏     | 5630/13317 [58:02<1:18:37,  1.63it/s]
 42%|████▏     | 5631/13317 [58:02<1:18:36,  1.63it/s]
                                                      
{'loss': 0.1678, 'grad_norm': 0.4764023721218109, 'learning_rate': 7.1473786193688464e-06, 'epoch': 1.27}

 42%|████▏     | 5631/13317 [58:02<1:18:36,  1.63it/s]
 42%|████▏     | 5632/13317 [58:03<1:18:36,  1.63it/s]
                                                      
{'loss': 0.1931, 'grad_norm': 0.49726513028144836, 'learning_rate': 7.146194939234947e-06, 'epoch': 1.27}

 42%|████▏     | 5632/13317 [58:03<1:18:36,  1.63it/s]
 42%|████▏     | 5633/13317 [58:03<1:18:39,  1.63it/s]
                                                      
{'loss': 0.1767, 'grad_norm': 0.418653279542923, 'learning_rate': 7.145011111634732e-06, 'epoch': 1.27}

 42%|████▏     | 5633/13317 [58:03<1:18:39,  1.63it/s]
 42%|████▏     | 5634/13317 [58:04<1:18:37,  1.63it/s]
                                                      
{'loss': 0.2376, 'grad_norm': 0.5954270958900452, 'learning_rate': 7.143827136649541e-06, 'epoch': 1.27}

 42%|████▏     | 5634/13317 [58:04<1:18:37,  1.63it/s]
 42%|████▏     | 5635/13317 [58:05<1:18:35,  1.63it/s]
                                                      
{'loss': 0.1619, 'grad_norm': 0.4264603853225708, 'learning_rate': 7.142643014360726e-06, 'epoch': 1.27}

 42%|████▏     | 5635/13317 [58:05<1:18:35,  1.63it/s]
 42%|████▏     | 5636/13317 [58:05<1:18:32,  1.63it/s]
                                                      
{'loss': 0.1632, 'grad_norm': 0.4395492970943451, 'learning_rate': 7.141458744849651e-06, 'epoch': 1.27}

 42%|████▏     | 5636/13317 [58:05<1:18:32,  1.63it/s]
 42%|████▏     | 5637/13317 [58:06<1:18:32,  1.63it/s]
                                                      
{'loss': 0.2039, 'grad_norm': 0.5320619940757751, 'learning_rate': 7.140274328197684e-06, 'epoch': 1.27}

 42%|████▏     | 5637/13317 [58:06<1:18:32,  1.63it/s]
 42%|████▏     | 5638/13317 [58:06<1:18:38,  1.63it/s]
                                                      
{'loss': 0.173, 'grad_norm': 0.44280675053596497, 'learning_rate': 7.13908976448621e-06, 'epoch': 1.27}

 42%|████▏     | 5638/13317 [58:07<1:18:38,  1.63it/s]
 42%|████▏     | 5639/13317 [58:07<1:18:33,  1.63it/s]
                                                      
{'loss': 0.1763, 'grad_norm': 0.5045804977416992, 'learning_rate': 7.1379050537966196e-06, 'epoch': 1.27}

 42%|████▏     | 5639/13317 [58:07<1:18:33,  1.63it/s]
 42%|████▏     | 5640/13317 [58:08<1:18:32,  1.63it/s]
                                                      
{'loss': 0.1977, 'grad_norm': 0.46197688579559326, 'learning_rate': 7.136720196210316e-06, 'epoch': 1.27}

 42%|████▏     | 5640/13317 [58:08<1:18:32,  1.63it/s]
 42%|████▏     | 5641/13317 [58:08<1:18:33,  1.63it/s]
                                                      
{'loss': 0.2433, 'grad_norm': 0.5534989833831787, 'learning_rate': 7.1355351918087115e-06, 'epoch': 1.27}

 42%|████▏     | 5641/13317 [58:08<1:18:33,  1.63it/s]
 42%|████▏     | 5642/13317 [58:09<1:18:32,  1.63it/s]
                                                      
{'loss': 0.2633, 'grad_norm': 0.623134970664978, 'learning_rate': 7.1343500406732284e-06, 'epoch': 1.27}

 42%|████▏     | 5642/13317 [58:09<1:18:32,  1.63it/s]
 42%|████▏     | 5643/13317 [58:09<1:18:31,  1.63it/s]
                                                      
{'loss': 0.1966, 'grad_norm': 0.4857393801212311, 'learning_rate': 7.133164742885296e-06, 'epoch': 1.27}

 42%|████▏     | 5643/13317 [58:10<1:18:31,  1.63it/s]
 42%|████▏     | 5644/13317 [58:10<1:18:31,  1.63it/s]
                                                      
{'loss': 0.1861, 'grad_norm': 0.5013343691825867, 'learning_rate': 7.131979298526361e-06, 'epoch': 1.27}

 42%|████▏     | 5644/13317 [58:10<1:18:31,  1.63it/s]
 42%|████▏     | 5645/13317 [58:11<1:18:28,  1.63it/s]
                                                      
{'loss': 0.1899, 'grad_norm': 0.5367646813392639, 'learning_rate': 7.130793707677875e-06, 'epoch': 1.27}

 42%|████▏     | 5645/13317 [58:11<1:18:28,  1.63it/s]
 42%|████▏     | 5646/13317 [58:11<1:18:26,  1.63it/s]
                                                      
{'loss': 0.2078, 'grad_norm': 0.45386219024658203, 'learning_rate': 7.129607970421299e-06, 'epoch': 1.27}

 42%|████▏     | 5646/13317 [58:11<1:18:26,  1.63it/s]
 42%|████▏     | 5647/13317 [58:12<1:18:25,  1.63it/s]
                                                      
{'loss': 0.2016, 'grad_norm': 0.5351231098175049, 'learning_rate': 7.1284220868381095e-06, 'epoch': 1.27}

 42%|████▏     | 5647/13317 [58:12<1:18:25,  1.63it/s]
 42%|████▏     | 5648/13317 [58:13<1:18:28,  1.63it/s]
                                                      
{'loss': 0.2184, 'grad_norm': 0.6548784971237183, 'learning_rate': 7.127236057009783e-06, 'epoch': 1.27}

 42%|████▏     | 5648/13317 [58:13<1:18:28,  1.63it/s]
 42%|████▏     | 5649/13317 [58:13<1:18:27,  1.63it/s]
                                                      
{'loss': 0.2818, 'grad_norm': 0.7229523658752441, 'learning_rate': 7.126049881017818e-06, 'epoch': 1.27}

 42%|████▏     | 5649/13317 [58:13<1:18:27,  1.63it/s]
 42%|████▏     | 5650/13317 [58:14<1:18:28,  1.63it/s]
                                                      
{'loss': 0.1602, 'grad_norm': 0.45266884565353394, 'learning_rate': 7.124863558943713e-06, 'epoch': 1.27}

 42%|████▏     | 5650/13317 [58:14<1:18:28,  1.63it/s]
 42%|████▏     | 5651/13317 [58:14<1:18:27,  1.63it/s]
                                                      
{'loss': 0.1529, 'grad_norm': 0.43648242950439453, 'learning_rate': 7.123677090868985e-06, 'epoch': 1.27}

 42%|████▏     | 5651/13317 [58:15<1:18:27,  1.63it/s]
 42%|████▏     | 5652/13317 [58:15<1:18:24,  1.63it/s]
                                                      
{'loss': 0.1828, 'grad_norm': 0.5447354912757874, 'learning_rate': 7.122490476875154e-06, 'epoch': 1.27}

 42%|████▏     | 5652/13317 [58:15<1:18:24,  1.63it/s]
 42%|████▏     | 5653/13317 [58:16<1:18:23,  1.63it/s]
                                                      
{'loss': 0.1838, 'grad_norm': 0.5033540725708008, 'learning_rate': 7.121303717043755e-06, 'epoch': 1.27}

 42%|████▏     | 5653/13317 [58:16<1:18:23,  1.63it/s]
 42%|████▏     | 5654/13317 [58:16<1:18:26,  1.63it/s]
                                                      
{'loss': 0.2029, 'grad_norm': 0.4927610456943512, 'learning_rate': 7.120116811456329e-06, 'epoch': 1.27}

 42%|████▏     | 5654/13317 [58:16<1:18:26,  1.63it/s]
 42%|████▏     | 5655/13317 [58:17<1:18:22,  1.63it/s]
                                                      
{'loss': 0.2447, 'grad_norm': 0.5882678031921387, 'learning_rate': 7.118929760194431e-06, 'epoch': 1.27}

 42%|████▏     | 5655/13317 [58:17<1:18:22,  1.63it/s]
 42%|████▏     | 5656/13317 [58:17<1:18:20,  1.63it/s]
                                                      
{'loss': 0.2363, 'grad_norm': 0.5618509650230408, 'learning_rate': 7.117742563339622e-06, 'epoch': 1.27}

 42%|████▏     | 5656/13317 [58:18<1:18:20,  1.63it/s]
 42%|████▏     | 5657/13317 [58:18<1:18:21,  1.63it/s]
                                                      
{'loss': 0.2048, 'grad_norm': 0.5229693651199341, 'learning_rate': 7.116555220973477e-06, 'epoch': 1.27}

 42%|████▏     | 5657/13317 [58:18<1:18:21,  1.63it/s]
 42%|████▏     | 5658/13317 [58:19<1:18:20,  1.63it/s]
                                                      
{'loss': 0.2406, 'grad_norm': 0.5579347014427185, 'learning_rate': 7.1153677331775765e-06, 'epoch': 1.27}

 42%|████▏     | 5658/13317 [58:19<1:18:20,  1.63it/s]
 42%|████▏     | 5659/13317 [58:19<1:18:22,  1.63it/s]
                                                      
{'loss': 0.2038, 'grad_norm': 0.5642748475074768, 'learning_rate': 7.114180100033515e-06, 'epoch': 1.27}

 42%|████▏     | 5659/13317 [58:19<1:18:22,  1.63it/s]
 43%|████▎     | 5660/13317 [58:20<1:18:20,  1.63it/s]
                                                      
{'loss': 0.1587, 'grad_norm': 0.48029062151908875, 'learning_rate': 7.112992321622896e-06, 'epoch': 1.28}

 43%|████▎     | 5660/13317 [58:20<1:18:20,  1.63it/s]
 43%|████▎     | 5661/13317 [58:21<1:18:16,  1.63it/s]
                                                      
{'loss': 0.1729, 'grad_norm': 0.5420202612876892, 'learning_rate': 7.111804398027331e-06, 'epoch': 1.28}

 43%|████▎     | 5661/13317 [58:21<1:18:16,  1.63it/s]
 43%|████▎     | 5662/13317 [58:21<1:18:16,  1.63it/s]
                                                      
{'loss': 0.1851, 'grad_norm': 0.5157152414321899, 'learning_rate': 7.110616329328443e-06, 'epoch': 1.28}

 43%|████▎     | 5662/13317 [58:21<1:18:16,  1.63it/s]
 43%|████▎     | 5663/13317 [58:22<1:18:15,  1.63it/s]
                                                      
{'loss': 0.1783, 'grad_norm': 0.5141317248344421, 'learning_rate': 7.109428115607868e-06, 'epoch': 1.28}

 43%|████▎     | 5663/13317 [58:22<1:18:15,  1.63it/s]
 43%|████▎     | 5664/13317 [58:22<1:18:14,  1.63it/s]
                                                      
{'loss': 0.2517, 'grad_norm': 0.6088061332702637, 'learning_rate': 7.108239756947246e-06, 'epoch': 1.28}

 43%|████▎     | 5664/13317 [58:23<1:18:14,  1.63it/s]
 43%|████▎     | 5665/13317 [58:23<1:18:12,  1.63it/s]
                                                      
{'loss': 0.2685, 'grad_norm': 0.6128682494163513, 'learning_rate': 7.107051253428229e-06, 'epoch': 1.28}

 43%|████▎     | 5665/13317 [58:23<1:18:12,  1.63it/s]
 43%|████▎     | 5666/13317 [58:24<1:18:12,  1.63it/s]
                                                      
{'loss': 0.2144, 'grad_norm': 0.5175162553787231, 'learning_rate': 7.105862605132481e-06, 'epoch': 1.28}

 43%|████▎     | 5666/13317 [58:24<1:18:12,  1.63it/s]
 43%|████▎     | 5667/13317 [58:24<1:18:11,  1.63it/s]
                                                      
{'loss': 0.1943, 'grad_norm': 0.5992981791496277, 'learning_rate': 7.104673812141676e-06, 'epoch': 1.28}

 43%|████▎     | 5667/13317 [58:24<1:18:11,  1.63it/s]
 43%|████▎     | 5668/13317 [58:25<1:18:10,  1.63it/s]
                                                      
{'loss': 0.1779, 'grad_norm': 0.5206093192100525, 'learning_rate': 7.103484874537494e-06, 'epoch': 1.28}

 43%|████▎     | 5668/13317 [58:25<1:18:10,  1.63it/s]
 43%|████▎     | 5669/13317 [58:25<1:18:13,  1.63it/s]
                                                      
{'loss': 0.2219, 'grad_norm': 0.534754753112793, 'learning_rate': 7.102295792401631e-06, 'epoch': 1.28}

 43%|████▎     | 5669/13317 [58:26<1:18:13,  1.63it/s]
 43%|████▎     | 5670/13317 [58:26<1:18:10,  1.63it/s]
                                                      
{'loss': 0.2006, 'grad_norm': 0.5158679485321045, 'learning_rate': 7.1011065658157855e-06, 'epoch': 1.28}

 43%|████▎     | 5670/13317 [58:26<1:18:10,  1.63it/s]
 43%|████▎     | 5671/13317 [58:27<1:18:10,  1.63it/s]
                                                      
{'loss': 0.2276, 'grad_norm': 0.5390929579734802, 'learning_rate': 7.099917194861674e-06, 'epoch': 1.28}

 43%|████▎     | 5671/13317 [58:27<1:18:10,  1.63it/s]
 43%|████▎     | 5672/13317 [58:27<1:18:08,  1.63it/s]
                                                      
{'loss': 0.2265, 'grad_norm': 0.5687742829322815, 'learning_rate': 7.098727679621016e-06, 'epoch': 1.28}

 43%|████▎     | 5672/13317 [58:27<1:18:08,  1.63it/s]
 43%|████▎     | 5673/13317 [58:28<1:18:05,  1.63it/s]
                                                      
{'loss': 0.1839, 'grad_norm': 0.4791073799133301, 'learning_rate': 7.097538020175544e-06, 'epoch': 1.28}

 43%|████▎     | 5673/13317 [58:28<1:18:05,  1.63it/s]
 43%|████▎     | 5674/13317 [58:29<1:18:08,  1.63it/s]
                                                      
{'loss': 0.1898, 'grad_norm': 0.5081892013549805, 'learning_rate': 7.096348216607003e-06, 'epoch': 1.28}

 43%|████▎     | 5674/13317 [58:29<1:18:08,  1.63it/s]
 43%|████▎     | 5675/13317 [58:29<1:18:08,  1.63it/s]
                                                      
{'loss': 0.2732, 'grad_norm': 0.6252506971359253, 'learning_rate': 7.09515826899714e-06, 'epoch': 1.28}

 43%|████▎     | 5675/13317 [58:29<1:18:08,  1.63it/s]
 43%|████▎     | 5676/13317 [58:30<1:18:07,  1.63it/s]
                                                      
{'loss': 0.1949, 'grad_norm': 0.45942771434783936, 'learning_rate': 7.093968177427722e-06, 'epoch': 1.28}

 43%|████▎     | 5676/13317 [58:30<1:18:07,  1.63it/s]
 43%|████▎     | 5677/13317 [58:30<1:18:08,  1.63it/s]
                                                      
{'loss': 0.2535, 'grad_norm': 0.6767109632492065, 'learning_rate': 7.092777941980517e-06, 'epoch': 1.28}

 43%|████▎     | 5677/13317 [58:30<1:18:08,  1.63it/s]
 43%|████▎     | 5678/13317 [58:31<1:18:08,  1.63it/s]
                                                      
{'loss': 0.2579, 'grad_norm': 0.5855813026428223, 'learning_rate': 7.091587562737312e-06, 'epoch': 1.28}

 43%|████▎     | 5678/13317 [58:31<1:18:08,  1.63it/s]
 43%|████▎     | 5679/13317 [58:32<1:18:09,  1.63it/s]
                                                      
{'loss': 0.183, 'grad_norm': 0.4984695315361023, 'learning_rate': 7.090397039779894e-06, 'epoch': 1.28}

 43%|████▎     | 5679/13317 [58:32<1:18:09,  1.63it/s]
 43%|████▎     | 5680/13317 [58:32<1:18:04,  1.63it/s]
                                                      
{'loss': 0.2064, 'grad_norm': 0.5420662760734558, 'learning_rate': 7.089206373190065e-06, 'epoch': 1.28}

 43%|████▎     | 5680/13317 [58:32<1:18:04,  1.63it/s]
 43%|████▎     | 5681/13317 [58:33<1:18:04,  1.63it/s]
                                                      
{'loss': 0.1952, 'grad_norm': 0.5061120390892029, 'learning_rate': 7.088015563049638e-06, 'epoch': 1.28}

 43%|████▎     | 5681/13317 [58:33<1:18:04,  1.63it/s]
 43%|████▎     | 5682/13317 [58:33<1:18:03,  1.63it/s]
                                                      
{'loss': 0.2105, 'grad_norm': 0.518169105052948, 'learning_rate': 7.086824609440434e-06, 'epoch': 1.28}

 43%|████▎     | 5682/13317 [58:34<1:18:03,  1.63it/s]
 43%|████▎     | 5683/13317 [58:34<1:18:02,  1.63it/s]
                                                      
{'loss': 0.1947, 'grad_norm': 0.5542079210281372, 'learning_rate': 7.085633512444283e-06, 'epoch': 1.28}

 43%|████▎     | 5683/13317 [58:34<1:18:02,  1.63it/s]
 43%|████▎     | 5684/13317 [58:35<1:18:02,  1.63it/s]
                                                      
{'loss': 0.2026, 'grad_norm': 0.4895102083683014, 'learning_rate': 7.084442272143026e-06, 'epoch': 1.28}

 43%|████▎     | 5684/13317 [58:35<1:18:02,  1.63it/s]
 43%|████▎     | 5685/13317 [58:35<1:18:00,  1.63it/s]
                                                      
{'loss': 0.1895, 'grad_norm': 0.5462807416915894, 'learning_rate': 7.0832508886185155e-06, 'epoch': 1.28}

 43%|████▎     | 5685/13317 [58:35<1:18:00,  1.63it/s]
 43%|████▎     | 5686/13317 [58:36<1:18:00,  1.63it/s]
                                                      
{'loss': 0.1991, 'grad_norm': 0.5684618353843689, 'learning_rate': 7.082059361952611e-06, 'epoch': 1.28}

 43%|████▎     | 5686/13317 [58:36<1:18:00,  1.63it/s]
 43%|████▎     | 5687/13317 [58:36<1:17:59,  1.63it/s]
                                                      
{'loss': 0.1894, 'grad_norm': 0.43562746047973633, 'learning_rate': 7.080867692227182e-06, 'epoch': 1.28}

 43%|████▎     | 5687/13317 [58:37<1:17:59,  1.63it/s]
 43%|████▎     | 5688/13317 [58:37<1:18:00,  1.63it/s]
                                                      
{'loss': 0.1597, 'grad_norm': 0.4821782410144806, 'learning_rate': 7.079675879524111e-06, 'epoch': 1.28}

 43%|████▎     | 5688/13317 [58:37<1:18:00,  1.63it/s]
 43%|████▎     | 5689/13317 [58:38<1:17:59,  1.63it/s]
                                                      
{'loss': 0.2572, 'grad_norm': 0.7311716079711914, 'learning_rate': 7.078483923925288e-06, 'epoch': 1.28}

 43%|████▎     | 5689/13317 [58:38<1:17:59,  1.63it/s]
 43%|████▎     | 5690/13317 [58:38<1:17:58,  1.63it/s]
                                                      
{'loss': 0.2071, 'grad_norm': 0.5197800397872925, 'learning_rate': 7.077291825512611e-06, 'epoch': 1.28}

 43%|████▎     | 5690/13317 [58:38<1:17:58,  1.63it/s]
 43%|████▎     | 5691/13317 [58:39<1:18:00,  1.63it/s]
                                                      
{'loss': 0.1636, 'grad_norm': 0.39492130279541016, 'learning_rate': 7.076099584367991e-06, 'epoch': 1.28}

 43%|████▎     | 5691/13317 [58:39<1:18:00,  1.63it/s]
 43%|████▎     | 5692/13317 [58:40<1:17:58,  1.63it/s]
                                                      
{'loss': 0.1997, 'grad_norm': 0.49247002601623535, 'learning_rate': 7.074907200573348e-06, 'epoch': 1.28}

 43%|████▎     | 5692/13317 [58:40<1:17:58,  1.63it/s]
 43%|████▎     | 5693/13317 [58:40<1:17:59,  1.63it/s]
                                                      
{'loss': 0.1738, 'grad_norm': 0.4859948754310608, 'learning_rate': 7.07371467421061e-06, 'epoch': 1.28}

 43%|████▎     | 5693/13317 [58:40<1:17:59,  1.63it/s]
 43%|████▎     | 5694/13317 [58:41<1:17:59,  1.63it/s]
                                                      
{'loss': 0.2383, 'grad_norm': 0.5415642857551575, 'learning_rate': 7.0725220053617175e-06, 'epoch': 1.28}

 43%|████▎     | 5694/13317 [58:41<1:17:59,  1.63it/s]
 43%|████▎     | 5695/13317 [58:41<1:17:54,  1.63it/s]
                                                      
{'loss': 0.1883, 'grad_norm': 0.5400936603546143, 'learning_rate': 7.071329194108618e-06, 'epoch': 1.28}

 43%|████▎     | 5695/13317 [58:42<1:17:54,  1.63it/s]
 43%|████▎     | 5696/13317 [58:42<1:17:54,  1.63it/s]
                                                      
{'loss': 0.2494, 'grad_norm': 0.5553784966468811, 'learning_rate': 7.070136240533274e-06, 'epoch': 1.28}

 43%|████▎     | 5696/13317 [58:42<1:17:54,  1.63it/s]
 43%|████▎     | 5697/13317 [58:43<1:17:57,  1.63it/s]
                                                      
{'loss': 0.1816, 'grad_norm': 0.5051487684249878, 'learning_rate': 7.068943144717648e-06, 'epoch': 1.28}

 43%|████▎     | 5697/13317 [58:43<1:17:57,  1.63it/s]
 43%|████▎     | 5698/13317 [58:43<1:17:56,  1.63it/s]
                                                      
{'loss': 0.1948, 'grad_norm': 0.5941091775894165, 'learning_rate': 7.067749906743724e-06, 'epoch': 1.28}

 43%|████▎     | 5698/13317 [58:43<1:17:56,  1.63it/s]
 43%|████▎     | 5699/13317 [58:44<1:17:55,  1.63it/s]
                                                      
{'loss': 0.2157, 'grad_norm': 0.5236202478408813, 'learning_rate': 7.0665565266934865e-06, 'epoch': 1.28}

 43%|████▎     | 5699/13317 [58:44<1:17:55,  1.63it/s]
 43%|████▎     | 5700/13317 [58:44<1:17:57,  1.63it/s]
                                                      
{'loss': 0.1692, 'grad_norm': 0.5104272365570068, 'learning_rate': 7.065363004648935e-06, 'epoch': 1.28}

 43%|████▎     | 5700/13317 [58:45<1:17:57,  1.63it/s]
 43%|████▎     | 5701/13317 [58:45<1:17:54,  1.63it/s]
                                                      
{'loss': 0.1804, 'grad_norm': 0.5845767259597778, 'learning_rate': 7.064169340692076e-06, 'epoch': 1.28}

 43%|████▎     | 5701/13317 [58:45<1:17:54,  1.63it/s]
 43%|████▎     | 5702/13317 [58:46<1:17:53,  1.63it/s]
                                                      
{'loss': 0.2668, 'grad_norm': 0.6548738479614258, 'learning_rate': 7.062975534904928e-06, 'epoch': 1.28}

 43%|████▎     | 5702/13317 [58:46<1:17:53,  1.63it/s]
 43%|████▎     | 5703/13317 [58:46<1:17:52,  1.63it/s]
                                                      
{'loss': 0.1717, 'grad_norm': 0.46839800477027893, 'learning_rate': 7.061781587369518e-06, 'epoch': 1.28}

 43%|████▎     | 5703/13317 [58:46<1:17:52,  1.63it/s]
 43%|████▎     | 5704/13317 [58:47<1:17:51,  1.63it/s]
                                                      
{'loss': 0.1501, 'grad_norm': 0.4087236821651459, 'learning_rate': 7.060587498167881e-06, 'epoch': 1.28}

 43%|████▎     | 5704/13317 [58:47<1:17:51,  1.63it/s]
 43%|████▎     | 5705/13317 [58:48<1:17:49,  1.63it/s]
                                                      
{'loss': 0.2103, 'grad_norm': 0.5237478017807007, 'learning_rate': 7.0593932673820686e-06, 'epoch': 1.29}

 43%|████▎     | 5705/13317 [58:48<1:17:49,  1.63it/s]
 43%|████▎     | 5706/13317 [58:48<1:17:48,  1.63it/s]
                                                      
{'loss': 0.191, 'grad_norm': 0.4682569205760956, 'learning_rate': 7.05819889509413e-06, 'epoch': 1.29}

 43%|████▎     | 5706/13317 [58:48<1:17:48,  1.63it/s]
 43%|████▎     | 5707/13317 [58:49<1:17:50,  1.63it/s]
                                                      
{'loss': 0.1926, 'grad_norm': 0.5603281855583191, 'learning_rate': 7.057004381386137e-06, 'epoch': 1.29}

 43%|████▎     | 5707/13317 [58:49<1:17:50,  1.63it/s]
 43%|████▎     | 5708/13317 [58:49<1:17:51,  1.63it/s]
                                                      
{'loss': 0.1874, 'grad_norm': 0.5248541831970215, 'learning_rate': 7.055809726340162e-06, 'epoch': 1.29}

 43%|████▎     | 5708/13317 [58:50<1:17:51,  1.63it/s]
 43%|████▎     | 5709/13317 [58:50<1:17:50,  1.63it/s]
                                                      
{'loss': 0.2322, 'grad_norm': 0.49021443724632263, 'learning_rate': 7.05461493003829e-06, 'epoch': 1.29}

 43%|████▎     | 5709/13317 [58:50<1:17:50,  1.63it/s]
 43%|████▎     | 5710/13317 [58:51<1:17:49,  1.63it/s]
                                                      
{'loss': 0.1546, 'grad_norm': 0.4348529875278473, 'learning_rate': 7.05341999256262e-06, 'epoch': 1.29}

 43%|████▎     | 5710/13317 [58:51<1:17:49,  1.63it/s]
 43%|████▎     | 5711/13317 [58:51<1:17:46,  1.63it/s]
                                                      
{'loss': 0.2188, 'grad_norm': 0.6540827751159668, 'learning_rate': 7.052224913995253e-06, 'epoch': 1.29}

 43%|████▎     | 5711/13317 [58:51<1:17:46,  1.63it/s]
 43%|████▎     | 5712/13317 [58:52<1:17:48,  1.63it/s]
                                                      
{'loss': 0.2081, 'grad_norm': 0.5484642386436462, 'learning_rate': 7.051029694418307e-06, 'epoch': 1.29}

 43%|████▎     | 5712/13317 [58:52<1:17:48,  1.63it/s]
 43%|████▎     | 5713/13317 [58:52<1:17:48,  1.63it/s]
                                                      
{'loss': 0.2783, 'grad_norm': 0.6492037177085876, 'learning_rate': 7.049834333913903e-06, 'epoch': 1.29}

 43%|████▎     | 5713/13317 [58:53<1:17:48,  1.63it/s]
 43%|████▎     | 5714/13317 [58:53<1:17:46,  1.63it/s]
                                                      
{'loss': 0.2268, 'grad_norm': 0.6197230815887451, 'learning_rate': 7.0486388325641764e-06, 'epoch': 1.29}

 43%|████▎     | 5714/13317 [58:53<1:17:46,  1.63it/s]
 43%|████▎     | 5715/13317 [58:54<1:17:45,  1.63it/s]
                                                      
{'loss': 0.1779, 'grad_norm': 0.4930244982242584, 'learning_rate': 7.047443190451271e-06, 'epoch': 1.29}

 43%|████▎     | 5715/13317 [58:54<1:17:45,  1.63it/s]
 43%|████▎     | 5716/13317 [58:54<1:17:41,  1.63it/s]
                                                      
{'loss': 0.2809, 'grad_norm': 0.7048065662384033, 'learning_rate': 7.046247407657338e-06, 'epoch': 1.29}

 43%|████▎     | 5716/13317 [58:54<1:17:41,  1.63it/s]
 43%|████▎     | 5717/13317 [58:55<1:17:40,  1.63it/s]
                                                      
{'loss': 0.1861, 'grad_norm': 0.5600019097328186, 'learning_rate': 7.045051484264545e-06, 'epoch': 1.29}

 43%|████▎     | 5717/13317 [58:55<1:17:40,  1.63it/s]
 43%|████▎     | 5718/13317 [58:56<1:17:39,  1.63it/s]
                                                      
{'loss': 0.2201, 'grad_norm': 0.5733352303504944, 'learning_rate': 7.04385542035506e-06, 'epoch': 1.29}

 43%|████▎     | 5718/13317 [58:56<1:17:39,  1.63it/s]
 43%|████▎     | 5719/13317 [58:56<1:17:41,  1.63it/s]
                                                      
{'loss': 0.2145, 'grad_norm': 0.6159857511520386, 'learning_rate': 7.042659216011067e-06, 'epoch': 1.29}

 43%|████▎     | 5719/13317 [58:56<1:17:41,  1.63it/s]
 43%|████▎     | 5720/13317 [58:57<1:17:40,  1.63it/s]
                                                      
{'loss': 0.1978, 'grad_norm': 0.5224574208259583, 'learning_rate': 7.041462871314758e-06, 'epoch': 1.29}

 43%|████▎     | 5720/13317 [58:57<1:17:40,  1.63it/s]
 43%|████▎     | 5721/13317 [58:57<1:17:38,  1.63it/s]
                                                      
{'loss': 0.1849, 'grad_norm': 0.49636971950531006, 'learning_rate': 7.040266386348333e-06, 'epoch': 1.29}

 43%|████▎     | 5721/13317 [58:57<1:17:38,  1.63it/s]
 43%|████▎     | 5722/13317 [58:58<1:17:42,  1.63it/s]
                                                      
{'loss': 0.1863, 'grad_norm': 0.5191925168037415, 'learning_rate': 7.039069761194006e-06, 'epoch': 1.29}

 43%|████▎     | 5722/13317 [58:58<1:17:42,  1.63it/s]
 43%|████▎     | 5723/13317 [58:59<1:17:40,  1.63it/s]
                                                      
{'loss': 0.226, 'grad_norm': 0.5733911395072937, 'learning_rate': 7.0378729959339965e-06, 'epoch': 1.29}

 43%|████▎     | 5723/13317 [58:59<1:17:40,  1.63it/s]
 43%|████▎     | 5724/13317 [58:59<1:17:38,  1.63it/s]
                                                      
{'loss': 0.2025, 'grad_norm': 0.5079026222229004, 'learning_rate': 7.036676090650535e-06, 'epoch': 1.29}

 43%|████▎     | 5724/13317 [58:59<1:17:38,  1.63it/s]
 43%|████▎     | 5725/13317 [59:00<1:17:38,  1.63it/s]
                                                      
{'loss': 0.2123, 'grad_norm': 0.5650913119316101, 'learning_rate': 7.035479045425862e-06, 'epoch': 1.29}

 43%|████▎     | 5725/13317 [59:00<1:17:38,  1.63it/s]
 43%|████▎     | 5726/13317 [59:00<1:17:36,  1.63it/s]
                                                      
{'loss': 0.1899, 'grad_norm': 0.5004392266273499, 'learning_rate': 7.0342818603422254e-06, 'epoch': 1.29}

 43%|████▎     | 5726/13317 [59:01<1:17:36,  1.63it/s]
 43%|████▎     | 5727/13317 [59:01<1:17:36,  1.63it/s]
                                                      
{'loss': 0.1817, 'grad_norm': 0.4673456847667694, 'learning_rate': 7.0330845354818855e-06, 'epoch': 1.29}

 43%|████▎     | 5727/13317 [59:01<1:17:36,  1.63it/s]
 43%|████▎     | 5728/13317 [59:02<1:17:35,  1.63it/s]
                                                      
{'loss': 0.1896, 'grad_norm': 0.4609438180923462, 'learning_rate': 7.0318870709271126e-06, 'epoch': 1.29}

 43%|████▎     | 5728/13317 [59:02<1:17:35,  1.63it/s]
 43%|████▎     | 5729/13317 [59:02<1:17:35,  1.63it/s]
                                                      
{'loss': 0.2356, 'grad_norm': 0.5239904522895813, 'learning_rate': 7.030689466760182e-06, 'epoch': 1.29}

 43%|████▎     | 5729/13317 [59:02<1:17:35,  1.63it/s]
 43%|████▎     | 5730/13317 [59:03<1:17:33,  1.63it/s]
                                                      
{'loss': 0.2435, 'grad_norm': 0.5579047203063965, 'learning_rate': 7.029491723063386e-06, 'epoch': 1.29}

 43%|████▎     | 5730/13317 [59:03<1:17:33,  1.63it/s]
 43%|████▎     | 5731/13317 [59:03<1:17:33,  1.63it/s]
                                                      
{'loss': 0.1909, 'grad_norm': 0.5093385577201843, 'learning_rate': 7.028293839919018e-06, 'epoch': 1.29}

 43%|████▎     | 5731/13317 [59:04<1:17:33,  1.63it/s]
 43%|████▎     | 5732/13317 [59:04<1:17:32,  1.63it/s]
                                                      
{'loss': 0.1867, 'grad_norm': 0.553626298904419, 'learning_rate': 7.027095817409388e-06, 'epoch': 1.29}

 43%|████▎     | 5732/13317 [59:04<1:17:32,  1.63it/s]
 43%|████▎     | 5733/13317 [59:05<1:17:32,  1.63it/s]
                                                      
{'loss': 0.2106, 'grad_norm': 0.6184285879135132, 'learning_rate': 7.025897655616812e-06, 'epoch': 1.29}

 43%|████▎     | 5733/13317 [59:05<1:17:32,  1.63it/s]
 43%|████▎     | 5734/13317 [59:05<1:17:37,  1.63it/s]
                                                      
{'loss': 0.2089, 'grad_norm': 0.5051038861274719, 'learning_rate': 7.024699354623618e-06, 'epoch': 1.29}

 43%|████▎     | 5734/13317 [59:05<1:17:37,  1.63it/s]
 43%|████▎     | 5735/13317 [59:06<1:17:34,  1.63it/s]
                                                      
{'loss': 0.172, 'grad_norm': 0.4330154061317444, 'learning_rate': 7.023500914512139e-06, 'epoch': 1.29}

 43%|████▎     | 5735/13317 [59:06<1:17:34,  1.63it/s]
 43%|████▎     | 5736/13317 [59:07<1:17:34,  1.63it/s]
                                                      
{'loss': 0.1675, 'grad_norm': 0.5224897265434265, 'learning_rate': 7.022302335364722e-06, 'epoch': 1.29}

 43%|████▎     | 5736/13317 [59:07<1:17:34,  1.63it/s]
 43%|████▎     | 5737/13317 [59:07<1:17:35,  1.63it/s]
                                                      
{'loss': 0.1921, 'grad_norm': 0.5507612824440002, 'learning_rate': 7.021103617263724e-06, 'epoch': 1.29}

 43%|████▎     | 5737/13317 [59:07<1:17:35,  1.63it/s]
 43%|████▎     | 5738/13317 [59:08<1:17:36,  1.63it/s]
                                                      
{'loss': 0.1646, 'grad_norm': 0.40742596983909607, 'learning_rate': 7.0199047602915045e-06, 'epoch': 1.29}

 43%|████▎     | 5738/13317 [59:08<1:17:36,  1.63it/s]
 43%|████▎     | 5739/13317 [59:08<1:17:32,  1.63it/s]
                                                      
{'loss': 0.2069, 'grad_norm': 0.5186066031455994, 'learning_rate': 7.018705764530444e-06, 'epoch': 1.29}

 43%|████▎     | 5739/13317 [59:09<1:17:32,  1.63it/s]
 43%|████▎     | 5740/13317 [59:09<1:17:30,  1.63it/s]
                                                      
{'loss': 0.2289, 'grad_norm': 0.5604831576347351, 'learning_rate': 7.01750663006292e-06, 'epoch': 1.29}

 43%|████▎     | 5740/13317 [59:09<1:17:30,  1.63it/s]
 43%|████▎     | 5741/13317 [59:10<1:17:30,  1.63it/s]
                                                      
{'loss': 0.1892, 'grad_norm': 0.5337096452713013, 'learning_rate': 7.01630735697133e-06, 'epoch': 1.29}

 43%|████▎     | 5741/13317 [59:10<1:17:30,  1.63it/s]
 43%|████▎     | 5742/13317 [59:10<1:17:28,  1.63it/s]
                                                      
{'loss': 0.1962, 'grad_norm': 0.5355504155158997, 'learning_rate': 7.0151079453380766e-06, 'epoch': 1.29}

 43%|████▎     | 5742/13317 [59:10<1:17:28,  1.63it/s]
 43%|████▎     | 5743/13317 [59:11<1:17:28,  1.63it/s]
                                                      
{'loss': 0.2172, 'grad_norm': 0.6506832242012024, 'learning_rate': 7.01390839524557e-06, 'epoch': 1.29}

 43%|████▎     | 5743/13317 [59:11<1:17:28,  1.63it/s]
 43%|████▎     | 5744/13317 [59:11<1:17:27,  1.63it/s]
                                                      
{'loss': 0.1997, 'grad_norm': 0.4811570942401886, 'learning_rate': 7.0127087067762335e-06, 'epoch': 1.29}

 43%|████▎     | 5744/13317 [59:12<1:17:27,  1.63it/s]
 43%|████▎     | 5745/13317 [59:12<1:17:24,  1.63it/s]
                                                      
{'loss': 0.1969, 'grad_norm': 0.538367509841919, 'learning_rate': 7.011508880012498e-06, 'epoch': 1.29}

 43%|████▎     | 5745/13317 [59:12<1:17:24,  1.63it/s]
 43%|████▎     | 5746/13317 [59:13<1:17:28,  1.63it/s]
                                                      
{'loss': 0.1824, 'grad_norm': 0.5186547040939331, 'learning_rate': 7.010308915036805e-06, 'epoch': 1.29}

 43%|████▎     | 5746/13317 [59:13<1:17:28,  1.63it/s]
 43%|████▎     | 5747/13317 [59:13<1:17:26,  1.63it/s]
                                                      
{'loss': 0.1714, 'grad_norm': 0.4770989716053009, 'learning_rate': 7.0091088119316035e-06, 'epoch': 1.29}

 43%|████▎     | 5747/13317 [59:13<1:17:26,  1.63it/s]
 43%|████▎     | 5748/13317 [59:14<1:17:25,  1.63it/s]
                                                      
{'loss': 0.1949, 'grad_norm': 0.5060436725616455, 'learning_rate': 7.007908570779352e-06, 'epoch': 1.29}

 43%|████▎     | 5748/13317 [59:14<1:17:25,  1.63it/s]
 43%|████▎     | 5749/13317 [59:15<1:17:29,  1.63it/s]
                                                      
{'loss': 0.1596, 'grad_norm': 0.4476640820503235, 'learning_rate': 7.006708191662524e-06, 'epoch': 1.3}

 43%|████▎     | 5749/13317 [59:15<1:17:29,  1.63it/s]
 43%|████▎     | 5750/13317 [59:15<1:17:25,  1.63it/s]
                                                      
{'loss': 0.2262, 'grad_norm': 0.48832622170448303, 'learning_rate': 7.005507674663594e-06, 'epoch': 1.3}

 43%|████▎     | 5750/13317 [59:15<1:17:25,  1.63it/s]
 43%|████▎     | 5751/13317 [59:16<1:17:22,  1.63it/s]
                                                      
{'loss': 0.1844, 'grad_norm': 0.5053837299346924, 'learning_rate': 7.004307019865053e-06, 'epoch': 1.3}

 43%|████▎     | 5751/13317 [59:16<1:17:22,  1.63it/s]
 43%|████▎     | 5752/13317 [59:16<1:17:22,  1.63it/s]
                                                      
{'loss': 0.1999, 'grad_norm': 0.5202120542526245, 'learning_rate': 7.003106227349399e-06, 'epoch': 1.3}

 43%|████▎     | 5752/13317 [59:17<1:17:22,  1.63it/s]
 43%|████▎     | 5753/13317 [59:17<1:17:52,  1.62it/s]
                                                      
{'loss': 0.2538, 'grad_norm': 0.5609796643257141, 'learning_rate': 7.0019052971991365e-06, 'epoch': 1.3}

 43%|████▎     | 5753/13317 [59:17<1:17:52,  1.62it/s]
 43%|████▎     | 5754/13317 [59:18<1:17:42,  1.62it/s]
                                                      
{'loss': 0.1876, 'grad_norm': 0.48115915060043335, 'learning_rate': 7.0007042294967845e-06, 'epoch': 1.3}

 43%|████▎     | 5754/13317 [59:18<1:17:42,  1.62it/s]
 43%|████▎     | 5755/13317 [59:18<1:17:33,  1.62it/s]
                                                      
{'loss': 0.1919, 'grad_norm': 0.5433835387229919, 'learning_rate': 6.999503024324867e-06, 'epoch': 1.3}

 43%|████▎     | 5755/13317 [59:18<1:17:33,  1.62it/s]
 43%|████▎     | 5756/13317 [59:19<1:17:30,  1.63it/s]
                                                      
{'loss': 0.1779, 'grad_norm': 0.48889434337615967, 'learning_rate': 6.998301681765921e-06, 'epoch': 1.3}

 43%|████▎     | 5756/13317 [59:19<1:17:30,  1.63it/s]
 43%|████▎     | 5757/13317 [59:19<1:17:25,  1.63it/s]
                                                      
{'loss': 0.208, 'grad_norm': 0.5602658987045288, 'learning_rate': 6.997100201902491e-06, 'epoch': 1.3}

 43%|████▎     | 5757/13317 [59:20<1:17:25,  1.63it/s]
 43%|████▎     | 5758/13317 [59:20<1:17:30,  1.63it/s]
                                                      
{'loss': 0.2089, 'grad_norm': 0.5309603214263916, 'learning_rate': 6.995898584817131e-06, 'epoch': 1.3}

 43%|████▎     | 5758/13317 [59:20<1:17:30,  1.63it/s]
 43%|████▎     | 5759/13317 [59:21<1:17:30,  1.63it/s]
                                                      
{'loss': 0.1582, 'grad_norm': 0.5772702097892761, 'learning_rate': 6.994696830592407e-06, 'epoch': 1.3}

 43%|████▎     | 5759/13317 [59:21<1:17:30,  1.63it/s]
 43%|████▎     | 5760/13317 [59:21<1:17:25,  1.63it/s]
                                                      
{'loss': 0.1834, 'grad_norm': 0.5637660622596741, 'learning_rate': 6.993494939310889e-06, 'epoch': 1.3}

 43%|████▎     | 5760/13317 [59:21<1:17:25,  1.63it/s]
 43%|████▎     | 5761/13317 [59:22<1:17:21,  1.63it/s]
                                                      
{'loss': 0.16, 'grad_norm': 0.47099289298057556, 'learning_rate': 6.992292911055163e-06, 'epoch': 1.3}

 43%|████▎     | 5761/13317 [59:22<1:17:21,  1.63it/s]
 43%|████▎     | 5762/13317 [59:23<1:17:19,  1.63it/s]
                                                      
{'loss': 0.194, 'grad_norm': 0.5082598924636841, 'learning_rate': 6.991090745907818e-06, 'epoch': 1.3}

 43%|████▎     | 5762/13317 [59:23<1:17:19,  1.63it/s]
 43%|████▎     | 5763/13317 [59:23<1:17:24,  1.63it/s]
                                                      
{'loss': 0.192, 'grad_norm': 0.595879852771759, 'learning_rate': 6.9898884439514555e-06, 'epoch': 1.3}

 43%|████▎     | 5763/13317 [59:23<1:17:24,  1.63it/s]
 43%|████▎     | 5764/13317 [59:24<1:17:18,  1.63it/s]
                                                      
{'loss': 0.2103, 'grad_norm': 0.5654543042182922, 'learning_rate': 6.98868600526869e-06, 'epoch': 1.3}

 43%|████▎     | 5764/13317 [59:24<1:17:18,  1.63it/s]
 43%|████▎     | 5765/13317 [59:24<1:17:14,  1.63it/s]
                                                      
{'loss': 0.2595, 'grad_norm': 0.6747859716415405, 'learning_rate': 6.987483429942138e-06, 'epoch': 1.3}

 43%|████▎     | 5765/13317 [59:25<1:17:14,  1.63it/s]
 43%|████▎     | 5766/13317 [59:25<1:17:14,  1.63it/s]
                                                      
{'loss': 0.2019, 'grad_norm': 0.5248414874076843, 'learning_rate': 6.986280718054432e-06, 'epoch': 1.3}

 43%|████▎     | 5766/13317 [59:25<1:17:14,  1.63it/s]
 43%|████▎     | 5767/13317 [59:26<1:17:14,  1.63it/s]
                                                      
{'loss': 0.2825, 'grad_norm': 0.8151994943618774, 'learning_rate': 6.985077869688207e-06, 'epoch': 1.3}

 43%|████▎     | 5767/13317 [59:26<1:17:14,  1.63it/s]
 43%|████▎     | 5768/13317 [59:26<1:17:19,  1.63it/s]
                                                      
{'loss': 0.1565, 'grad_norm': 0.43403360247612, 'learning_rate': 6.983874884926116e-06, 'epoch': 1.3}

 43%|████▎     | 5768/13317 [59:26<1:17:19,  1.63it/s]
 43%|████▎     | 5769/13317 [59:27<1:17:12,  1.63it/s]
                                                      
{'loss': 0.2203, 'grad_norm': 0.5993577241897583, 'learning_rate': 6.982671763850814e-06, 'epoch': 1.3}

 43%|████▎     | 5769/13317 [59:27<1:17:12,  1.63it/s]
 43%|████▎     | 5770/13317 [59:27<1:17:12,  1.63it/s]
                                                      
{'loss': 0.1617, 'grad_norm': 0.40508583188056946, 'learning_rate': 6.981468506544969e-06, 'epoch': 1.3}

 43%|████▎     | 5770/13317 [59:28<1:17:12,  1.63it/s]
 43%|████▎     | 5771/13317 [59:28<1:17:13,  1.63it/s]
                                                      
{'loss': 0.2412, 'grad_norm': 0.5913175940513611, 'learning_rate': 6.980265113091258e-06, 'epoch': 1.3}

 43%|████▎     | 5771/13317 [59:28<1:17:13,  1.63it/s]
 43%|████▎     | 5772/13317 [59:29<1:17:08,  1.63it/s]
                                                      
{'loss': 0.2025, 'grad_norm': 0.528123676776886, 'learning_rate': 6.979061583572366e-06, 'epoch': 1.3}

 43%|████▎     | 5772/13317 [59:29<1:17:08,  1.63it/s]
 43%|████▎     | 5773/13317 [59:29<1:17:15,  1.63it/s]
                                                      
{'loss': 0.2259, 'grad_norm': 0.5871632099151611, 'learning_rate': 6.9778579180709896e-06, 'epoch': 1.3}

 43%|████▎     | 5773/13317 [59:29<1:17:15,  1.63it/s]
 43%|████▎     | 5774/13317 [59:30<1:17:11,  1.63it/s]
                                                      
{'loss': 0.1565, 'grad_norm': 0.4391154646873474, 'learning_rate': 6.9766541166698324e-06, 'epoch': 1.3}

 43%|████▎     | 5774/13317 [59:30<1:17:11,  1.63it/s]
 43%|████▎     | 5775/13317 [59:31<1:17:09,  1.63it/s]
                                                      
{'loss': 0.1719, 'grad_norm': 0.43595486879348755, 'learning_rate': 6.975450179451606e-06, 'epoch': 1.3}

 43%|████▎     | 5775/13317 [59:31<1:17:09,  1.63it/s]
 43%|████▎     | 5776/13317 [59:31<1:17:07,  1.63it/s]
                                                      
{'loss': 0.1565, 'grad_norm': 0.4378621578216553, 'learning_rate': 6.974246106499037e-06, 'epoch': 1.3}

 43%|████▎     | 5776/13317 [59:31<1:17:07,  1.63it/s]
 43%|████▎     | 5777/13317 [59:32<1:17:12,  1.63it/s]
                                                      
{'loss': 0.1674, 'grad_norm': 0.39267244935035706, 'learning_rate': 6.973041897894856e-06, 'epoch': 1.3}

 43%|████▎     | 5777/13317 [59:32<1:17:12,  1.63it/s]
 43%|████▎     | 5778/13317 [59:32<1:17:31,  1.62it/s]
                                                      
{'loss': 0.1821, 'grad_norm': 0.47243478894233704, 'learning_rate': 6.971837553721808e-06, 'epoch': 1.3}

 43%|████▎     | 5778/13317 [59:32<1:17:31,  1.62it/s]
 43%|████▎     | 5779/13317 [59:33<1:17:24,  1.62it/s]
                                                      
{'loss': 0.222, 'grad_norm': 0.6749874949455261, 'learning_rate': 6.970633074062641e-06, 'epoch': 1.3}

 43%|████▎     | 5779/13317 [59:33<1:17:24,  1.62it/s]
 43%|████▎     | 5780/13317 [59:34<1:17:20,  1.62it/s]
                                                      
{'loss': 0.1682, 'grad_norm': 0.485115110874176, 'learning_rate': 6.969428459000116e-06, 'epoch': 1.3}

 43%|████▎     | 5780/13317 [59:34<1:17:20,  1.62it/s]
 43%|████▎     | 5781/13317 [59:34<1:17:14,  1.63it/s]
                                                      
{'loss': 0.1661, 'grad_norm': 0.5310882925987244, 'learning_rate': 6.968223708617004e-06, 'epoch': 1.3}

 43%|████▎     | 5781/13317 [59:34<1:17:14,  1.63it/s]
 43%|████▎     | 5782/13317 [59:35<1:17:10,  1.63it/s]
                                                      
{'loss': 0.2565, 'grad_norm': 0.6550129652023315, 'learning_rate': 6.967018822996081e-06, 'epoch': 1.3}

 43%|████▎     | 5782/13317 [59:35<1:17:10,  1.63it/s]
 43%|████▎     | 5783/13317 [59:35<1:17:14,  1.63it/s]
                                                      
{'loss': 0.1736, 'grad_norm': 0.5353598594665527, 'learning_rate': 6.965813802220139e-06, 'epoch': 1.3}

 43%|████▎     | 5783/13317 [59:36<1:17:14,  1.63it/s]
 43%|████▎     | 5784/13317 [59:36<1:17:12,  1.63it/s]
                                                      
{'loss': 0.192, 'grad_norm': 0.5253236293792725, 'learning_rate': 6.964608646371975e-06, 'epoch': 1.3}

 43%|████▎     | 5784/13317 [59:36<1:17:12,  1.63it/s]
 43%|████▎     | 5785/13317 [59:37<1:17:11,  1.63it/s]
                                                      
{'loss': 0.2469, 'grad_norm': 0.5344371795654297, 'learning_rate': 6.963403355534395e-06, 'epoch': 1.3}

 43%|████▎     | 5785/13317 [59:37<1:17:11,  1.63it/s]
 43%|████▎     | 5786/13317 [59:37<1:17:09,  1.63it/s]
                                                      
{'loss': 0.2215, 'grad_norm': 0.5515610575675964, 'learning_rate': 6.962197929790216e-06, 'epoch': 1.3}

 43%|████▎     | 5786/13317 [59:37<1:17:09,  1.63it/s]
 43%|████▎     | 5787/13317 [59:38<1:17:05,  1.63it/s]
                                                      
{'loss': 0.2106, 'grad_norm': 0.5441672205924988, 'learning_rate': 6.960992369222261e-06, 'epoch': 1.3}

 43%|████▎     | 5787/13317 [59:38<1:17:05,  1.63it/s]
 43%|████▎     | 5788/13317 [59:39<1:17:06,  1.63it/s]
                                                      
{'loss': 0.2159, 'grad_norm': 0.5094538927078247, 'learning_rate': 6.959786673913371e-06, 'epoch': 1.3}

 43%|████▎     | 5788/13317 [59:39<1:17:06,  1.63it/s]
 43%|████▎     | 5789/13317 [59:39<1:17:06,  1.63it/s]
                                                      
{'loss': 0.1903, 'grad_norm': 0.543373703956604, 'learning_rate': 6.958580843946383e-06, 'epoch': 1.3}

 43%|████▎     | 5789/13317 [59:39<1:17:06,  1.63it/s]
 43%|████▎     | 5790/13317 [59:40<1:17:01,  1.63it/s]
                                                      
{'loss': 0.1612, 'grad_norm': 0.49636173248291016, 'learning_rate': 6.957374879404154e-06, 'epoch': 1.3}

 43%|████▎     | 5790/13317 [59:40<1:17:01,  1.63it/s]
 43%|████▎     | 5791/13317 [59:40<1:16:59,  1.63it/s]
                                                      
{'loss': 0.2258, 'grad_norm': 0.5361677408218384, 'learning_rate': 6.956168780369546e-06, 'epoch': 1.3}

 43%|████▎     | 5791/13317 [59:40<1:16:59,  1.63it/s]
 43%|████▎     | 5792/13317 [59:41<1:16:56,  1.63it/s]
                                                      
{'loss': 0.1955, 'grad_norm': 0.5069985389709473, 'learning_rate': 6.95496254692543e-06, 'epoch': 1.3}

 43%|████▎     | 5792/13317 [59:41<1:16:56,  1.63it/s]
 44%|████▎     | 5793/13317 [59:42<1:17:03,  1.63it/s]
                                                      
{'loss': 0.1881, 'grad_norm': 0.4437817931175232, 'learning_rate': 6.953756179154688e-06, 'epoch': 1.31}

 44%|████▎     | 5793/13317 [59:42<1:17:03,  1.63it/s]
 44%|████▎     | 5794/13317 [59:42<1:17:00,  1.63it/s]
                                                      
{'loss': 0.1632, 'grad_norm': 0.4513434171676636, 'learning_rate': 6.952549677140209e-06, 'epoch': 1.31}

 44%|████▎     | 5794/13317 [59:42<1:17:00,  1.63it/s]
 44%|████▎     | 5795/13317 [59:43<1:16:58,  1.63it/s]
                                                      
{'loss': 0.1893, 'grad_norm': 0.5769416689872742, 'learning_rate': 6.951343040964894e-06, 'epoch': 1.31}

 44%|████▎     | 5795/13317 [59:43<1:16:58,  1.63it/s]
 44%|████▎     | 5796/13317 [59:43<1:16:59,  1.63it/s]
                                                      
{'loss': 0.2265, 'grad_norm': 0.5968804359436035, 'learning_rate': 6.950136270711652e-06, 'epoch': 1.31}

 44%|████▎     | 5796/13317 [59:44<1:16:59,  1.63it/s]
 44%|████▎     | 5797/13317 [59:44<1:16:57,  1.63it/s]
                                                      
{'loss': 0.1985, 'grad_norm': 0.4956411123275757, 'learning_rate': 6.948929366463397e-06, 'epoch': 1.31}

 44%|████▎     | 5797/13317 [59:44<1:16:57,  1.63it/s]
 44%|████▎     | 5798/13317 [59:45<1:17:03,  1.63it/s]
                                                      
{'loss': 0.1932, 'grad_norm': 0.44473934173583984, 'learning_rate': 6.947722328303062e-06, 'epoch': 1.31}

 44%|████▎     | 5798/13317 [59:45<1:17:03,  1.63it/s]
 44%|████▎     | 5799/13317 [59:45<1:17:02,  1.63it/s]
                                                      
{'loss': 0.2934, 'grad_norm': 0.6195372939109802, 'learning_rate': 6.946515156313578e-06, 'epoch': 1.31}

 44%|████▎     | 5799/13317 [59:45<1:17:02,  1.63it/s]
 44%|████▎     | 5800/13317 [59:46<1:16:58,  1.63it/s]
                                                      
{'loss': 0.2338, 'grad_norm': 0.5951836705207825, 'learning_rate': 6.945307850577894e-06, 'epoch': 1.31}

 44%|████▎     | 5800/13317 [59:46<1:16:58,  1.63it/s]
 44%|████▎     | 5801/13317 [59:46<1:16:54,  1.63it/s]
                                                      
{'loss': 0.1441, 'grad_norm': 0.4219872057437897, 'learning_rate': 6.944100411178962e-06, 'epoch': 1.31}

 44%|████▎     | 5801/13317 [59:47<1:16:54,  1.63it/s]
 44%|████▎     | 5802/13317 [59:47<1:16:55,  1.63it/s]
                                                      
{'loss': 0.1478, 'grad_norm': 0.40886032581329346, 'learning_rate': 6.942892838199749e-06, 'epoch': 1.31}

 44%|████▎     | 5802/13317 [59:47<1:16:55,  1.63it/s]
 44%|████▎     | 5803/13317 [59:48<1:17:01,  1.63it/s]
                                                      
{'loss': 0.2253, 'grad_norm': 0.6520893573760986, 'learning_rate': 6.941685131723225e-06, 'epoch': 1.31}

 44%|████▎     | 5803/13317 [59:48<1:17:01,  1.63it/s]
 44%|████▎     | 5804/13317 [59:48<1:16:57,  1.63it/s]
                                                      
{'loss': 0.3715, 'grad_norm': 0.807641327381134, 'learning_rate': 6.940477291832374e-06, 'epoch': 1.31}

 44%|████▎     | 5804/13317 [59:48<1:16:57,  1.63it/s]
 44%|████▎     | 5805/13317 [59:49<1:16:56,  1.63it/s]
                                                      
{'loss': 0.2079, 'grad_norm': 0.5741638541221619, 'learning_rate': 6.939269318610187e-06, 'epoch': 1.31}

 44%|████▎     | 5805/13317 [59:49<1:16:56,  1.63it/s]
 44%|████▎     | 5806/13317 [59:50<1:16:54,  1.63it/s]
                                                      
{'loss': 0.1275, 'grad_norm': 0.38165590167045593, 'learning_rate': 6.938061212139664e-06, 'epoch': 1.31}

 44%|████▎     | 5806/13317 [59:50<1:16:54,  1.63it/s]
 44%|████▎     | 5807/13317 [59:50<1:16:52,  1.63it/s]
                                                      
{'loss': 0.1547, 'grad_norm': 0.49947354197502136, 'learning_rate': 6.9368529725038166e-06, 'epoch': 1.31}

 44%|████▎     | 5807/13317 [59:50<1:16:52,  1.63it/s]
 44%|████▎     | 5808/13317 [59:51<1:16:59,  1.63it/s]
                                                      
{'loss': 0.1745, 'grad_norm': 0.4497690498828888, 'learning_rate': 6.935644599785661e-06, 'epoch': 1.31}

 44%|████▎     | 5808/13317 [59:51<1:16:59,  1.63it/s]
 44%|████▎     | 5809/13317 [59:51<1:16:53,  1.63it/s]
                                                      
{'loss': 0.2117, 'grad_norm': 0.5438433289527893, 'learning_rate': 6.9344360940682276e-06, 'epoch': 1.31}

 44%|████▎     | 5809/13317 [59:52<1:16:53,  1.63it/s]
 44%|████▎     | 5810/13317 [59:52<1:16:49,  1.63it/s]
                                                      
{'loss': 0.2736, 'grad_norm': 0.529308021068573, 'learning_rate': 6.933227455434552e-06, 'epoch': 1.31}

 44%|████▎     | 5810/13317 [59:52<1:16:49,  1.63it/s]
 44%|████▎     | 5811/13317 [59:53<1:16:48,  1.63it/s]
                                                      
{'loss': 0.2096, 'grad_norm': 0.5030333995819092, 'learning_rate': 6.932018683967679e-06, 'epoch': 1.31}

 44%|████▎     | 5811/13317 [59:53<1:16:48,  1.63it/s]
 44%|████▎     | 5812/13317 [59:53<1:16:47,  1.63it/s]
                                                      
{'loss': 0.1663, 'grad_norm': 0.42222172021865845, 'learning_rate': 6.930809779750668e-06, 'epoch': 1.31}

 44%|████▎     | 5812/13317 [59:53<1:16:47,  1.63it/s]
 44%|████▎     | 5813/13317 [59:54<1:16:55,  1.63it/s]
                                                      
{'loss': 0.2196, 'grad_norm': 0.5732181072235107, 'learning_rate': 6.929600742866581e-06, 'epoch': 1.31}

 44%|████▎     | 5813/13317 [59:54<1:16:55,  1.63it/s]
 44%|████▎     | 5814/13317 [59:54<1:16:52,  1.63it/s]
                                                      
{'loss': 0.1971, 'grad_norm': 0.5274275541305542, 'learning_rate': 6.928391573398491e-06, 'epoch': 1.31}

 44%|████▎     | 5814/13317 [59:55<1:16:52,  1.63it/s]
 44%|████▎     | 5815/13317 [59:55<1:16:49,  1.63it/s]
                                                      
{'loss': 0.2222, 'grad_norm': 0.5919705629348755, 'learning_rate': 6.927182271429483e-06, 'epoch': 1.31}

 44%|████▎     | 5815/13317 [59:55<1:16:49,  1.63it/s]
 44%|████▎     | 5816/13317 [59:56<1:16:46,  1.63it/s]
                                                      
{'loss': 0.2122, 'grad_norm': 0.5177388787269592, 'learning_rate': 6.925972837042647e-06, 'epoch': 1.31}

 44%|████▎     | 5816/13317 [59:56<1:16:46,  1.63it/s]
 44%|████▎     | 5817/13317 [59:56<1:16:43,  1.63it/s]
                                                      
{'loss': 0.1966, 'grad_norm': 0.45854538679122925, 'learning_rate': 6.924763270321086e-06, 'epoch': 1.31}

 44%|████▎     | 5817/13317 [59:56<1:16:43,  1.63it/s]
 44%|████▎     | 5818/13317 [59:57<1:16:51,  1.63it/s]
                                                      
{'loss': 0.1566, 'grad_norm': 0.43911153078079224, 'learning_rate': 6.923553571347907e-06, 'epoch': 1.31}

 44%|████▎     | 5818/13317 [59:57<1:16:51,  1.63it/s]
 44%|████▎     | 5819/13317 [59:58<1:16:51,  1.63it/s]
                                                      
{'loss': 0.2015, 'grad_norm': 0.5063163638114929, 'learning_rate': 6.92234374020623e-06, 'epoch': 1.31}

 44%|████▎     | 5819/13317 [59:58<1:16:51,  1.63it/s]
 44%|████▎     | 5820/13317 [59:58<1:16:48,  1.63it/s]
                                                      
{'loss': 0.2051, 'grad_norm': 0.5594156980514526, 'learning_rate': 6.921133776979186e-06, 'epoch': 1.31}

 44%|████▎     | 5820/13317 [59:58<1:16:48,  1.63it/s]
 44%|████▎     | 5821/13317 [59:59<1:16:47,  1.63it/s]
                                                      
{'loss': 0.1938, 'grad_norm': 0.4733172357082367, 'learning_rate': 6.919923681749908e-06, 'epoch': 1.31}

 44%|████▎     | 5821/13317 [59:59<1:16:47,  1.63it/s]
 44%|████▎     | 5822/13317 [59:59<1:16:47,  1.63it/s]
                                                      
{'loss': 0.1828, 'grad_norm': 0.43083497881889343, 'learning_rate': 6.918713454601547e-06, 'epoch': 1.31}

 44%|████▎     | 5822/13317 [1:00:00<1:16:47,  1.63it/s]
 44%|████▎     | 5823/13317 [1:00:00<1:19:07,  1.58it/s]
                                                        
{'loss': 0.2018, 'grad_norm': 0.570564866065979, 'learning_rate': 6.917503095617255e-06, 'epoch': 1.31}

 44%|████▎     | 5823/13317 [1:00:00<1:19:07,  1.58it/s]
 44%|████▎     | 5824/13317 [1:00:01<1:18:21,  1.59it/s]
                                                        
{'loss': 0.2159, 'grad_norm': 0.5570433139801025, 'learning_rate': 6.916292604880197e-06, 'epoch': 1.31}

 44%|████▎     | 5824/13317 [1:00:01<1:18:21,  1.59it/s]
 44%|████▎     | 5825/13317 [1:00:01<1:17:51,  1.60it/s]
                                                        
{'loss': 0.2135, 'grad_norm': 0.5403984785079956, 'learning_rate': 6.915081982473548e-06, 'epoch': 1.31}

 44%|████▎     | 5825/13317 [1:00:01<1:17:51,  1.60it/s]
 44%|████▎     | 5826/13317 [1:00:02<1:17:28,  1.61it/s]
                                                        
{'loss': 0.1955, 'grad_norm': 0.5417100787162781, 'learning_rate': 6.913871228480489e-06, 'epoch': 1.31}

 44%|████▎     | 5826/13317 [1:00:02<1:17:28,  1.61it/s]
 44%|████▍     | 5827/13317 [1:00:03<1:17:15,  1.62it/s]
                                                        
{'loss': 0.277, 'grad_norm': 0.675613522529602, 'learning_rate': 6.912660342984212e-06, 'epoch': 1.31}

 44%|████▍     | 5827/13317 [1:00:03<1:17:15,  1.62it/s]
 44%|████▍     | 5828/13317 [1:00:03<1:17:08,  1.62it/s]
                                                        
{'loss': 0.23, 'grad_norm': 0.6311497092247009, 'learning_rate': 6.911449326067917e-06, 'epoch': 1.31}

 44%|████▍     | 5828/13317 [1:00:03<1:17:08,  1.62it/s]
 44%|████▍     | 5829/13317 [1:00:04<1:16:59,  1.62it/s]
                                                        
{'loss': 0.2098, 'grad_norm': 0.5096721053123474, 'learning_rate': 6.910238177814816e-06, 'epoch': 1.31}

 44%|████▍     | 5829/13317 [1:00:04<1:16:59,  1.62it/s]
 44%|████▍     | 5830/13317 [1:00:04<1:16:53,  1.62it/s]
                                                        
{'loss': 0.288, 'grad_norm': 0.6555179357528687, 'learning_rate': 6.909026898308126e-06, 'epoch': 1.31}

 44%|████▍     | 5830/13317 [1:00:05<1:16:53,  1.62it/s]
 44%|████▍     | 5831/13317 [1:00:05<1:16:50,  1.62it/s]
                                                        
{'loss': 0.2348, 'grad_norm': 0.6066633462905884, 'learning_rate': 6.907815487631075e-06, 'epoch': 1.31}

 44%|████▍     | 5831/13317 [1:00:05<1:16:50,  1.62it/s]
 44%|████▍     | 5832/13317 [1:00:06<1:16:47,  1.62it/s]
                                                        
{'loss': 0.1702, 'grad_norm': 0.4573027789592743, 'learning_rate': 6.9066039458668996e-06, 'epoch': 1.31}

 44%|████▍     | 5832/13317 [1:00:06<1:16:47,  1.62it/s]
 44%|████▍     | 5833/13317 [1:00:06<1:16:52,  1.62it/s]
                                                        
{'loss': 0.1958, 'grad_norm': 0.519775390625, 'learning_rate': 6.905392273098846e-06, 'epoch': 1.31}

 44%|████▍     | 5833/13317 [1:00:06<1:16:52,  1.62it/s]
 44%|████▍     | 5834/13317 [1:00:07<1:16:49,  1.62it/s]
                                                        
{'loss': 0.3054, 'grad_norm': 0.6034708619117737, 'learning_rate': 6.904180469410167e-06, 'epoch': 1.31}

 44%|████▍     | 5834/13317 [1:00:07<1:16:49,  1.62it/s]
 44%|████▍     | 5835/13317 [1:00:07<1:16:46,  1.62it/s]
                                                        
{'loss': 0.2989, 'grad_norm': 0.6979843378067017, 'learning_rate': 6.9029685348841315e-06, 'epoch': 1.31}

 44%|████▍     | 5835/13317 [1:00:08<1:16:46,  1.62it/s]
 44%|████▍     | 5836/13317 [1:00:08<1:16:44,  1.62it/s]
                                                        
{'loss': 0.3091, 'grad_norm': 0.6512997150421143, 'learning_rate': 6.901756469604005e-06, 'epoch': 1.31}

 44%|████▍     | 5836/13317 [1:00:08<1:16:44,  1.62it/s]
 44%|████▍     | 5837/13317 [1:00:09<1:16:41,  1.63it/s]
                                                        
{'loss': 0.1468, 'grad_norm': 0.4604126214981079, 'learning_rate': 6.9005442736530745e-06, 'epoch': 1.31}

 44%|████▍     | 5837/13317 [1:00:09<1:16:41,  1.63it/s]
 44%|████▍     | 5838/13317 [1:00:09<1:16:43,  1.62it/s]
                                                        
{'loss': 0.2258, 'grad_norm': 0.5316182374954224, 'learning_rate': 6.899331947114629e-06, 'epoch': 1.32}

 44%|████▍     | 5838/13317 [1:00:09<1:16:43,  1.62it/s]
 44%|████▍     | 5839/13317 [1:00:10<1:16:41,  1.63it/s]
                                                        
{'loss': 0.2158, 'grad_norm': 0.603419840335846, 'learning_rate': 6.898119490071968e-06, 'epoch': 1.32}

 44%|████▍     | 5839/13317 [1:00:10<1:16:41,  1.63it/s]
 44%|████▍     | 5840/13317 [1:00:11<1:16:36,  1.63it/s]
                                                        
{'loss': 0.1835, 'grad_norm': 0.4779863953590393, 'learning_rate': 6.896906902608402e-06, 'epoch': 1.32}

 44%|████▍     | 5840/13317 [1:00:11<1:16:36,  1.63it/s]
 44%|████▍     | 5841/13317 [1:00:11<1:16:31,  1.63it/s]
                                                        
{'loss': 0.1714, 'grad_norm': 0.4629531800746918, 'learning_rate': 6.895694184807244e-06, 'epoch': 1.32}

 44%|████▍     | 5841/13317 [1:00:11<1:16:31,  1.63it/s]
 44%|████▍     | 5842/13317 [1:00:12<1:16:28,  1.63it/s]
                                                        
{'loss': 0.1702, 'grad_norm': 0.4930160939693451, 'learning_rate': 6.894481336751826e-06, 'epoch': 1.32}

 44%|████▍     | 5842/13317 [1:00:12<1:16:28,  1.63it/s]
 44%|████▍     | 5843/13317 [1:00:12<1:16:33,  1.63it/s]
                                                        
{'loss': 0.2398, 'grad_norm': 0.5567155480384827, 'learning_rate': 6.89326835852548e-06, 'epoch': 1.32}

 44%|████▍     | 5843/13317 [1:00:12<1:16:33,  1.63it/s]
 44%|████▍     | 5844/13317 [1:00:13<1:16:30,  1.63it/s]
                                                        
{'loss': 0.1845, 'grad_norm': 0.47384342551231384, 'learning_rate': 6.892055250211552e-06, 'epoch': 1.32}

 44%|████▍     | 5844/13317 [1:00:13<1:16:30,  1.63it/s]
 44%|████▍     | 5845/13317 [1:00:14<1:16:27,  1.63it/s]
                                                        
{'loss': 0.2489, 'grad_norm': 0.6420910358428955, 'learning_rate': 6.890842011893395e-06, 'epoch': 1.32}

 44%|████▍     | 5845/13317 [1:00:14<1:16:27,  1.63it/s]
 44%|████▍     | 5846/13317 [1:00:14<1:16:25,  1.63it/s]
                                                        
{'loss': 0.1633, 'grad_norm': 0.4638146162033081, 'learning_rate': 6.889628643654371e-06, 'epoch': 1.32}

 44%|████▍     | 5846/13317 [1:00:14<1:16:25,  1.63it/s]
 44%|████▍     | 5847/13317 [1:00:15<1:16:24,  1.63it/s]
                                                        
{'loss': 0.2058, 'grad_norm': 0.5400323867797852, 'learning_rate': 6.888415145577851e-06, 'epoch': 1.32}

 44%|████▍     | 5847/13317 [1:00:15<1:16:24,  1.63it/s]
 44%|████▍     | 5848/13317 [1:00:15<1:16:29,  1.63it/s]
                                                        
{'loss': 0.1843, 'grad_norm': 0.4907037019729614, 'learning_rate': 6.887201517747216e-06, 'epoch': 1.32}

 44%|████▍     | 5848/13317 [1:00:16<1:16:29,  1.63it/s]
 44%|████▍     | 5849/13317 [1:00:16<1:16:28,  1.63it/s]
                                                        
{'loss': 0.262, 'grad_norm': 0.6044946312904358, 'learning_rate': 6.885987760245856e-06, 'epoch': 1.32}

 44%|████▍     | 5849/13317 [1:00:16<1:16:28,  1.63it/s]
 44%|████▍     | 5850/13317 [1:00:17<1:16:25,  1.63it/s]
                                                        
{'loss': 0.2249, 'grad_norm': 0.5670246481895447, 'learning_rate': 6.8847738731571645e-06, 'epoch': 1.32}

 44%|████▍     | 5850/13317 [1:00:17<1:16:25,  1.63it/s]
 44%|████▍     | 5851/13317 [1:00:17<1:16:21,  1.63it/s]
                                                        
{'loss': 0.2125, 'grad_norm': 0.5649557113647461, 'learning_rate': 6.883559856564555e-06, 'epoch': 1.32}

 44%|████▍     | 5851/13317 [1:00:17<1:16:21,  1.63it/s]
 44%|████▍     | 5852/13317 [1:00:18<1:16:19,  1.63it/s]
                                                        
{'loss': 0.194, 'grad_norm': 0.5408152341842651, 'learning_rate': 6.882345710551439e-06, 'epoch': 1.32}

 44%|████▍     | 5852/13317 [1:00:18<1:16:19,  1.63it/s]
 44%|████▍     | 5853/13317 [1:00:19<1:16:30,  1.63it/s]
                                                        
{'loss': 0.1821, 'grad_norm': 0.48573222756385803, 'learning_rate': 6.8811314352012405e-06, 'epoch': 1.32}

 44%|████▍     | 5853/13317 [1:00:19<1:16:30,  1.63it/s]
 44%|████▍     | 5854/13317 [1:00:19<1:16:24,  1.63it/s]
                                                        
{'loss': 0.1602, 'grad_norm': 0.44109687209129333, 'learning_rate': 6.879917030597397e-06, 'epoch': 1.32}

 44%|████▍     | 5854/13317 [1:00:19<1:16:24,  1.63it/s]
 44%|████▍     | 5855/13317 [1:00:20<1:16:22,  1.63it/s]
                                                        
{'loss': 0.1592, 'grad_norm': 0.3904225528240204, 'learning_rate': 6.878702496823346e-06, 'epoch': 1.32}

 44%|████▍     | 5855/13317 [1:00:20<1:16:22,  1.63it/s]
 44%|████▍     | 5856/13317 [1:00:20<1:16:21,  1.63it/s]
                                                        
{'loss': 0.1748, 'grad_norm': 0.4535578191280365, 'learning_rate': 6.877487833962544e-06, 'epoch': 1.32}

 44%|████▍     | 5856/13317 [1:00:20<1:16:21,  1.63it/s]
 44%|████▍     | 5857/13317 [1:00:21<1:16:22,  1.63it/s]
                                                        
{'loss': 0.2812, 'grad_norm': 0.5212286710739136, 'learning_rate': 6.876273042098447e-06, 'epoch': 1.32}

 44%|████▍     | 5857/13317 [1:00:21<1:16:22,  1.63it/s]
 44%|████▍     | 5858/13317 [1:00:22<1:16:41,  1.62it/s]
                                                        
{'loss': 0.2268, 'grad_norm': 0.5284751057624817, 'learning_rate': 6.875058121314526e-06, 'epoch': 1.32}

 44%|████▍     | 5858/13317 [1:00:22<1:16:41,  1.62it/s]
 44%|████▍     | 5859/13317 [1:00:22<1:16:32,  1.62it/s]
                                                        
{'loss': 0.2242, 'grad_norm': 0.6112297177314758, 'learning_rate': 6.87384307169426e-06, 'epoch': 1.32}

 44%|████▍     | 5859/13317 [1:00:22<1:16:32,  1.62it/s]
 44%|████▍     | 5860/13317 [1:00:23<1:16:25,  1.63it/s]
                                                        
{'loss': 0.179, 'grad_norm': 0.4657804071903229, 'learning_rate': 6.872627893321132e-06, 'epoch': 1.32}

 44%|████▍     | 5860/13317 [1:00:23<1:16:25,  1.63it/s]
 44%|████▍     | 5861/13317 [1:00:23<1:16:22,  1.63it/s]
                                                        
{'loss': 0.1689, 'grad_norm': 0.464062362909317, 'learning_rate': 6.871412586278642e-06, 'epoch': 1.32}

 44%|████▍     | 5861/13317 [1:00:24<1:16:22,  1.63it/s]
 44%|████▍     | 5862/13317 [1:00:24<1:16:19,  1.63it/s]
                                                        
{'loss': 0.2257, 'grad_norm': 0.5616302490234375, 'learning_rate': 6.870197150650292e-06, 'epoch': 1.32}

 44%|████▍     | 5862/13317 [1:00:24<1:16:19,  1.63it/s]
 44%|████▍     | 5863/13317 [1:00:25<1:16:24,  1.63it/s]
                                                        
{'loss': 0.1551, 'grad_norm': 0.41526180505752563, 'learning_rate': 6.868981586519595e-06, 'epoch': 1.32}

 44%|████▍     | 5863/13317 [1:00:25<1:16:24,  1.63it/s]
 44%|████▍     | 5864/13317 [1:00:25<1:16:23,  1.63it/s]
                                                        
{'loss': 0.1913, 'grad_norm': 0.46299150586128235, 'learning_rate': 6.867765893970075e-06, 'epoch': 1.32}

 44%|████▍     | 5864/13317 [1:00:25<1:16:23,  1.63it/s]
 44%|████▍     | 5865/13317 [1:00:26<1:16:22,  1.63it/s]
                                                        
{'loss': 0.2807, 'grad_norm': 0.6161587238311768, 'learning_rate': 6.866550073085261e-06, 'epoch': 1.32}

 44%|████▍     | 5865/13317 [1:00:26<1:16:22,  1.63it/s]
 44%|████▍     | 5866/13317 [1:00:27<1:16:18,  1.63it/s]
                                                        
{'loss': 0.2029, 'grad_norm': 0.45026695728302, 'learning_rate': 6.865334123948693e-06, 'epoch': 1.32}

 44%|████▍     | 5866/13317 [1:00:27<1:16:18,  1.63it/s]
 44%|████▍     | 5867/13317 [1:00:27<1:16:14,  1.63it/s]
                                                        
{'loss': 0.2148, 'grad_norm': 0.5775617957115173, 'learning_rate': 6.864118046643922e-06, 'epoch': 1.32}

 44%|████▍     | 5867/13317 [1:00:27<1:16:14,  1.63it/s]
 44%|████▍     | 5868/13317 [1:00:28<1:16:20,  1.63it/s]
                                                        
{'loss': 0.2315, 'grad_norm': 0.5631265044212341, 'learning_rate': 6.862901841254502e-06, 'epoch': 1.32}

 44%|████▍     | 5868/13317 [1:00:28<1:16:20,  1.63it/s]
 44%|████▍     | 5869/13317 [1:00:28<1:16:16,  1.63it/s]
                                                        
{'loss': 0.2075, 'grad_norm': 0.6705900430679321, 'learning_rate': 6.861685507864003e-06, 'epoch': 1.32}

 44%|████▍     | 5869/13317 [1:00:28<1:16:16,  1.63it/s]
 44%|████▍     | 5870/13317 [1:00:29<1:16:12,  1.63it/s]
                                                        
{'loss': 0.1974, 'grad_norm': 0.5494195222854614, 'learning_rate': 6.860469046555996e-06, 'epoch': 1.32}

 44%|████▍     | 5870/13317 [1:00:29<1:16:12,  1.63it/s]
 44%|████▍     | 5871/13317 [1:00:30<1:16:12,  1.63it/s]
                                                        
{'loss': 0.2372, 'grad_norm': 0.5829907059669495, 'learning_rate': 6.859252457414067e-06, 'epoch': 1.32}

 44%|████▍     | 5871/13317 [1:00:30<1:16:12,  1.63it/s]
 44%|████▍     | 5872/13317 [1:00:30<1:16:09,  1.63it/s]
                                                        
{'loss': 0.2513, 'grad_norm': 0.5756157636642456, 'learning_rate': 6.858035740521808e-06, 'epoch': 1.32}

 44%|████▍     | 5872/13317 [1:00:30<1:16:09,  1.63it/s]
 44%|████▍     | 5873/13317 [1:00:31<1:16:13,  1.63it/s]
                                                        
{'loss': 0.1939, 'grad_norm': 0.5280535221099854, 'learning_rate': 6.85681889596282e-06, 'epoch': 1.32}

 44%|████▍     | 5873/13317 [1:00:31<1:16:13,  1.63it/s]
 44%|████▍     | 5874/13317 [1:00:31<1:16:10,  1.63it/s]
                                                        
{'loss': 0.1555, 'grad_norm': 0.3857589364051819, 'learning_rate': 6.855601923820715e-06, 'epoch': 1.32}

 44%|████▍     | 5874/13317 [1:00:32<1:16:10,  1.63it/s]
 44%|████▍     | 5875/13317 [1:00:32<1:16:08,  1.63it/s]
                                                        
{'loss': 0.2459, 'grad_norm': 0.5939980149269104, 'learning_rate': 6.854384824179108e-06, 'epoch': 1.32}

 44%|████▍     | 5875/13317 [1:00:32<1:16:08,  1.63it/s]
 44%|████▍     | 5876/13317 [1:00:33<1:16:05,  1.63it/s]
                                                        
{'loss': 0.224, 'grad_norm': 0.5446463823318481, 'learning_rate': 6.8531675971216305e-06, 'epoch': 1.32}

 44%|████▍     | 5876/13317 [1:00:33<1:16:05,  1.63it/s]
 44%|████▍     | 5877/13317 [1:00:33<1:16:08,  1.63it/s]
                                                        
{'loss': 0.1405, 'grad_norm': 0.412882924079895, 'learning_rate': 6.851950242731917e-06, 'epoch': 1.32}

 44%|████▍     | 5877/13317 [1:00:33<1:16:08,  1.63it/s]
 44%|████▍     | 5878/13317 [1:00:34<1:16:12,  1.63it/s]
                                                        
{'loss': 0.1753, 'grad_norm': 0.45104849338531494, 'learning_rate': 6.850732761093613e-06, 'epoch': 1.32}

 44%|████▍     | 5878/13317 [1:00:34<1:16:12,  1.63it/s]
 44%|████▍     | 5879/13317 [1:00:34<1:16:10,  1.63it/s]
                                                        
{'loss': 0.234, 'grad_norm': 0.5092732906341553, 'learning_rate': 6.849515152290373e-06, 'epoch': 1.32}

 44%|████▍     | 5879/13317 [1:00:35<1:16:10,  1.63it/s]
 44%|████▍     | 5880/13317 [1:00:35<1:16:06,  1.63it/s]
                                                        
{'loss': 0.2678, 'grad_norm': 0.6163591742515564, 'learning_rate': 6.848297416405859e-06, 'epoch': 1.32}

 44%|████▍     | 5880/13317 [1:00:35<1:16:06,  1.63it/s]
 44%|████▍     | 5881/13317 [1:00:36<1:16:04,  1.63it/s]
                                                        
{'loss': 0.2235, 'grad_norm': 0.5260871052742004, 'learning_rate': 6.847079553523743e-06, 'epoch': 1.32}

 44%|████▍     | 5881/13317 [1:00:36<1:16:04,  1.63it/s]
 44%|████▍     | 5882/13317 [1:00:36<1:16:00,  1.63it/s]
                                                        
{'loss': 0.2635, 'grad_norm': 0.6010370254516602, 'learning_rate': 6.8458615637277025e-06, 'epoch': 1.33}

 44%|████▍     | 5882/13317 [1:00:36<1:16:00,  1.63it/s]
 44%|████▍     | 5883/13317 [1:00:37<1:16:07,  1.63it/s]
                                                        
{'loss': 0.2026, 'grad_norm': 0.5640497803688049, 'learning_rate': 6.844643447101431e-06, 'epoch': 1.33}

 44%|████▍     | 5883/13317 [1:00:37<1:16:07,  1.63it/s]
 44%|████▍     | 5884/13317 [1:00:38<1:16:03,  1.63it/s]
                                                        
{'loss': 0.207, 'grad_norm': 0.48955756425857544, 'learning_rate': 6.843425203728622e-06, 'epoch': 1.33}

 44%|████▍     | 5884/13317 [1:00:38<1:16:03,  1.63it/s]
 44%|████▍     | 5885/13317 [1:00:38<1:16:01,  1.63it/s]
                                                        
{'loss': 0.1707, 'grad_norm': 0.4887620210647583, 'learning_rate': 6.842206833692982e-06, 'epoch': 1.33}

 44%|████▍     | 5885/13317 [1:00:38<1:16:01,  1.63it/s]
 44%|████▍     | 5886/13317 [1:00:39<1:15:59,  1.63it/s]
                                                        
{'loss': 0.243, 'grad_norm': 0.49797189235687256, 'learning_rate': 6.840988337078228e-06, 'epoch': 1.33}

 44%|████▍     | 5886/13317 [1:00:39<1:15:59,  1.63it/s]
 44%|████▍     | 5887/13317 [1:00:39<1:15:59,  1.63it/s]
                                                        
{'loss': 0.1613, 'grad_norm': 0.4172806143760681, 'learning_rate': 6.839769713968082e-06, 'epoch': 1.33}

 44%|████▍     | 5887/13317 [1:00:40<1:15:59,  1.63it/s]
 44%|████▍     | 5888/13317 [1:00:40<1:16:07,  1.63it/s]
                                                        
{'loss': 0.2249, 'grad_norm': 0.5539928078651428, 'learning_rate': 6.838550964446277e-06, 'epoch': 1.33}

 44%|████▍     | 5888/13317 [1:00:40<1:16:07,  1.63it/s]
 44%|████▍     | 5889/13317 [1:00:41<1:16:01,  1.63it/s]
                                                        
{'loss': 0.1738, 'grad_norm': 0.4736337661743164, 'learning_rate': 6.8373320885965535e-06, 'epoch': 1.33}

 44%|████▍     | 5889/13317 [1:00:41<1:16:01,  1.63it/s]
 44%|████▍     | 5890/13317 [1:00:41<1:15:58,  1.63it/s]
                                                        
{'loss': 0.1618, 'grad_norm': 0.4625905752182007, 'learning_rate': 6.836113086502662e-06, 'epoch': 1.33}

 44%|████▍     | 5890/13317 [1:00:41<1:15:58,  1.63it/s]
 44%|████▍     | 5891/13317 [1:00:42<1:15:59,  1.63it/s]
                                                        
{'loss': 0.1539, 'grad_norm': 0.4520246088504791, 'learning_rate': 6.834893958248361e-06, 'epoch': 1.33}

 44%|████▍     | 5891/13317 [1:00:42<1:15:59,  1.63it/s]
 44%|████▍     | 5892/13317 [1:00:42<1:15:57,  1.63it/s]
                                                        
{'loss': 0.1422, 'grad_norm': 0.3931047022342682, 'learning_rate': 6.833674703917417e-06, 'epoch': 1.33}

 44%|████▍     | 5892/13317 [1:00:43<1:15:57,  1.63it/s]
 44%|████▍     | 5893/13317 [1:00:43<1:16:00,  1.63it/s]
                                                        
{'loss': 0.2074, 'grad_norm': 0.4883286654949188, 'learning_rate': 6.832455323593604e-06, 'epoch': 1.33}

 44%|████▍     | 5893/13317 [1:00:43<1:16:00,  1.63it/s]
 44%|████▍     | 5894/13317 [1:00:44<1:15:58,  1.63it/s]
                                                        
{'loss': 0.1837, 'grad_norm': 0.4729725420475006, 'learning_rate': 6.831235817360709e-06, 'epoch': 1.33}

 44%|████▍     | 5894/13317 [1:00:44<1:15:58,  1.63it/s]
 44%|████▍     | 5895/13317 [1:00:44<1:15:55,  1.63it/s]
                                                        
{'loss': 0.216, 'grad_norm': 0.5433443784713745, 'learning_rate': 6.830016185302524e-06, 'epoch': 1.33}

 44%|████▍     | 5895/13317 [1:00:44<1:15:55,  1.63it/s]
 44%|████▍     | 5896/13317 [1:00:45<1:15:57,  1.63it/s]
                                                        
{'loss': 0.2142, 'grad_norm': 0.5383272767066956, 'learning_rate': 6.828796427502851e-06, 'epoch': 1.33}

 44%|████▍     | 5896/13317 [1:00:45<1:15:57,  1.63it/s]
 44%|████▍     | 5897/13317 [1:00:46<1:15:53,  1.63it/s]
                                                        
{'loss': 0.2348, 'grad_norm': 0.6299175024032593, 'learning_rate': 6.8275765440454985e-06, 'epoch': 1.33}

 44%|████▍     | 5897/13317 [1:00:46<1:15:53,  1.63it/s]
 44%|████▍     | 5898/13317 [1:00:46<1:15:51,  1.63it/s]
                                                        
{'loss': 0.1943, 'grad_norm': 0.5002802014350891, 'learning_rate': 6.826356535014287e-06, 'epoch': 1.33}

 44%|████▍     | 5898/13317 [1:00:46<1:15:51,  1.63it/s]
 44%|████▍     | 5899/13317 [1:00:47<1:15:53,  1.63it/s]
                                                        
{'loss': 0.2088, 'grad_norm': 0.5810014605522156, 'learning_rate': 6.825136400493044e-06, 'epoch': 1.33}

 44%|████▍     | 5899/13317 [1:00:47<1:15:53,  1.63it/s]
 44%|████▍     | 5900/13317 [1:00:47<1:15:53,  1.63it/s]
                                                        
{'loss': 0.245, 'grad_norm': 0.5598465800285339, 'learning_rate': 6.823916140565605e-06, 'epoch': 1.33}

 44%|████▍     | 5900/13317 [1:00:48<1:15:53,  1.63it/s]
 44%|████▍     | 5901/13317 [1:00:48<1:15:49,  1.63it/s]
                                                        
{'loss': 0.5523, 'grad_norm': 0.6137849688529968, 'learning_rate': 6.822695755315817e-06, 'epoch': 1.33}

 44%|████▍     | 5901/13317 [1:00:48<1:15:49,  1.63it/s]
 44%|████▍     | 5902/13317 [1:00:49<1:15:51,  1.63it/s]
                                                        
{'loss': 0.2147, 'grad_norm': 0.5612524151802063, 'learning_rate': 6.821475244827529e-06, 'epoch': 1.33}

 44%|████▍     | 5902/13317 [1:00:49<1:15:51,  1.63it/s]
 44%|████▍     | 5903/13317 [1:00:49<1:15:48,  1.63it/s]
                                                        
{'loss': 0.2079, 'grad_norm': 0.5647463202476501, 'learning_rate': 6.820254609184608e-06, 'epoch': 1.33}

 44%|████▍     | 5903/13317 [1:00:49<1:15:48,  1.63it/s]
 44%|████▍     | 5904/13317 [1:00:50<1:15:47,  1.63it/s]
                                                        
{'loss': 0.2111, 'grad_norm': 0.5020962953567505, 'learning_rate': 6.819033848470919e-06, 'epoch': 1.33}

 44%|████▍     | 5904/13317 [1:00:50<1:15:47,  1.63it/s]
 44%|████▍     | 5905/13317 [1:00:50<1:15:49,  1.63it/s]
                                                        
{'loss': 0.2076, 'grad_norm': 0.49955716729164124, 'learning_rate': 6.817812962770348e-06, 'epoch': 1.33}

 44%|████▍     | 5905/13317 [1:00:51<1:15:49,  1.63it/s]
 44%|████▍     | 5906/13317 [1:00:51<1:15:50,  1.63it/s]
                                                        
{'loss': 0.198, 'grad_norm': 0.5116398334503174, 'learning_rate': 6.816591952166778e-06, 'epoch': 1.33}

 44%|████▍     | 5906/13317 [1:00:51<1:15:50,  1.63it/s]
 44%|████▍     | 5907/13317 [1:00:52<1:15:49,  1.63it/s]
                                                        
{'loss': 0.2604, 'grad_norm': 0.5719709396362305, 'learning_rate': 6.815370816744106e-06, 'epoch': 1.33}

 44%|████▍     | 5907/13317 [1:00:52<1:15:49,  1.63it/s]
 44%|████▍     | 5908/13317 [1:00:52<1:15:47,  1.63it/s]
                                                        
{'loss': 0.2162, 'grad_norm': 0.5469120740890503, 'learning_rate': 6.814149556586238e-06, 'epoch': 1.33}

 44%|████▍     | 5908/13317 [1:00:52<1:15:47,  1.63it/s]
 44%|████▍     | 5909/13317 [1:00:53<1:15:47,  1.63it/s]
                                                        
{'loss': 0.1799, 'grad_norm': 0.47919341921806335, 'learning_rate': 6.812928171777086e-06, 'epoch': 1.33}

 44%|████▍     | 5909/13317 [1:00:53<1:15:47,  1.63it/s]
 44%|████▍     | 5910/13317 [1:00:54<1:15:44,  1.63it/s]
                                                        
{'loss': 0.1574, 'grad_norm': 0.4187549650669098, 'learning_rate': 6.811706662400575e-06, 'epoch': 1.33}

 44%|████▍     | 5910/13317 [1:00:54<1:15:44,  1.63it/s]
 44%|████▍     | 5911/13317 [1:00:54<1:15:46,  1.63it/s]
                                                        
{'loss': 0.2644, 'grad_norm': 0.6692149043083191, 'learning_rate': 6.8104850285406295e-06, 'epoch': 1.33}

 44%|████▍     | 5911/13317 [1:00:54<1:15:46,  1.63it/s]
 44%|████▍     | 5912/13317 [1:00:55<1:15:45,  1.63it/s]
                                                        
{'loss': 0.1887, 'grad_norm': 0.5135143995285034, 'learning_rate': 6.8092632702811964e-06, 'epoch': 1.33}

 44%|████▍     | 5912/13317 [1:00:55<1:15:45,  1.63it/s]
 44%|████▍     | 5913/13317 [1:00:55<1:15:44,  1.63it/s]
                                                        
{'loss': 0.1929, 'grad_norm': 0.47011902928352356, 'learning_rate': 6.808041387706219e-06, 'epoch': 1.33}

 44%|████▍     | 5913/13317 [1:00:55<1:15:44,  1.63it/s]
 44%|████▍     | 5914/13317 [1:00:56<1:15:46,  1.63it/s]
                                                        
{'loss': 0.146, 'grad_norm': 0.41465505957603455, 'learning_rate': 6.806819380899654e-06, 'epoch': 1.33}

 44%|████▍     | 5914/13317 [1:00:56<1:15:46,  1.63it/s]
 44%|████▍     | 5915/13317 [1:00:57<1:15:45,  1.63it/s]
                                                        
{'loss': 0.2337, 'grad_norm': 0.5866001844406128, 'learning_rate': 6.805597249945467e-06, 'epoch': 1.33}

 44%|████▍     | 5915/13317 [1:00:57<1:15:45,  1.63it/s]
 44%|████▍     | 5916/13317 [1:00:57<1:15:45,  1.63it/s]
                                                        
{'loss': 0.1693, 'grad_norm': 0.46328866481781006, 'learning_rate': 6.804374994927631e-06, 'epoch': 1.33}

 44%|████▍     | 5916/13317 [1:00:57<1:15:45,  1.63it/s]
 44%|████▍     | 5917/13317 [1:00:58<1:15:42,  1.63it/s]
                                                        
{'loss': 0.1909, 'grad_norm': 0.49140822887420654, 'learning_rate': 6.803152615930129e-06, 'epoch': 1.33}

 44%|████▍     | 5917/13317 [1:00:58<1:15:42,  1.63it/s]
 44%|████▍     | 5918/13317 [1:00:58<1:15:42,  1.63it/s]
                                                        
{'loss': 0.2347, 'grad_norm': 0.6610428690910339, 'learning_rate': 6.801930113036948e-06, 'epoch': 1.33}

 44%|████▍     | 5918/13317 [1:00:59<1:15:42,  1.63it/s]
 44%|████▍     | 5919/13317 [1:00:59<1:15:44,  1.63it/s]
                                                        
{'loss': 0.2032, 'grad_norm': 0.5199065208435059, 'learning_rate': 6.80070748633209e-06, 'epoch': 1.33}

 44%|████▍     | 5919/13317 [1:00:59<1:15:44,  1.63it/s]
 44%|████▍     | 5920/13317 [1:01:00<1:15:41,  1.63it/s]
                                                        
{'loss': 0.2243, 'grad_norm': 0.5439605712890625, 'learning_rate': 6.799484735899562e-06, 'epoch': 1.33}

 44%|████▍     | 5920/13317 [1:01:00<1:15:41,  1.63it/s]
 44%|████▍     | 5921/13317 [1:01:00<1:15:40,  1.63it/s]
                                                        
{'loss': 0.1855, 'grad_norm': 0.4839344620704651, 'learning_rate': 6.798261861823379e-06, 'epoch': 1.33}

 44%|████▍     | 5921/13317 [1:01:00<1:15:40,  1.63it/s]
 44%|████▍     | 5922/13317 [1:01:01<1:15:34,  1.63it/s]
                                                        
{'loss': 0.1906, 'grad_norm': 0.5329304933547974, 'learning_rate': 6.797038864187564e-06, 'epoch': 1.33}

 44%|████▍     | 5922/13317 [1:01:01<1:15:34,  1.63it/s]
 44%|████▍     | 5923/13317 [1:01:01<1:15:33,  1.63it/s]
                                                        
{'loss': 0.1804, 'grad_norm': 0.46459704637527466, 'learning_rate': 6.795815743076155e-06, 'epoch': 1.33}

 44%|████▍     | 5923/13317 [1:01:02<1:15:33,  1.63it/s]
 44%|████▍     | 5924/13317 [1:01:02<1:15:33,  1.63it/s]
                                                        
{'loss': 0.1764, 'grad_norm': 0.4968605637550354, 'learning_rate': 6.794592498573187e-06, 'epoch': 1.33}

 44%|████▍     | 5924/13317 [1:01:02<1:15:33,  1.63it/s]
 44%|████▍     | 5925/13317 [1:01:03<1:15:34,  1.63it/s]
                                                        
{'loss': 0.228, 'grad_norm': 0.5658459663391113, 'learning_rate': 6.793369130762715e-06, 'epoch': 1.33}

 44%|████▍     | 5925/13317 [1:01:03<1:15:34,  1.63it/s]
 44%|████▍     | 5926/13317 [1:01:03<1:15:33,  1.63it/s]
                                                        
{'loss': 0.1733, 'grad_norm': 0.49877408146858215, 'learning_rate': 6.7921456397287946e-06, 'epoch': 1.33}

 44%|████▍     | 5926/13317 [1:01:03<1:15:33,  1.63it/s]
 45%|████▍     | 5927/13317 [1:01:04<1:15:32,  1.63it/s]
                                                        
{'loss': 0.2463, 'grad_norm': 0.5562933683395386, 'learning_rate': 6.790922025555493e-06, 'epoch': 1.34}

 45%|████▍     | 5927/13317 [1:01:04<1:15:32,  1.63it/s]
 45%|████▍     | 5928/13317 [1:01:05<1:15:33,  1.63it/s]
                                                        
{'loss': 0.1712, 'grad_norm': 0.46784868836402893, 'learning_rate': 6.789698288326886e-06, 'epoch': 1.34}

 45%|████▍     | 5928/13317 [1:01:05<1:15:33,  1.63it/s]
 45%|████▍     | 5929/13317 [1:01:05<1:15:30,  1.63it/s]
                                                        
{'loss': 0.1789, 'grad_norm': 0.5054362416267395, 'learning_rate': 6.788474428127056e-06, 'epoch': 1.34}

 45%|████▍     | 5929/13317 [1:01:05<1:15:30,  1.63it/s]
 45%|████▍     | 5930/13317 [1:01:06<1:15:31,  1.63it/s]
                                                        
{'loss': 0.2368, 'grad_norm': 0.5968835949897766, 'learning_rate': 6.787250445040098e-06, 'epoch': 1.34}

 45%|████▍     | 5930/13317 [1:01:06<1:15:31,  1.63it/s]
 45%|████▍     | 5931/13317 [1:01:06<1:15:31,  1.63it/s]
                                                        
{'loss': 0.312, 'grad_norm': 0.6435225605964661, 'learning_rate': 6.786026339150109e-06, 'epoch': 1.34}

 45%|████▍     | 5931/13317 [1:01:07<1:15:31,  1.63it/s]
 45%|████▍     | 5932/13317 [1:01:07<1:15:27,  1.63it/s]
                                                        
{'loss': 0.2301, 'grad_norm': 0.5486071109771729, 'learning_rate': 6.784802110541201e-06, 'epoch': 1.34}

 45%|████▍     | 5932/13317 [1:01:07<1:15:27,  1.63it/s]
 45%|████▍     | 5933/13317 [1:01:08<1:15:27,  1.63it/s]
                                                        
{'loss': 0.1636, 'grad_norm': 0.4996718168258667, 'learning_rate': 6.783577759297489e-06, 'epoch': 1.34}

 45%|████▍     | 5933/13317 [1:01:08<1:15:27,  1.63it/s]
 45%|████▍     | 5934/13317 [1:01:08<1:15:27,  1.63it/s]
                                                        
{'loss': 0.2559, 'grad_norm': 0.5738811492919922, 'learning_rate': 6.782353285503101e-06, 'epoch': 1.34}

 45%|████▍     | 5934/13317 [1:01:08<1:15:27,  1.63it/s]
 45%|████▍     | 5935/13317 [1:01:09<1:15:28,  1.63it/s]
                                                        
{'loss': 0.178, 'grad_norm': 0.5493126511573792, 'learning_rate': 6.781128689242169e-06, 'epoch': 1.34}

 45%|████▍     | 5935/13317 [1:01:09<1:15:28,  1.63it/s]
 45%|████▍     | 5936/13317 [1:01:09<1:15:29,  1.63it/s]
                                                        
{'loss': 0.2056, 'grad_norm': 0.5193482637405396, 'learning_rate': 6.779903970598839e-06, 'epoch': 1.34}

 45%|████▍     | 5936/13317 [1:01:10<1:15:29,  1.63it/s]
 45%|████▍     | 5937/13317 [1:01:10<1:15:27,  1.63it/s]
                                                        
{'loss': 0.2117, 'grad_norm': 0.5625088214874268, 'learning_rate': 6.7786791296572594e-06, 'epoch': 1.34}

 45%|████▍     | 5937/13317 [1:01:10<1:15:27,  1.63it/s]
 45%|████▍     | 5938/13317 [1:01:11<1:15:27,  1.63it/s]
                                                        
{'loss': 0.1953, 'grad_norm': 0.5498471260070801, 'learning_rate': 6.77745416650159e-06, 'epoch': 1.34}

 45%|████▍     | 5938/13317 [1:01:11<1:15:27,  1.63it/s]
 45%|████▍     | 5939/13317 [1:01:11<1:15:25,  1.63it/s]
                                                        
{'loss': 0.2151, 'grad_norm': 0.5480607748031616, 'learning_rate': 6.776229081216001e-06, 'epoch': 1.34}

 45%|████▍     | 5939/13317 [1:01:11<1:15:25,  1.63it/s]
 45%|████▍     | 5940/13317 [1:01:12<1:15:26,  1.63it/s]
                                                        
{'loss': 0.1974, 'grad_norm': 0.5605814456939697, 'learning_rate': 6.775003873884667e-06, 'epoch': 1.34}

 45%|████▍     | 5940/13317 [1:01:12<1:15:26,  1.63it/s]
 45%|████▍     | 5941/13317 [1:01:13<1:15:25,  1.63it/s]
                                                        
{'loss': 0.195, 'grad_norm': 0.5438852906227112, 'learning_rate': 6.7737785445917715e-06, 'epoch': 1.34}

 45%|████▍     | 5941/13317 [1:01:13<1:15:25,  1.63it/s]
 45%|████▍     | 5942/13317 [1:01:13<1:15:24,  1.63it/s]
                                                        
{'loss': 0.1895, 'grad_norm': 0.4757399559020996, 'learning_rate': 6.77255309342151e-06, 'epoch': 1.34}

 45%|████▍     | 5942/13317 [1:01:13<1:15:24,  1.63it/s]
 45%|████▍     | 5943/13317 [1:01:14<1:15:23,  1.63it/s]
                                                        
{'loss': 0.1676, 'grad_norm': 0.4548255503177643, 'learning_rate': 6.771327520458082e-06, 'epoch': 1.34}

 45%|████▍     | 5943/13317 [1:01:14<1:15:23,  1.63it/s]
 45%|████▍     | 5944/13317 [1:01:14<1:15:21,  1.63it/s]
                                                        
{'loss': 0.1939, 'grad_norm': 0.49927058815956116, 'learning_rate': 6.770101825785698e-06, 'epoch': 1.34}

 45%|████▍     | 5944/13317 [1:01:15<1:15:21,  1.63it/s]
 45%|████▍     | 5945/13317 [1:01:15<1:15:20,  1.63it/s]
                                                        
{'loss': 0.1948, 'grad_norm': 0.5106679201126099, 'learning_rate': 6.768876009488579e-06, 'epoch': 1.34}

 45%|████▍     | 5945/13317 [1:01:15<1:15:20,  1.63it/s]
 45%|████▍     | 5946/13317 [1:01:16<1:15:21,  1.63it/s]
                                                        
{'loss': 0.2064, 'grad_norm': 0.4887790083885193, 'learning_rate': 6.767650071650946e-06, 'epoch': 1.34}

 45%|████▍     | 5946/13317 [1:01:16<1:15:21,  1.63it/s]
 45%|████▍     | 5947/13317 [1:01:16<1:15:20,  1.63it/s]
                                                        
{'loss': 0.2103, 'grad_norm': 0.5555021166801453, 'learning_rate': 6.766424012357038e-06, 'epoch': 1.34}

 45%|████▍     | 5947/13317 [1:01:16<1:15:20,  1.63it/s]
 45%|████▍     | 5948/13317 [1:01:17<1:15:21,  1.63it/s]
                                                        
{'loss': 0.1982, 'grad_norm': 0.5024904608726501, 'learning_rate': 6.765197831691097e-06, 'epoch': 1.34}

 45%|████▍     | 5948/13317 [1:01:17<1:15:21,  1.63it/s]
 45%|████▍     | 5949/13317 [1:01:17<1:15:21,  1.63it/s]
                                                        
{'loss': 0.1734, 'grad_norm': 0.4760306775569916, 'learning_rate': 6.763971529737374e-06, 'epoch': 1.34}

 45%|████▍     | 5949/13317 [1:01:18<1:15:21,  1.63it/s]
 45%|████▍     | 5950/13317 [1:01:18<1:15:26,  1.63it/s]
                                                        
{'loss': 0.162, 'grad_norm': 0.4174610376358032, 'learning_rate': 6.762745106580131e-06, 'epoch': 1.34}

 45%|████▍     | 5950/13317 [1:01:18<1:15:26,  1.63it/s]
 45%|████▍     | 5951/13317 [1:01:19<1:15:27,  1.63it/s]
                                                        
{'loss': 0.215, 'grad_norm': 0.5870687365531921, 'learning_rate': 6.761518562303634e-06, 'epoch': 1.34}

 45%|████▍     | 5951/13317 [1:01:19<1:15:27,  1.63it/s]
 45%|████▍     | 5952/13317 [1:01:19<1:15:25,  1.63it/s]
                                                        
{'loss': 0.1956, 'grad_norm': 0.5011380314826965, 'learning_rate': 6.760291896992162e-06, 'epoch': 1.34}

 45%|████▍     | 5952/13317 [1:01:19<1:15:25,  1.63it/s]
 45%|████▍     | 5953/13317 [1:01:20<1:15:19,  1.63it/s]
                                                        
{'loss': 0.1554, 'grad_norm': 0.4828644096851349, 'learning_rate': 6.759065110729997e-06, 'epoch': 1.34}

 45%|████▍     | 5953/13317 [1:01:20<1:15:19,  1.63it/s]
 45%|████▍     | 5954/13317 [1:01:21<1:15:19,  1.63it/s]
                                                        
{'loss': 0.2114, 'grad_norm': 0.586219072341919, 'learning_rate': 6.757838203601435e-06, 'epoch': 1.34}

 45%|████▍     | 5954/13317 [1:01:21<1:15:19,  1.63it/s]
 45%|████▍     | 5955/13317 [1:01:21<1:15:20,  1.63it/s]
                                                        
{'loss': 0.2403, 'grad_norm': 0.6119095087051392, 'learning_rate': 6.756611175690773e-06, 'epoch': 1.34}

 45%|████▍     | 5955/13317 [1:01:21<1:15:20,  1.63it/s]
 45%|████▍     | 5956/13317 [1:01:22<1:15:19,  1.63it/s]
                                                        
{'loss': 0.1951, 'grad_norm': 0.48854902386665344, 'learning_rate': 6.755384027082327e-06, 'epoch': 1.34}

 45%|████▍     | 5956/13317 [1:01:22<1:15:19,  1.63it/s]
 45%|████▍     | 5957/13317 [1:01:22<1:15:18,  1.63it/s]
                                                        
{'loss': 0.1633, 'grad_norm': 0.4403230547904968, 'learning_rate': 6.754156757860412e-06, 'epoch': 1.34}

 45%|████▍     | 5957/13317 [1:01:22<1:15:18,  1.63it/s]
 45%|████▍     | 5958/13317 [1:01:23<1:15:22,  1.63it/s]
                                                        
{'loss': 0.2284, 'grad_norm': 0.6418771147727966, 'learning_rate': 6.752929368109352e-06, 'epoch': 1.34}

 45%|████▍     | 5958/13317 [1:01:23<1:15:22,  1.63it/s]
 45%|████▍     | 5959/13317 [1:01:24<1:15:17,  1.63it/s]
                                                        
{'loss': 0.1568, 'grad_norm': 0.4517025351524353, 'learning_rate': 6.7517018579134864e-06, 'epoch': 1.34}

 45%|████▍     | 5959/13317 [1:01:24<1:15:17,  1.63it/s]
 45%|████▍     | 5960/13317 [1:01:24<1:15:19,  1.63it/s]
                                                        
{'loss': 0.1712, 'grad_norm': 0.5831912755966187, 'learning_rate': 6.7504742273571555e-06, 'epoch': 1.34}

 45%|████▍     | 5960/13317 [1:01:24<1:15:19,  1.63it/s]
 45%|████▍     | 5961/13317 [1:01:25<1:15:16,  1.63it/s]
                                                        
{'loss': 0.2427, 'grad_norm': 0.5849778056144714, 'learning_rate': 6.74924647652471e-06, 'epoch': 1.34}

 45%|████▍     | 5961/13317 [1:01:25<1:15:16,  1.63it/s]
 45%|████▍     | 5962/13317 [1:01:25<1:15:14,  1.63it/s]
                                                        
{'loss': 0.1491, 'grad_norm': 0.3964293599128723, 'learning_rate': 6.74801860550051e-06, 'epoch': 1.34}

 45%|████▍     | 5962/13317 [1:01:26<1:15:14,  1.63it/s]
 45%|████▍     | 5963/13317 [1:01:26<1:15:23,  1.63it/s]
                                                        
{'loss': 0.179, 'grad_norm': 0.5296375155448914, 'learning_rate': 6.746790614368925e-06, 'epoch': 1.34}

 45%|████▍     | 5963/13317 [1:01:26<1:15:23,  1.63it/s]
 45%|████▍     | 5964/13317 [1:01:27<1:15:18,  1.63it/s]
                                                        
{'loss': 0.1614, 'grad_norm': 0.46145743131637573, 'learning_rate': 6.745562503214329e-06, 'epoch': 1.34}

 45%|████▍     | 5964/13317 [1:01:27<1:15:18,  1.63it/s]
 45%|████▍     | 5965/13317 [1:01:27<1:15:18,  1.63it/s]
                                                        
{'loss': 0.1825, 'grad_norm': 0.5610266327857971, 'learning_rate': 6.744334272121104e-06, 'epoch': 1.34}

 45%|████▍     | 5965/13317 [1:01:27<1:15:18,  1.63it/s]
 45%|████▍     | 5966/13317 [1:01:28<1:15:16,  1.63it/s]
                                                        
{'loss': 0.2011, 'grad_norm': 0.488066703081131, 'learning_rate': 6.743105921173649e-06, 'epoch': 1.34}

 45%|████▍     | 5966/13317 [1:01:28<1:15:16,  1.63it/s]
 45%|████▍     | 5967/13317 [1:01:28<1:15:12,  1.63it/s]
                                                        
{'loss': 0.2651, 'grad_norm': 0.6157221794128418, 'learning_rate': 6.74187745045636e-06, 'epoch': 1.34}

 45%|████▍     | 5967/13317 [1:01:29<1:15:12,  1.63it/s]
 45%|████▍     | 5968/13317 [1:01:29<1:15:20,  1.63it/s]
                                                        
{'loss': 0.1654, 'grad_norm': 0.5184832215309143, 'learning_rate': 6.740648860053646e-06, 'epoch': 1.34}

 45%|████▍     | 5968/13317 [1:01:29<1:15:20,  1.63it/s]
 45%|████▍     | 5969/13317 [1:01:30<1:15:15,  1.63it/s]
                                                        
{'loss': 0.1953, 'grad_norm': 0.47133657336235046, 'learning_rate': 6.739420150049925e-06, 'epoch': 1.34}

 45%|████▍     | 5969/13317 [1:01:30<1:15:15,  1.63it/s]
 45%|████▍     | 5970/13317 [1:01:30<1:15:13,  1.63it/s]
                                                        
{'loss': 0.1979, 'grad_norm': 0.4841718375682831, 'learning_rate': 6.738191320529623e-06, 'epoch': 1.34}

 45%|████▍     | 5970/13317 [1:01:30<1:15:13,  1.63it/s]
 45%|████▍     | 5971/13317 [1:01:31<1:15:09,  1.63it/s]
                                                        
{'loss': 0.1922, 'grad_norm': 0.4849417209625244, 'learning_rate': 6.736962371577173e-06, 'epoch': 1.35}

 45%|████▍     | 5971/13317 [1:01:31<1:15:09,  1.63it/s]
 45%|████▍     | 5972/13317 [1:01:32<1:15:07,  1.63it/s]
                                                        
{'loss': 0.1719, 'grad_norm': 0.50641268491745, 'learning_rate': 6.735733303277015e-06, 'epoch': 1.35}

 45%|████▍     | 5972/13317 [1:01:32<1:15:07,  1.63it/s]
 45%|████▍     | 5973/13317 [1:01:32<1:15:16,  1.63it/s]
                                                        
{'loss': 0.2642, 'grad_norm': 0.730030357837677, 'learning_rate': 6.7345041157136035e-06, 'epoch': 1.35}

 45%|████▍     | 5973/13317 [1:01:32<1:15:16,  1.63it/s]
 45%|████▍     | 5974/13317 [1:01:33<1:15:11,  1.63it/s]
                                                        
{'loss': 0.1846, 'grad_norm': 0.5069400668144226, 'learning_rate': 6.733274808971393e-06, 'epoch': 1.35}

 45%|████▍     | 5974/13317 [1:01:33<1:15:11,  1.63it/s]
 45%|████▍     | 5975/13317 [1:01:33<1:15:10,  1.63it/s]
                                                        
{'loss': 0.1912, 'grad_norm': 0.5063022375106812, 'learning_rate': 6.732045383134851e-06, 'epoch': 1.35}

 45%|████▍     | 5975/13317 [1:01:34<1:15:10,  1.63it/s]
 45%|████▍     | 5976/13317 [1:01:34<1:15:09,  1.63it/s]
                                                        
{'loss': 0.3241, 'grad_norm': 0.63201904296875, 'learning_rate': 6.730815838288451e-06, 'epoch': 1.35}

 45%|████▍     | 5976/13317 [1:01:34<1:15:09,  1.63it/s]
 45%|████▍     | 5977/13317 [1:01:35<1:15:08,  1.63it/s]
                                                        
{'loss': 0.2157, 'grad_norm': 0.49016550183296204, 'learning_rate': 6.729586174516678e-06, 'epoch': 1.35}

 45%|████▍     | 5977/13317 [1:01:35<1:15:08,  1.63it/s]
 45%|████▍     | 5978/13317 [1:01:35<1:15:09,  1.63it/s]
                                                        
{'loss': 0.1834, 'grad_norm': 0.5333173274993896, 'learning_rate': 6.728356391904021e-06, 'epoch': 1.35}

 45%|████▍     | 5978/13317 [1:01:35<1:15:09,  1.63it/s]
 45%|████▍     | 5979/13317 [1:01:36<1:15:07,  1.63it/s]
                                                        
{'loss': 0.2041, 'grad_norm': 0.5418321490287781, 'learning_rate': 6.727126490534981e-06, 'epoch': 1.35}

 45%|████▍     | 5979/13317 [1:01:36<1:15:07,  1.63it/s]
 45%|████▍     | 5980/13317 [1:01:36<1:15:05,  1.63it/s]
                                                        
{'loss': 0.2212, 'grad_norm': 0.5318682193756104, 'learning_rate': 6.725896470494063e-06, 'epoch': 1.35}

 45%|████▍     | 5980/13317 [1:01:37<1:15:05,  1.63it/s]
 45%|████▍     | 5981/13317 [1:01:37<1:15:03,  1.63it/s]
                                                        
{'loss': 0.2027, 'grad_norm': 0.5368024110794067, 'learning_rate': 6.724666331865784e-06, 'epoch': 1.35}

 45%|████▍     | 5981/13317 [1:01:37<1:15:03,  1.63it/s]
 45%|████▍     | 5982/13317 [1:01:38<1:15:03,  1.63it/s]
                                                        
{'loss': 0.2492, 'grad_norm': 0.6214893460273743, 'learning_rate': 6.723436074734665e-06, 'epoch': 1.35}

 45%|████▍     | 5982/13317 [1:01:38<1:15:03,  1.63it/s]
 45%|████▍     | 5983/13317 [1:01:38<1:15:24,  1.62it/s]
                                                        
{'loss': 0.1633, 'grad_norm': 0.4604928195476532, 'learning_rate': 6.722205699185241e-06, 'epoch': 1.35}

 45%|████▍     | 5983/13317 [1:01:38<1:15:24,  1.62it/s]
 45%|████▍     | 5984/13317 [1:01:39<1:15:16,  1.62it/s]
                                                        
{'loss': 0.2172, 'grad_norm': 0.5372785329818726, 'learning_rate': 6.720975205302051e-06, 'epoch': 1.35}

 45%|████▍     | 5984/13317 [1:01:39<1:15:16,  1.62it/s]
 45%|████▍     | 5985/13317 [1:01:40<1:15:10,  1.63it/s]
                                                        
{'loss': 0.21, 'grad_norm': 0.5843274593353271, 'learning_rate': 6.719744593169642e-06, 'epoch': 1.35}

 45%|████▍     | 5985/13317 [1:01:40<1:15:10,  1.63it/s]
 45%|████▍     | 5986/13317 [1:01:40<1:15:06,  1.63it/s]
                                                        
{'loss': 0.1799, 'grad_norm': 0.5381581783294678, 'learning_rate': 6.71851386287257e-06, 'epoch': 1.35}

 45%|████▍     | 5986/13317 [1:01:40<1:15:06,  1.63it/s]
 45%|████▍     | 5987/13317 [1:01:41<1:15:02,  1.63it/s]
                                                        
{'loss': 0.1937, 'grad_norm': 0.49498510360717773, 'learning_rate': 6.7172830144953995e-06, 'epoch': 1.35}

 45%|████▍     | 5987/13317 [1:01:41<1:15:02,  1.63it/s]
 45%|████▍     | 5988/13317 [1:01:41<1:15:05,  1.63it/s]
                                                        
{'loss': 0.1561, 'grad_norm': 0.48447391390800476, 'learning_rate': 6.716052048122704e-06, 'epoch': 1.35}

 45%|████▍     | 5988/13317 [1:01:42<1:15:05,  1.63it/s]
 45%|████▍     | 5989/13317 [1:01:42<1:15:07,  1.63it/s]
                                                        
{'loss': 0.1693, 'grad_norm': 0.40012696385383606, 'learning_rate': 6.714820963839063e-06, 'epoch': 1.35}

 45%|████▍     | 5989/13317 [1:01:42<1:15:07,  1.63it/s]
 45%|████▍     | 5990/13317 [1:01:43<1:15:02,  1.63it/s]
                                                        
{'loss': 0.1972, 'grad_norm': 0.4847389757633209, 'learning_rate': 6.713589761729063e-06, 'epoch': 1.35}

 45%|████▍     | 5990/13317 [1:01:43<1:15:02,  1.63it/s]
 45%|████▍     | 5991/13317 [1:01:43<1:15:00,  1.63it/s]
                                                        
{'loss': 0.192, 'grad_norm': 0.5359996557235718, 'learning_rate': 6.712358441877306e-06, 'epoch': 1.35}

 45%|████▍     | 5991/13317 [1:01:43<1:15:00,  1.63it/s]
 45%|████▍     | 5992/13317 [1:01:44<1:14:59,  1.63it/s]
                                                        
{'loss': 0.1989, 'grad_norm': 0.495774507522583, 'learning_rate': 6.71112700436839e-06, 'epoch': 1.35}

 45%|████▍     | 5992/13317 [1:01:44<1:14:59,  1.63it/s]
 45%|████▌     | 5993/13317 [1:01:44<1:15:22,  1.62it/s]
                                                        
{'loss': 0.2319, 'grad_norm': 0.5822529792785645, 'learning_rate': 6.709895449286933e-06, 'epoch': 1.35}

 45%|████▌     | 5993/13317 [1:01:45<1:15:22,  1.62it/s]
 45%|████▌     | 5994/13317 [1:01:45<1:15:17,  1.62it/s]
                                                        
{'loss': 0.2066, 'grad_norm': 0.5644437670707703, 'learning_rate': 6.708663776717552e-06, 'epoch': 1.35}

 45%|████▌     | 5994/13317 [1:01:45<1:15:17,  1.62it/s]
 45%|████▌     | 5995/13317 [1:01:46<1:15:11,  1.62it/s]
                                                        
{'loss': 0.1909, 'grad_norm': 0.6420048475265503, 'learning_rate': 6.707431986744881e-06, 'epoch': 1.35}

 45%|████▌     | 5995/13317 [1:01:46<1:15:11,  1.62it/s]
 45%|████▌     | 5996/13317 [1:01:46<1:15:03,  1.63it/s]
                                                        
{'loss': 0.2075, 'grad_norm': 0.5667726397514343, 'learning_rate': 6.706200079453552e-06, 'epoch': 1.35}

 45%|████▌     | 5996/13317 [1:01:46<1:15:03,  1.63it/s]
 45%|████▌     | 5997/13317 [1:01:47<1:15:00,  1.63it/s]
                                                        
{'loss': 0.165, 'grad_norm': 0.4361223876476288, 'learning_rate': 6.7049680549282114e-06, 'epoch': 1.35}

 45%|████▌     | 5997/13317 [1:01:47<1:15:00,  1.63it/s]
 45%|████▌     | 5998/13317 [1:01:48<1:15:04,  1.62it/s]
                                                        
{'loss': 0.2074, 'grad_norm': 0.5532160401344299, 'learning_rate': 6.7037359132535144e-06, 'epoch': 1.35}

 45%|████▌     | 5998/13317 [1:01:48<1:15:04,  1.62it/s]
 45%|████▌     | 5999/13317 [1:01:48<1:14:59,  1.63it/s]
                                                        
{'loss': 0.2446, 'grad_norm': 0.6721919775009155, 'learning_rate': 6.70250365451412e-06, 'epoch': 1.35}

 45%|████▌     | 5999/13317 [1:01:48<1:14:59,  1.63it/s]
 45%|████▌     | 6000/13317 [1:01:49<1:14:56,  1.63it/s]
                                                        
{'loss': 0.1906, 'grad_norm': 0.5108140707015991, 'learning_rate': 6.701271278794699e-06, 'epoch': 1.35}

 45%|████▌     | 6000/13317 [1:01:49<1:14:56,  1.63it/s]

  0%|          | 0/13 [00:00<?, ?it/s]

 15%|█▌        | 2/13 [00:00<00:00, 13.36it/s]

 31%|███       | 4/13 [00:00<00:01,  8.37it/s]

 38%|███▊      | 5/13 [00:00<00:01,  7.78it/s]

 46%|████▌     | 6/13 [00:00<00:00,  7.41it/s]

 54%|█████▍    | 7/13 [00:00<00:00,  7.17it/s]

 62%|██████▏   | 8/13 [00:01<00:00,  7.00it/s]

 69%|██████▉   | 9/13 [00:01<00:00,  6.90it/s]

 77%|███████▋  | 10/13 [00:01<00:00,  6.83it/s]

 85%|████████▍ | 11/13 [00:01<00:00,  6.78it/s]

 92%|█████████▏| 12/13 [00:01<00:00,  6.74it/s]

100%|██████████| 13/13 [00:01<00:00,  6.75it/s]
                                                        

                                               
{'eval_loss': 0.5849972367286682, 'eval_runtime': 1.9548, 'eval_samples_per_second': 102.314, 'eval_steps_per_second': 6.65, 'epoch': 1.35}

 45%|████▌     | 6000/13317 [1:01:51<1:14:56,  1.63it/s]

100%|██████████| 13/13 [00:01<00:00,  6.75it/s]

                                               
 45%|████▌     | 6001/13317 [1:01:51<2:26:35,  1.20s/it]
                                                        
{'loss': 0.1973, 'grad_norm': 0.5080292224884033, 'learning_rate': 6.700038786179927e-06, 'epoch': 1.35}

 45%|████▌     | 6001/13317 [1:01:51<2:26:35,  1.20s/it]
 45%|████▌     | 6002/13317 [1:01:52<2:05:02,  1.03s/it]
                                                        
{'loss': 0.2238, 'grad_norm': 0.5738369226455688, 'learning_rate': 6.69880617675449e-06, 'epoch': 1.35}

 45%|████▌     | 6002/13317 [1:01:52<2:05:02,  1.03s/it]
 45%|████▌     | 6003/13317 [1:01:53<1:50:03,  1.11it/s]
                                                        
{'loss': 0.2108, 'grad_norm': 0.5404796004295349, 'learning_rate': 6.697573450603081e-06, 'epoch': 1.35}

 45%|████▌     | 6003/13317 [1:01:53<1:50:03,  1.11it/s]
 45%|████▌     | 6004/13317 [1:01:53<1:39:34,  1.22it/s]
                                                        
{'loss': 0.2183, 'grad_norm': 0.5701006650924683, 'learning_rate': 6.696340607810402e-06, 'epoch': 1.35}

 45%|████▌     | 6004/13317 [1:01:53<1:39:34,  1.22it/s]
 45%|████▌     | 6005/13317 [1:01:54<1:32:07,  1.32it/s]
                                                        
{'loss': 0.2255, 'grad_norm': 0.5846425294876099, 'learning_rate': 6.695107648461162e-06, 'epoch': 1.35}

 45%|████▌     | 6005/13317 [1:01:54<1:32:07,  1.32it/s]
 45%|████▌     | 6006/13317 [1:01:54<1:26:53,  1.40it/s]
                                                        
{'loss': 0.2297, 'grad_norm': 0.7290287613868713, 'learning_rate': 6.69387457264008e-06, 'epoch': 1.35}

 45%|████▌     | 6006/13317 [1:01:55<1:26:53,  1.40it/s]
 45%|████▌     | 6007/13317 [1:01:55<1:23:14,  1.46it/s]
                                                        
{'loss': 0.1561, 'grad_norm': 0.46610280871391296, 'learning_rate': 6.692641380431879e-06, 'epoch': 1.35}

 45%|████▌     | 6007/13317 [1:01:55<1:23:14,  1.46it/s]
 45%|████▌     | 6008/13317 [1:01:56<1:20:46,  1.51it/s]
                                                        
{'loss': 0.1515, 'grad_norm': 0.5381866097450256, 'learning_rate': 6.6914080719212925e-06, 'epoch': 1.35}

 45%|████▌     | 6008/13317 [1:01:56<1:20:46,  1.51it/s]
 45%|████▌     | 6009/13317 [1:01:56<1:19:05,  1.54it/s]
                                                        
{'loss': 0.1949, 'grad_norm': 0.5321829319000244, 'learning_rate': 6.6901746471930625e-06, 'epoch': 1.35}

 45%|████▌     | 6009/13317 [1:01:56<1:19:05,  1.54it/s]
 45%|████▌     | 6010/13317 [1:01:57<1:17:46,  1.57it/s]
                                                        
{'loss': 0.246, 'grad_norm': 0.5893944501876831, 'learning_rate': 6.688941106331939e-06, 'epoch': 1.35}

 45%|████▌     | 6010/13317 [1:01:57<1:17:46,  1.57it/s]
 45%|████▌     | 6011/13317 [1:01:58<1:16:51,  1.58it/s]
                                                        
{'loss': 0.2448, 'grad_norm': 0.572230339050293, 'learning_rate': 6.687707449422679e-06, 'epoch': 1.35}

 45%|████▌     | 6011/13317 [1:01:58<1:16:51,  1.58it/s]
 45%|████▌     | 6012/13317 [1:01:58<1:16:11,  1.60it/s]
                                                        
{'loss': 0.1742, 'grad_norm': 0.5325755476951599, 'learning_rate': 6.686473676550046e-06, 'epoch': 1.35}

 45%|████▌     | 6012/13317 [1:01:58<1:16:11,  1.60it/s]
 45%|████▌     | 6013/13317 [1:01:59<1:15:52,  1.60it/s]
                                                        
{'loss': 0.1832, 'grad_norm': 0.5434613227844238, 'learning_rate': 6.685239787798816e-06, 'epoch': 1.35}

 45%|████▌     | 6013/13317 [1:01:59<1:15:52,  1.60it/s]
 45%|████▌     | 6014/13317 [1:01:59<1:15:37,  1.61it/s]
                                                        
{'loss': 0.1714, 'grad_norm': 0.4623665511608124, 'learning_rate': 6.684005783253769e-06, 'epoch': 1.35}

 45%|████▌     | 6014/13317 [1:01:59<1:15:37,  1.61it/s]
 45%|████▌     | 6015/13317 [1:02:00<1:15:21,  1.61it/s]
                                                        
{'loss': 0.2072, 'grad_norm': 0.5003488063812256, 'learning_rate': 6.682771662999694e-06, 'epoch': 1.36}

 45%|████▌     | 6015/13317 [1:02:00<1:15:21,  1.61it/s]
 45%|████▌     | 6016/13317 [1:02:01<1:15:09,  1.62it/s]
                                                        
{'loss': 0.1463, 'grad_norm': 0.4420963525772095, 'learning_rate': 6.681537427121388e-06, 'epoch': 1.36}

 45%|████▌     | 6016/13317 [1:02:01<1:15:09,  1.62it/s]
 45%|████▌     | 6017/13317 [1:02:01<1:14:57,  1.62it/s]
                                                        
{'loss': 0.1638, 'grad_norm': 0.4615480601787567, 'learning_rate': 6.680303075703656e-06, 'epoch': 1.36}

 45%|████▌     | 6017/13317 [1:02:01<1:14:57,  1.62it/s]
 45%|████▌     | 6018/13317 [1:02:02<1:15:15,  1.62it/s]
                                                        
{'loss': 0.2013, 'grad_norm': 0.4912770092487335, 'learning_rate': 6.679068608831313e-06, 'epoch': 1.36}

 45%|████▌     | 6018/13317 [1:02:02<1:15:15,  1.62it/s]
 45%|████▌     | 6019/13317 [1:02:02<1:15:07,  1.62it/s]
                                                        
{'loss': 0.1889, 'grad_norm': 0.5859134197235107, 'learning_rate': 6.677834026589176e-06, 'epoch': 1.36}

 45%|████▌     | 6019/13317 [1:02:03<1:15:07,  1.62it/s]
 45%|████▌     | 6020/13317 [1:02:03<1:14:59,  1.62it/s]
                                                        
{'loss': 0.2005, 'grad_norm': 0.5488490462303162, 'learning_rate': 6.676599329062079e-06, 'epoch': 1.36}

 45%|████▌     | 6020/13317 [1:02:03<1:14:59,  1.62it/s]
 45%|████▌     | 6021/13317 [1:02:04<1:14:50,  1.62it/s]
                                                        
{'loss': 0.201, 'grad_norm': 0.5253309607505798, 'learning_rate': 6.6753645163348526e-06, 'epoch': 1.36}

 45%|████▌     | 6021/13317 [1:02:04<1:14:50,  1.62it/s]
 45%|████▌     | 6022/13317 [1:02:04<1:14:46,  1.63it/s]
                                                        
{'loss': 0.2269, 'grad_norm': 0.6246716380119324, 'learning_rate': 6.674129588492348e-06, 'epoch': 1.36}

 45%|████▌     | 6022/13317 [1:02:04<1:14:46,  1.63it/s]
 45%|████▌     | 6023/13317 [1:02:05<1:15:05,  1.62it/s]
                                                        
{'loss': 0.1762, 'grad_norm': 0.5003761053085327, 'learning_rate': 6.6728945456194124e-06, 'epoch': 1.36}

 45%|████▌     | 6023/13317 [1:02:05<1:15:05,  1.62it/s]
 45%|████▌     | 6024/13317 [1:02:06<1:14:56,  1.62it/s]
                                                        
{'loss': 0.1168, 'grad_norm': 0.34494829177856445, 'learning_rate': 6.67165938780091e-06, 'epoch': 1.36}

 45%|████▌     | 6024/13317 [1:02:06<1:14:56,  1.62it/s]
 45%|████▌     | 6025/13317 [1:02:06<1:14:47,  1.62it/s]
                                                        
{'loss': 0.1862, 'grad_norm': 0.571480393409729, 'learning_rate': 6.670424115121708e-06, 'epoch': 1.36}

 45%|████▌     | 6025/13317 [1:02:06<1:14:47,  1.62it/s]
 45%|████▌     | 6026/13317 [1:02:07<1:14:43,  1.63it/s]
                                                        
{'loss': 0.1906, 'grad_norm': 0.4389052093029022, 'learning_rate': 6.669188727666681e-06, 'epoch': 1.36}

 45%|████▌     | 6026/13317 [1:02:07<1:14:43,  1.63it/s]
 45%|████▌     | 6027/13317 [1:02:07<1:14:40,  1.63it/s]
                                                        
{'loss': 0.1724, 'grad_norm': 0.472394198179245, 'learning_rate': 6.667953225520716e-06, 'epoch': 1.36}

 45%|████▌     | 6027/13317 [1:02:07<1:14:40,  1.63it/s]
 45%|████▌     | 6028/13317 [1:02:08<1:14:45,  1.63it/s]
                                                        
{'loss': 0.2314, 'grad_norm': 0.5991856455802917, 'learning_rate': 6.666717608768704e-06, 'epoch': 1.36}

 45%|████▌     | 6028/13317 [1:02:08<1:14:45,  1.63it/s]
 45%|████▌     | 6029/13317 [1:02:09<1:14:40,  1.63it/s]
                                                        
{'loss': 0.2154, 'grad_norm': 0.571690022945404, 'learning_rate': 6.665481877495544e-06, 'epoch': 1.36}

 45%|████▌     | 6029/13317 [1:02:09<1:14:40,  1.63it/s]
 45%|████▌     | 6030/13317 [1:02:09<1:14:34,  1.63it/s]
                                                        
{'loss': 0.2223, 'grad_norm': 0.619949460029602, 'learning_rate': 6.664246031786144e-06, 'epoch': 1.36}

 45%|████▌     | 6030/13317 [1:02:09<1:14:34,  1.63it/s]
 45%|████▌     | 6031/13317 [1:02:10<1:14:36,  1.63it/s]
                                                        
{'loss': 0.189, 'grad_norm': 0.47886475920677185, 'learning_rate': 6.663010071725422e-06, 'epoch': 1.36}

 45%|████▌     | 6031/13317 [1:02:10<1:14:36,  1.63it/s]
 45%|████▌     | 6032/13317 [1:02:10<1:14:48,  1.62it/s]
                                                        
{'loss': 0.214, 'grad_norm': 0.5369566082954407, 'learning_rate': 6.6617739973982985e-06, 'epoch': 1.36}

 45%|████▌     | 6032/13317 [1:02:11<1:14:48,  1.62it/s]
 45%|████▌     | 6033/13317 [1:02:11<1:14:43,  1.62it/s]
                                                        
{'loss': 0.1897, 'grad_norm': 0.4957960844039917, 'learning_rate': 6.660537808889706e-06, 'epoch': 1.36}

 45%|████▌     | 6033/13317 [1:02:11<1:14:43,  1.62it/s]
 45%|████▌     | 6034/13317 [1:02:12<1:14:39,  1.63it/s]
                                                        
{'loss': 0.173, 'grad_norm': 0.4571419954299927, 'learning_rate': 6.659301506284587e-06, 'epoch': 1.36}

 45%|████▌     | 6034/13317 [1:02:12<1:14:39,  1.63it/s]
 45%|████▌     | 6035/13317 [1:02:12<1:14:34,  1.63it/s]
                                                        
{'loss': 0.1714, 'grad_norm': 0.5546301603317261, 'learning_rate': 6.658065089667882e-06, 'epoch': 1.36}

 45%|████▌     | 6035/13317 [1:02:12<1:14:34,  1.63it/s]
 45%|████▌     | 6036/13317 [1:02:13<1:14:33,  1.63it/s]
                                                        
{'loss': 0.1812, 'grad_norm': 0.5584671497344971, 'learning_rate': 6.656828559124551e-06, 'epoch': 1.36}

 45%|████▌     | 6036/13317 [1:02:13<1:14:33,  1.63it/s]
 45%|████▌     | 6037/13317 [1:02:14<1:14:33,  1.63it/s]
                                                        
{'loss': 0.2303, 'grad_norm': 0.5877978205680847, 'learning_rate': 6.655591914739555e-06, 'epoch': 1.36}

 45%|████▌     | 6037/13317 [1:02:14<1:14:33,  1.63it/s]
 45%|████▌     | 6038/13317 [1:02:14<1:14:39,  1.62it/s]
                                                        
{'loss': 0.1625, 'grad_norm': 0.4377649426460266, 'learning_rate': 6.654355156597864e-06, 'epoch': 1.36}

 45%|████▌     | 6038/13317 [1:02:14<1:14:39,  1.62it/s]
 45%|████▌     | 6039/13317 [1:02:15<1:14:36,  1.63it/s]
                                                        
{'loss': 0.1744, 'grad_norm': 0.43697911500930786, 'learning_rate': 6.653118284784459e-06, 'epoch': 1.36}

 45%|████▌     | 6039/13317 [1:02:15<1:14:36,  1.63it/s]
 45%|████▌     | 6040/13317 [1:02:15<1:14:29,  1.63it/s]
                                                        
{'loss': 0.1869, 'grad_norm': 0.52436763048172, 'learning_rate': 6.651881299384324e-06, 'epoch': 1.36}

 45%|████▌     | 6040/13317 [1:02:15<1:14:29,  1.63it/s]
 45%|████▌     | 6041/13317 [1:02:16<1:14:26,  1.63it/s]
                                                        
{'loss': 0.221, 'grad_norm': 0.5182859897613525, 'learning_rate': 6.650644200482451e-06, 'epoch': 1.36}

 45%|████▌     | 6041/13317 [1:02:16<1:14:26,  1.63it/s]
 45%|████▌     | 6042/13317 [1:02:17<1:14:24,  1.63it/s]
                                                        
{'loss': 0.2203, 'grad_norm': 0.5480126142501831, 'learning_rate': 6.649406988163846e-06, 'epoch': 1.36}

 45%|████▌     | 6042/13317 [1:02:17<1:14:24,  1.63it/s]
 45%|████▌     | 6043/13317 [1:02:17<1:14:29,  1.63it/s]
                                                        
{'loss': 0.2007, 'grad_norm': 0.5167403817176819, 'learning_rate': 6.6481696625135175e-06, 'epoch': 1.36}

 45%|████▌     | 6043/13317 [1:02:17<1:14:29,  1.63it/s]
 45%|████▌     | 6044/13317 [1:02:18<1:14:28,  1.63it/s]
                                                        
{'loss': 0.3093, 'grad_norm': 0.6146815419197083, 'learning_rate': 6.64693222361648e-06, 'epoch': 1.36}

 45%|████▌     | 6044/13317 [1:02:18<1:14:28,  1.63it/s]
 45%|████▌     | 6045/13317 [1:02:18<1:14:24,  1.63it/s]
                                                        
{'loss': 0.1902, 'grad_norm': 0.5146430134773254, 'learning_rate': 6.645694671557763e-06, 'epoch': 1.36}

 45%|████▌     | 6045/13317 [1:02:19<1:14:24,  1.63it/s]
 45%|████▌     | 6046/13317 [1:02:19<1:14:23,  1.63it/s]
                                                        
{'loss': 0.1776, 'grad_norm': 0.4336983859539032, 'learning_rate': 6.644457006422396e-06, 'epoch': 1.36}

 45%|████▌     | 6046/13317 [1:02:19<1:14:23,  1.63it/s]
 45%|████▌     | 6047/13317 [1:02:20<1:14:20,  1.63it/s]
                                                        
{'loss': 0.1851, 'grad_norm': 0.5018842816352844, 'learning_rate': 6.6432192282954214e-06, 'epoch': 1.36}

 45%|████▌     | 6047/13317 [1:02:20<1:14:20,  1.63it/s]
 45%|████▌     | 6048/13317 [1:02:20<1:14:24,  1.63it/s]
                                                        
{'loss': 0.2005, 'grad_norm': 0.5280170440673828, 'learning_rate': 6.6419813372618844e-06, 'epoch': 1.36}

 45%|████▌     | 6048/13317 [1:02:20<1:14:24,  1.63it/s]
 45%|████▌     | 6049/13317 [1:02:21<1:14:22,  1.63it/s]
                                                        
{'loss': 0.1641, 'grad_norm': 0.45089200139045715, 'learning_rate': 6.640743333406847e-06, 'epoch': 1.36}

 45%|████▌     | 6049/13317 [1:02:21<1:14:22,  1.63it/s]
 45%|████▌     | 6050/13317 [1:02:21<1:14:22,  1.63it/s]
                                                        
{'loss': 0.1848, 'grad_norm': 0.4812179505825043, 'learning_rate': 6.639505216815369e-06, 'epoch': 1.36}

 45%|████▌     | 6050/13317 [1:02:22<1:14:22,  1.63it/s]
 45%|████▌     | 6051/13317 [1:02:22<1:14:19,  1.63it/s]
                                                        
{'loss': 0.2278, 'grad_norm': 0.6642034649848938, 'learning_rate': 6.638266987572523e-06, 'epoch': 1.36}

 45%|████▌     | 6051/13317 [1:02:22<1:14:19,  1.63it/s]
 45%|████▌     | 6052/13317 [1:02:23<1:14:16,  1.63it/s]
                                                        
{'loss': 0.1761, 'grad_norm': 0.46759647130966187, 'learning_rate': 6.637028645763389e-06, 'epoch': 1.36}

 45%|████▌     | 6052/13317 [1:02:23<1:14:16,  1.63it/s]
 45%|████▌     | 6053/13317 [1:02:23<1:14:18,  1.63it/s]
                                                        
{'loss': 0.2088, 'grad_norm': 0.5061015486717224, 'learning_rate': 6.635790191473055e-06, 'epoch': 1.36}

 45%|████▌     | 6053/13317 [1:02:23<1:14:18,  1.63it/s]
 45%|████▌     | 6054/13317 [1:02:24<1:14:20,  1.63it/s]
                                                        
{'loss': 0.1596, 'grad_norm': 0.43166980147361755, 'learning_rate': 6.634551624786614e-06, 'epoch': 1.36}

 45%|████▌     | 6054/13317 [1:02:24<1:14:20,  1.63it/s]
 45%|████▌     | 6055/13317 [1:02:25<1:14:17,  1.63it/s]
                                                        
{'loss': 0.172, 'grad_norm': 0.4543709456920624, 'learning_rate': 6.633312945789169e-06, 'epoch': 1.36}

 45%|████▌     | 6055/13317 [1:02:25<1:14:17,  1.63it/s]
 45%|████▌     | 6056/13317 [1:02:25<1:14:18,  1.63it/s]
                                                        
{'loss': 0.2369, 'grad_norm': 0.609501838684082, 'learning_rate': 6.632074154565832e-06, 'epoch': 1.36}

 45%|████▌     | 6056/13317 [1:02:25<1:14:18,  1.63it/s]
 45%|████▌     | 6057/13317 [1:02:26<1:14:18,  1.63it/s]
                                                        
{'loss': 0.233, 'grad_norm': 0.530981719493866, 'learning_rate': 6.6308352512017185e-06, 'epoch': 1.36}

 45%|████▌     | 6057/13317 [1:02:26<1:14:18,  1.63it/s]
 45%|████▌     | 6058/13317 [1:02:26<1:14:22,  1.63it/s]
                                                        
{'loss': 0.2307, 'grad_norm': 0.5561216473579407, 'learning_rate': 6.629596235781956e-06, 'epoch': 1.36}

 45%|████▌     | 6058/13317 [1:02:27<1:14:22,  1.63it/s]
 45%|████▌     | 6059/13317 [1:02:27<1:14:22,  1.63it/s]
                                                        
{'loss': 0.1762, 'grad_norm': 0.4594549238681793, 'learning_rate': 6.628357108391678e-06, 'epoch': 1.36}

 45%|████▌     | 6059/13317 [1:02:27<1:14:22,  1.63it/s]
 46%|████▌     | 6060/13317 [1:02:28<1:14:16,  1.63it/s]
                                                        
{'loss': 0.2036, 'grad_norm': 0.4846983253955841, 'learning_rate': 6.627117869116026e-06, 'epoch': 1.37}

 46%|████▌     | 6060/13317 [1:02:28<1:14:16,  1.63it/s]
 46%|████▌     | 6061/13317 [1:02:28<1:14:16,  1.63it/s]
                                                        
{'loss': 0.1936, 'grad_norm': 0.4942977726459503, 'learning_rate': 6.625878518040147e-06, 'epoch': 1.37}

 46%|████▌     | 6061/13317 [1:02:28<1:14:16,  1.63it/s]
 46%|████▌     | 6062/13317 [1:02:29<1:14:18,  1.63it/s]
                                                        
{'loss': 0.2315, 'grad_norm': 0.6687050461769104, 'learning_rate': 6.624639055249199e-06, 'epoch': 1.37}

 46%|████▌     | 6062/13317 [1:02:29<1:14:18,  1.63it/s]
 46%|████▌     | 6063/13317 [1:02:29<1:14:37,  1.62it/s]
                                                        
{'loss': 0.264, 'grad_norm': 0.6547748446464539, 'learning_rate': 6.623399480828346e-06, 'epoch': 1.37}

 46%|████▌     | 6063/13317 [1:02:30<1:14:37,  1.62it/s]
 46%|████▌     | 6064/13317 [1:02:30<1:14:28,  1.62it/s]
                                                        
{'loss': 0.1964, 'grad_norm': 0.4843771457672119, 'learning_rate': 6.62215979486276e-06, 'epoch': 1.37}

 46%|████▌     | 6064/13317 [1:02:30<1:14:28,  1.62it/s]
 46%|████▌     | 6065/13317 [1:02:31<1:14:24,  1.62it/s]
                                                        
{'loss': 0.1899, 'grad_norm': 0.4932829737663269, 'learning_rate': 6.6209199974376204e-06, 'epoch': 1.37}

 46%|████▌     | 6065/13317 [1:02:31<1:14:24,  1.62it/s]
 46%|████▌     | 6066/13317 [1:02:31<1:14:18,  1.63it/s]
                                                        
{'loss': 0.2002, 'grad_norm': 0.45596131682395935, 'learning_rate': 6.619680088638114e-06, 'epoch': 1.37}

 46%|████▌     | 6066/13317 [1:02:31<1:14:18,  1.63it/s]
 46%|████▌     | 6067/13317 [1:02:32<1:14:14,  1.63it/s]
                                                        
{'loss': 0.2312, 'grad_norm': 0.5432839393615723, 'learning_rate': 6.618440068549437e-06, 'epoch': 1.37}

 46%|████▌     | 6067/13317 [1:02:32<1:14:14,  1.63it/s]
 46%|████▌     | 6068/13317 [1:02:33<1:14:17,  1.63it/s]
                                                        
{'loss': 0.2096, 'grad_norm': 0.5462842583656311, 'learning_rate': 6.61719993725679e-06, 'epoch': 1.37}

 46%|████▌     | 6068/13317 [1:02:33<1:14:17,  1.63it/s]
 46%|████▌     | 6069/13317 [1:02:33<1:14:12,  1.63it/s]
                                                        
{'loss': 0.1676, 'grad_norm': 0.5002124309539795, 'learning_rate': 6.615959694845385e-06, 'epoch': 1.37}

 46%|████▌     | 6069/13317 [1:02:33<1:14:12,  1.63it/s]
 46%|████▌     | 6070/13317 [1:02:34<1:14:08,  1.63it/s]
                                                        
{'loss': 0.1718, 'grad_norm': 0.4621431529521942, 'learning_rate': 6.614719341400438e-06, 'epoch': 1.37}

 46%|████▌     | 6070/13317 [1:02:34<1:14:08,  1.63it/s]
 46%|████▌     | 6071/13317 [1:02:34<1:14:07,  1.63it/s]
                                                        
{'loss': 0.1732, 'grad_norm': 0.5190293192863464, 'learning_rate': 6.613478877007176e-06, 'epoch': 1.37}

 46%|████▌     | 6071/13317 [1:02:35<1:14:07,  1.63it/s]
 46%|████▌     | 6072/13317 [1:02:35<1:14:06,  1.63it/s]
                                                        
{'loss': 0.1533, 'grad_norm': 0.4316597580909729, 'learning_rate': 6.612238301750832e-06, 'epoch': 1.37}

 46%|████▌     | 6072/13317 [1:02:35<1:14:06,  1.63it/s]
 46%|████▌     | 6073/13317 [1:02:36<1:14:07,  1.63it/s]
                                                        
{'loss': 0.2497, 'grad_norm': 0.6051799654960632, 'learning_rate': 6.6109976157166435e-06, 'epoch': 1.37}

 46%|████▌     | 6073/13317 [1:02:36<1:14:07,  1.63it/s]
 46%|████▌     | 6074/13317 [1:02:36<1:14:09,  1.63it/s]
                                                        
{'loss': 0.2006, 'grad_norm': 0.5412940382957458, 'learning_rate': 6.609756818989864e-06, 'epoch': 1.37}

 46%|████▌     | 6074/13317 [1:02:36<1:14:09,  1.63it/s]
 46%|████▌     | 6075/13317 [1:02:37<1:14:09,  1.63it/s]
                                                        
{'loss': 0.1424, 'grad_norm': 0.4018043577671051, 'learning_rate': 6.608515911655744e-06, 'epoch': 1.37}

 46%|████▌     | 6075/13317 [1:02:37<1:14:09,  1.63it/s]
 46%|████▌     | 6076/13317 [1:02:37<1:14:06,  1.63it/s]
                                                        
{'loss': 0.2145, 'grad_norm': 0.5604133009910583, 'learning_rate': 6.607274893799553e-06, 'epoch': 1.37}

 46%|████▌     | 6076/13317 [1:02:38<1:14:06,  1.63it/s]
 46%|████▌     | 6077/13317 [1:02:38<1:14:08,  1.63it/s]
                                                        
{'loss': 0.1765, 'grad_norm': 0.48434099555015564, 'learning_rate': 6.606033765506556e-06, 'epoch': 1.37}

 46%|████▌     | 6077/13317 [1:02:38<1:14:08,  1.63it/s]
 46%|████▌     | 6078/13317 [1:02:39<1:14:09,  1.63it/s]
                                                        
{'loss': 0.2059, 'grad_norm': 0.4965876042842865, 'learning_rate': 6.6047925268620365e-06, 'epoch': 1.37}

 46%|████▌     | 6078/13317 [1:02:39<1:14:09,  1.63it/s]
 46%|████▌     | 6079/13317 [1:02:39<1:14:03,  1.63it/s]
                                                        
{'loss': 0.1966, 'grad_norm': 0.4747132956981659, 'learning_rate': 6.6035511779512764e-06, 'epoch': 1.37}

 46%|████▌     | 6079/13317 [1:02:39<1:14:03,  1.63it/s]
 46%|████▌     | 6080/13317 [1:02:40<1:14:03,  1.63it/s]
                                                        
{'loss': 0.194, 'grad_norm': 0.5133777856826782, 'learning_rate': 6.602309718859572e-06, 'epoch': 1.37}

 46%|████▌     | 6080/13317 [1:02:40<1:14:03,  1.63it/s]
 46%|████▌     | 6081/13317 [1:02:41<1:14:01,  1.63it/s]
                                                        
{'loss': 0.2328, 'grad_norm': 0.5699354410171509, 'learning_rate': 6.601068149672226e-06, 'epoch': 1.37}

 46%|████▌     | 6081/13317 [1:02:41<1:14:01,  1.63it/s]
 46%|████▌     | 6082/13317 [1:02:41<1:13:59,  1.63it/s]
                                                        
{'loss': 0.2577, 'grad_norm': 0.6009747982025146, 'learning_rate': 6.599826470474544e-06, 'epoch': 1.37}

 46%|████▌     | 6082/13317 [1:02:41<1:13:59,  1.63it/s]
 46%|████▌     | 6083/13317 [1:02:42<1:14:21,  1.62it/s]
                                                        
{'loss': 0.2005, 'grad_norm': 0.5423474907875061, 'learning_rate': 6.598584681351846e-06, 'epoch': 1.37}

 46%|████▌     | 6083/13317 [1:02:42<1:14:21,  1.62it/s]
 46%|████▌     | 6084/13317 [1:02:42<1:14:13,  1.62it/s]
                                                        
{'loss': 0.2021, 'grad_norm': 0.512998640537262, 'learning_rate': 6.597342782389453e-06, 'epoch': 1.37}

 46%|████▌     | 6084/13317 [1:02:43<1:14:13,  1.62it/s]
 46%|████▌     | 6085/13317 [1:02:43<1:14:06,  1.63it/s]
                                                        
{'loss': 0.2349, 'grad_norm': 0.5287438035011292, 'learning_rate': 6.596100773672698e-06, 'epoch': 1.37}

 46%|████▌     | 6085/13317 [1:02:43<1:14:06,  1.63it/s]
 46%|████▌     | 6086/13317 [1:02:44<1:14:04,  1.63it/s]
                                                        
{'loss': 0.2654, 'grad_norm': 0.6230257749557495, 'learning_rate': 6.5948586552869195e-06, 'epoch': 1.37}

 46%|████▌     | 6086/13317 [1:02:44<1:14:04,  1.63it/s]
 46%|████▌     | 6087/13317 [1:02:44<1:14:00,  1.63it/s]
                                                        
{'loss': 0.1916, 'grad_norm': 0.5407658219337463, 'learning_rate': 6.5936164273174665e-06, 'epoch': 1.37}

 46%|████▌     | 6087/13317 [1:02:44<1:14:00,  1.63it/s]
 46%|████▌     | 6088/13317 [1:02:45<1:14:03,  1.63it/s]
                                                        
{'loss': 0.1773, 'grad_norm': 0.4597419798374176, 'learning_rate': 6.59237408984969e-06, 'epoch': 1.37}

 46%|████▌     | 6088/13317 [1:02:45<1:14:03,  1.63it/s]
 46%|████▌     | 6089/13317 [1:02:45<1:14:06,  1.63it/s]
                                                        
{'loss': 0.1908, 'grad_norm': 0.5325137376785278, 'learning_rate': 6.5911316429689545e-06, 'epoch': 1.37}

 46%|████▌     | 6089/13317 [1:02:46<1:14:06,  1.63it/s]
 46%|████▌     | 6090/13317 [1:02:46<1:14:03,  1.63it/s]
                                                        
{'loss': 0.2126, 'grad_norm': 0.5534863471984863, 'learning_rate': 6.589889086760627e-06, 'epoch': 1.37}

 46%|████▌     | 6090/13317 [1:02:46<1:14:03,  1.63it/s]
 46%|████▌     | 6091/13317 [1:02:47<1:13:58,  1.63it/s]
                                                        
{'loss': 0.2592, 'grad_norm': 0.6190635561943054, 'learning_rate': 6.588646421310086e-06, 'epoch': 1.37}

 46%|████▌     | 6091/13317 [1:02:47<1:13:58,  1.63it/s]
 46%|████▌     | 6092/13317 [1:02:47<1:13:57,  1.63it/s]
                                                        
{'loss': 0.2186, 'grad_norm': 0.5848522782325745, 'learning_rate': 6.5874036467027135e-06, 'epoch': 1.37}

 46%|████▌     | 6092/13317 [1:02:47<1:13:57,  1.63it/s]
 46%|████▌     | 6093/13317 [1:02:48<1:14:02,  1.63it/s]
                                                        
{'loss': 0.167, 'grad_norm': 0.44525253772735596, 'learning_rate': 6.586160763023903e-06, 'epoch': 1.37}

 46%|████▌     | 6093/13317 [1:02:48<1:14:02,  1.63it/s]
 46%|████▌     | 6094/13317 [1:02:49<1:13:58,  1.63it/s]
                                                        
{'loss': 0.2315, 'grad_norm': 0.5477690696716309, 'learning_rate': 6.584917770359056e-06, 'epoch': 1.37}

 46%|████▌     | 6094/13317 [1:02:49<1:13:58,  1.63it/s]
 46%|████▌     | 6095/13317 [1:02:49<1:13:57,  1.63it/s]
                                                        
{'loss': 0.2541, 'grad_norm': 0.5468505024909973, 'learning_rate': 6.5836746687935735e-06, 'epoch': 1.37}

 46%|████▌     | 6095/13317 [1:02:49<1:13:57,  1.63it/s]
 46%|████▌     | 6096/13317 [1:02:50<1:13:55,  1.63it/s]
                                                        
{'loss': 0.2163, 'grad_norm': 0.5607853531837463, 'learning_rate': 6.5824314584128755e-06, 'epoch': 1.37}

 46%|████▌     | 6096/13317 [1:02:50<1:13:55,  1.63it/s]
 46%|████▌     | 6097/13317 [1:02:50<1:13:51,  1.63it/s]
                                                        
{'loss': 0.2093, 'grad_norm': 0.5851812958717346, 'learning_rate': 6.58118813930238e-06, 'epoch': 1.37}

 46%|████▌     | 6097/13317 [1:02:51<1:13:51,  1.63it/s]
 46%|████▌     | 6098/13317 [1:02:51<1:13:57,  1.63it/s]
                                                        
{'loss': 0.2291, 'grad_norm': 0.5256723761558533, 'learning_rate': 6.579944711547518e-06, 'epoch': 1.37}

 46%|████▌     | 6098/13317 [1:02:51<1:13:57,  1.63it/s]
 46%|████▌     | 6099/13317 [1:02:52<1:14:01,  1.63it/s]
                                                        
{'loss': 0.2057, 'grad_norm': 0.5401893258094788, 'learning_rate': 6.578701175233725e-06, 'epoch': 1.37}

 46%|████▌     | 6099/13317 [1:02:52<1:14:01,  1.63it/s]
 46%|████▌     | 6100/13317 [1:02:52<1:13:57,  1.63it/s]
                                                        
{'loss': 0.1917, 'grad_norm': 0.4538441598415375, 'learning_rate': 6.577457530446445e-06, 'epoch': 1.37}

 46%|████▌     | 6100/13317 [1:02:52<1:13:57,  1.63it/s]
 46%|████▌     | 6101/13317 [1:02:53<1:13:53,  1.63it/s]
                                                        
{'loss': 0.2014, 'grad_norm': 0.4820728898048401, 'learning_rate': 6.576213777271133e-06, 'epoch': 1.37}

 46%|████▌     | 6101/13317 [1:02:53<1:13:53,  1.63it/s]
 46%|████▌     | 6102/13317 [1:02:53<1:13:51,  1.63it/s]
                                                        
{'loss': 0.1957, 'grad_norm': 0.5631095767021179, 'learning_rate': 6.574969915793242e-06, 'epoch': 1.37}

 46%|████▌     | 6102/13317 [1:02:54<1:13:51,  1.63it/s]
 46%|████▌     | 6103/13317 [1:02:54<1:13:55,  1.63it/s]
                                                        
{'loss': 0.2072, 'grad_norm': 0.5354728698730469, 'learning_rate': 6.573725946098244e-06, 'epoch': 1.37}

 46%|████▌     | 6103/13317 [1:02:54<1:13:55,  1.63it/s]
 46%|████▌     | 6104/13317 [1:02:55<1:13:54,  1.63it/s]
                                                        
{'loss': 0.1907, 'grad_norm': 0.5212271809577942, 'learning_rate': 6.57248186827161e-06, 'epoch': 1.38}

 46%|████▌     | 6104/13317 [1:02:55<1:13:54,  1.63it/s]
 46%|████▌     | 6105/13317 [1:02:55<1:13:54,  1.63it/s]
                                                        
{'loss': 0.1699, 'grad_norm': 0.4379768371582031, 'learning_rate': 6.571237682398822e-06, 'epoch': 1.38}

 46%|████▌     | 6105/13317 [1:02:55<1:13:54,  1.63it/s]
 46%|████▌     | 6106/13317 [1:02:56<1:13:52,  1.63it/s]
                                                        
{'loss': 0.2883, 'grad_norm': 0.6907129883766174, 'learning_rate': 6.569993388565367e-06, 'epoch': 1.38}

 46%|████▌     | 6106/13317 [1:02:56<1:13:52,  1.63it/s]
 46%|████▌     | 6107/13317 [1:02:57<1:13:51,  1.63it/s]
                                                        
{'loss': 0.224, 'grad_norm': 0.6377509832382202, 'learning_rate': 6.568748986856746e-06, 'epoch': 1.38}

 46%|████▌     | 6107/13317 [1:02:57<1:13:51,  1.63it/s]
 46%|████▌     | 6108/13317 [1:02:57<1:13:53,  1.63it/s]
                                                        
{'loss': 0.2214, 'grad_norm': 0.5339236855506897, 'learning_rate': 6.567504477358459e-06, 'epoch': 1.38}

 46%|████▌     | 6108/13317 [1:02:57<1:13:53,  1.63it/s]
 46%|████▌     | 6109/13317 [1:02:58<1:13:54,  1.63it/s]
                                                        
{'loss': 0.2552, 'grad_norm': 0.7251378893852234, 'learning_rate': 6.566259860156015e-06, 'epoch': 1.38}

 46%|████▌     | 6109/13317 [1:02:58<1:13:54,  1.63it/s]
 46%|████▌     | 6110/13317 [1:02:58<1:13:49,  1.63it/s]
                                                        
{'loss': 0.1862, 'grad_norm': 0.4650956988334656, 'learning_rate': 6.5650151353349365e-06, 'epoch': 1.38}

 46%|████▌     | 6110/13317 [1:02:58<1:13:49,  1.63it/s]
 46%|████▌     | 6111/13317 [1:02:59<1:13:46,  1.63it/s]
                                                        
{'loss': 0.1653, 'grad_norm': 0.5067366361618042, 'learning_rate': 6.563770302980746e-06, 'epoch': 1.38}

 46%|████▌     | 6111/13317 [1:02:59<1:13:46,  1.63it/s]
 46%|████▌     | 6112/13317 [1:03:00<1:13:44,  1.63it/s]
                                                        
{'loss': 0.1473, 'grad_norm': 0.36685606837272644, 'learning_rate': 6.56252536317898e-06, 'epoch': 1.38}

 46%|████▌     | 6112/13317 [1:03:00<1:13:44,  1.63it/s]
 46%|████▌     | 6113/13317 [1:03:00<1:13:50,  1.63it/s]
                                                        
{'loss': 0.1619, 'grad_norm': 0.45806288719177246, 'learning_rate': 6.561280316015176e-06, 'epoch': 1.38}

 46%|████▌     | 6113/13317 [1:03:00<1:13:50,  1.63it/s]
 46%|████▌     | 6114/13317 [1:03:01<1:13:49,  1.63it/s]
                                                        
{'loss': 0.2079, 'grad_norm': 0.4632258713245392, 'learning_rate': 6.560035161574884e-06, 'epoch': 1.38}

 46%|████▌     | 6114/13317 [1:03:01<1:13:49,  1.63it/s]
 46%|████▌     | 6115/13317 [1:03:01<1:13:46,  1.63it/s]
                                                        
{'loss': 0.1592, 'grad_norm': 0.42662325501441956, 'learning_rate': 6.558789899943659e-06, 'epoch': 1.38}

 46%|████▌     | 6115/13317 [1:03:02<1:13:46,  1.63it/s]
 46%|████▌     | 6116/13317 [1:03:02<1:13:44,  1.63it/s]
                                                        
{'loss': 0.2258, 'grad_norm': 0.5549092292785645, 'learning_rate': 6.557544531207061e-06, 'epoch': 1.38}

 46%|████▌     | 6116/13317 [1:03:02<1:13:44,  1.63it/s]
 46%|████▌     | 6117/13317 [1:03:03<1:13:39,  1.63it/s]
                                                        
{'loss': 0.1967, 'grad_norm': 0.6034212112426758, 'learning_rate': 6.556299055450665e-06, 'epoch': 1.38}

 46%|████▌     | 6117/13317 [1:03:03<1:13:39,  1.63it/s]
 46%|████▌     | 6118/13317 [1:03:03<1:14:07,  1.62it/s]
                                                        
{'loss': 0.1764, 'grad_norm': 0.4624853730201721, 'learning_rate': 6.5550534727600434e-06, 'epoch': 1.38}

 46%|████▌     | 6118/13317 [1:03:03<1:14:07,  1.62it/s]
 46%|████▌     | 6119/13317 [1:03:04<1:13:59,  1.62it/s]
                                                        
{'loss': 0.2602, 'grad_norm': 0.655986487865448, 'learning_rate': 6.553807783220784e-06, 'epoch': 1.38}

 46%|████▌     | 6119/13317 [1:03:04<1:13:59,  1.62it/s]
 46%|████▌     | 6120/13317 [1:03:05<1:13:52,  1.62it/s]
                                                        
{'loss': 0.2204, 'grad_norm': 0.4717502295970917, 'learning_rate': 6.552561986918477e-06, 'epoch': 1.38}

 46%|████▌     | 6120/13317 [1:03:05<1:13:52,  1.62it/s]
 46%|████▌     | 6121/13317 [1:03:05<1:13:47,  1.63it/s]
                                                        
{'loss': 0.1781, 'grad_norm': 0.4892796277999878, 'learning_rate': 6.551316083938724e-06, 'epoch': 1.38}

 46%|████▌     | 6121/13317 [1:03:05<1:13:47,  1.63it/s]
 46%|████▌     | 6122/13317 [1:03:06<1:13:47,  1.63it/s]
                                                        
{'loss': 0.2741, 'grad_norm': 0.5767796039581299, 'learning_rate': 6.550070074367129e-06, 'epoch': 1.38}

 46%|████▌     | 6122/13317 [1:03:06<1:13:47,  1.63it/s]
 46%|████▌     | 6123/13317 [1:03:06<1:13:48,  1.62it/s]
                                                        
{'loss': 0.1792, 'grad_norm': 0.49986451864242554, 'learning_rate': 6.54882395828931e-06, 'epoch': 1.38}

 46%|████▌     | 6123/13317 [1:03:06<1:13:48,  1.62it/s]
 46%|████▌     | 6124/13317 [1:03:07<1:13:46,  1.62it/s]
                                                        
{'loss': 0.1407, 'grad_norm': 0.39438197016716003, 'learning_rate': 6.547577735790883e-06, 'epoch': 1.38}

 46%|████▌     | 6124/13317 [1:03:07<1:13:46,  1.62it/s]
 46%|████▌     | 6125/13317 [1:03:08<1:13:41,  1.63it/s]
                                                        
{'loss': 0.19, 'grad_norm': 0.5088313221931458, 'learning_rate': 6.546331406957481e-06, 'epoch': 1.38}

 46%|████▌     | 6125/13317 [1:03:08<1:13:41,  1.63it/s]
 46%|████▌     | 6126/13317 [1:03:08<1:13:38,  1.63it/s]
                                                        
{'loss': 0.2065, 'grad_norm': 0.5282174944877625, 'learning_rate': 6.545084971874738e-06, 'epoch': 1.38}

 46%|████▌     | 6126/13317 [1:03:08<1:13:38,  1.63it/s]
 46%|████▌     | 6127/13317 [1:03:09<1:13:35,  1.63it/s]
                                                        
{'loss': 0.1986, 'grad_norm': 0.49787232279777527, 'learning_rate': 6.5438384306282975e-06, 'epoch': 1.38}

 46%|████▌     | 6127/13317 [1:03:09<1:13:35,  1.63it/s]
 46%|████▌     | 6128/13317 [1:03:09<1:13:38,  1.63it/s]
                                                        
{'loss': 0.2917, 'grad_norm': 0.6113601922988892, 'learning_rate': 6.542591783303812e-06, 'epoch': 1.38}

 46%|████▌     | 6128/13317 [1:03:10<1:13:38,  1.63it/s]
 46%|████▌     | 6129/13317 [1:03:10<1:13:42,  1.63it/s]
                                                        
{'loss': 0.2102, 'grad_norm': 0.5178261995315552, 'learning_rate': 6.541345029986935e-06, 'epoch': 1.38}

 46%|████▌     | 6129/13317 [1:03:10<1:13:42,  1.63it/s]
 46%|████▌     | 6130/13317 [1:03:11<1:13:37,  1.63it/s]
                                                        
{'loss': 0.1698, 'grad_norm': 0.4901188910007477, 'learning_rate': 6.540098170763337e-06, 'epoch': 1.38}

 46%|████▌     | 6130/13317 [1:03:11<1:13:37,  1.63it/s]
 46%|████▌     | 6131/13317 [1:03:11<1:13:35,  1.63it/s]
                                                        
{'loss': 0.2356, 'grad_norm': 0.5737875699996948, 'learning_rate': 6.5388512057186855e-06, 'epoch': 1.38}

 46%|████▌     | 6131/13317 [1:03:11<1:13:35,  1.63it/s]
 46%|████▌     | 6132/13317 [1:03:12<1:13:32,  1.63it/s]
                                                        
{'loss': 0.1981, 'grad_norm': 0.586671769618988, 'learning_rate': 6.537604134938666e-06, 'epoch': 1.38}

 46%|████▌     | 6132/13317 [1:03:12<1:13:32,  1.63it/s]
 46%|████▌     | 6133/13317 [1:03:13<1:13:38,  1.63it/s]
                                                        
{'loss': 0.1736, 'grad_norm': 0.457695871591568, 'learning_rate': 6.536356958508959e-06, 'epoch': 1.38}

 46%|████▌     | 6133/13317 [1:03:13<1:13:38,  1.63it/s]
 46%|████▌     | 6134/13317 [1:03:13<1:13:42,  1.62it/s]
                                                        
{'loss': 0.1847, 'grad_norm': 0.5247774720191956, 'learning_rate': 6.535109676515263e-06, 'epoch': 1.38}

 46%|████▌     | 6134/13317 [1:03:13<1:13:42,  1.62it/s]
 46%|████▌     | 6135/13317 [1:03:14<1:13:39,  1.62it/s]
                                                        
{'loss': 0.1991, 'grad_norm': 0.5338635444641113, 'learning_rate': 6.533862289043279e-06, 'epoch': 1.38}

 46%|████▌     | 6135/13317 [1:03:14<1:13:39,  1.62it/s]
 46%|████▌     | 6136/13317 [1:03:14<1:13:34,  1.63it/s]
                                                        
{'loss': 0.2388, 'grad_norm': 0.5547301769256592, 'learning_rate': 6.532614796178712e-06, 'epoch': 1.38}

 46%|████▌     | 6136/13317 [1:03:14<1:13:34,  1.63it/s]
 46%|████▌     | 6137/13317 [1:03:15<1:13:31,  1.63it/s]
                                                        
{'loss': 0.2164, 'grad_norm': 0.49006643891334534, 'learning_rate': 6.531367198007285e-06, 'epoch': 1.38}

 46%|████▌     | 6137/13317 [1:03:15<1:13:31,  1.63it/s]
 46%|████▌     | 6138/13317 [1:03:16<1:13:49,  1.62it/s]
                                                        
{'loss': 0.2398, 'grad_norm': 0.6400893330574036, 'learning_rate': 6.530119494614713e-06, 'epoch': 1.38}

 46%|████▌     | 6138/13317 [1:03:16<1:13:49,  1.62it/s]
 46%|████▌     | 6139/13317 [1:03:16<1:14:00,  1.62it/s]
                                                        
{'loss': 0.1682, 'grad_norm': 0.44849953055381775, 'learning_rate': 6.528871686086733e-06, 'epoch': 1.38}

 46%|████▌     | 6139/13317 [1:03:16<1:14:00,  1.62it/s]
 46%|████▌     | 6140/13317 [1:03:17<1:13:48,  1.62it/s]
                                                        
{'loss': 0.4051, 'grad_norm': 0.6095478534698486, 'learning_rate': 6.52762377250908e-06, 'epoch': 1.38}

 46%|████▌     | 6140/13317 [1:03:17<1:13:48,  1.62it/s]
 46%|████▌     | 6141/13317 [1:03:17<1:13:41,  1.62it/s]
                                                        
{'loss': 0.1319, 'grad_norm': 0.44715437293052673, 'learning_rate': 6.526375753967499e-06, 'epoch': 1.38}

 46%|████▌     | 6141/13317 [1:03:18<1:13:41,  1.62it/s]
 46%|████▌     | 6142/13317 [1:03:18<1:13:34,  1.63it/s]
                                                        
{'loss': 0.2253, 'grad_norm': 0.5900529623031616, 'learning_rate': 6.525127630547741e-06, 'epoch': 1.38}

 46%|████▌     | 6142/13317 [1:03:18<1:13:34,  1.63it/s]
 46%|████▌     | 6143/13317 [1:03:19<1:13:35,  1.62it/s]
                                                        
{'loss': 0.1514, 'grad_norm': 0.4187045991420746, 'learning_rate': 6.523879402335568e-06, 'epoch': 1.38}

 46%|████▌     | 6143/13317 [1:03:19<1:13:35,  1.62it/s]
 46%|████▌     | 6144/13317 [1:03:19<1:13:35,  1.62it/s]
                                                        
{'loss': 0.1959, 'grad_norm': 0.5171992182731628, 'learning_rate': 6.522631069416745e-06, 'epoch': 1.38}

 46%|████▌     | 6144/13317 [1:03:19<1:13:35,  1.62it/s]
 46%|████▌     | 6145/13317 [1:03:20<1:13:30,  1.63it/s]
                                                        
{'loss': 0.2318, 'grad_norm': 0.5537123084068298, 'learning_rate': 6.521382631877044e-06, 'epoch': 1.38}

 46%|████▌     | 6145/13317 [1:03:20<1:13:30,  1.63it/s]
 46%|████▌     | 6146/13317 [1:03:21<1:13:26,  1.63it/s]
                                                        
{'loss': 0.1895, 'grad_norm': 0.49466434121131897, 'learning_rate': 6.520134089802249e-06, 'epoch': 1.38}

 46%|████▌     | 6146/13317 [1:03:21<1:13:26,  1.63it/s]
 46%|████▌     | 6147/13317 [1:03:21<1:13:22,  1.63it/s]
                                                        
{'loss': 0.2449, 'grad_norm': 0.57984459400177, 'learning_rate': 6.518885443278147e-06, 'epoch': 1.38}

 46%|████▌     | 6147/13317 [1:03:21<1:13:22,  1.63it/s]
 46%|████▌     | 6148/13317 [1:03:22<1:13:25,  1.63it/s]
                                                        
{'loss': 0.2116, 'grad_norm': 0.5787141919136047, 'learning_rate': 6.517636692390531e-06, 'epoch': 1.38}

 46%|████▌     | 6148/13317 [1:03:22<1:13:25,  1.63it/s]
 46%|████▌     | 6149/13317 [1:03:22<1:13:31,  1.62it/s]
                                                        
{'loss': 0.2126, 'grad_norm': 0.5081918239593506, 'learning_rate': 6.516387837225206e-06, 'epoch': 1.39}

 46%|████▌     | 6149/13317 [1:03:22<1:13:31,  1.62it/s]
 46%|████▌     | 6150/13317 [1:03:23<1:13:28,  1.63it/s]
                                                        
{'loss': 0.2367, 'grad_norm': 0.5669789910316467, 'learning_rate': 6.515138877867983e-06, 'epoch': 1.39}

 46%|████▌     | 6150/13317 [1:03:23<1:13:28,  1.63it/s]
 46%|████▌     | 6151/13317 [1:03:24<1:13:23,  1.63it/s]
                                                        
{'loss': 0.2444, 'grad_norm': 0.6476265788078308, 'learning_rate': 6.5138898144046735e-06, 'epoch': 1.39}

 46%|████▌     | 6151/13317 [1:03:24<1:13:23,  1.63it/s]
 46%|████▌     | 6152/13317 [1:03:24<1:13:21,  1.63it/s]
                                                        
{'loss': 0.2058, 'grad_norm': 0.461168110370636, 'learning_rate': 6.5126406469211055e-06, 'epoch': 1.39}

 46%|████▌     | 6152/13317 [1:03:24<1:13:21,  1.63it/s]
 46%|████▌     | 6153/13317 [1:03:25<1:13:23,  1.63it/s]
                                                        
{'loss': 0.1819, 'grad_norm': 0.46047720313072205, 'learning_rate': 6.5113913755031085e-06, 'epoch': 1.39}

 46%|████▌     | 6153/13317 [1:03:25<1:13:23,  1.63it/s]
 46%|████▌     | 6154/13317 [1:03:25<1:13:27,  1.63it/s]
                                                        
{'loss': 0.2455, 'grad_norm': 0.6072326302528381, 'learning_rate': 6.510142000236521e-06, 'epoch': 1.39}

 46%|████▌     | 6154/13317 [1:03:26<1:13:27,  1.63it/s]
 46%|████▌     | 6155/13317 [1:03:26<1:13:23,  1.63it/s]
                                                        
{'loss': 0.1817, 'grad_norm': 0.4485973119735718, 'learning_rate': 6.508892521207191e-06, 'epoch': 1.39}

 46%|████▌     | 6155/13317 [1:03:26<1:13:23,  1.63it/s]
 46%|████▌     | 6156/13317 [1:03:27<1:13:20,  1.63it/s]
                                                        
{'loss': 0.1652, 'grad_norm': 0.45094382762908936, 'learning_rate': 6.507642938500966e-06, 'epoch': 1.39}

 46%|████▌     | 6156/13317 [1:03:27<1:13:20,  1.63it/s]
 46%|████▌     | 6157/13317 [1:03:27<1:13:15,  1.63it/s]
                                                        
{'loss': 0.1755, 'grad_norm': 0.45338064432144165, 'learning_rate': 6.506393252203709e-06, 'epoch': 1.39}

 46%|████▌     | 6157/13317 [1:03:27<1:13:15,  1.63it/s]
 46%|████▌     | 6158/13317 [1:03:28<1:13:19,  1.63it/s]
                                                        
{'loss': 0.2237, 'grad_norm': 0.5157650113105774, 'learning_rate': 6.505143462401284e-06, 'epoch': 1.39}

 46%|████▌     | 6158/13317 [1:03:28<1:13:19,  1.63it/s]
 46%|████▌     | 6159/13317 [1:03:29<1:13:25,  1.62it/s]
                                                        
{'loss': 0.1734, 'grad_norm': 0.4531031548976898, 'learning_rate': 6.503893569179568e-06, 'epoch': 1.39}

 46%|████▌     | 6159/13317 [1:03:29<1:13:25,  1.62it/s]
 46%|████▋     | 6160/13317 [1:03:29<1:13:23,  1.63it/s]
                                                        
{'loss': 0.1731, 'grad_norm': 0.4275969862937927, 'learning_rate': 6.502643572624439e-06, 'epoch': 1.39}

 46%|████▋     | 6160/13317 [1:03:29<1:13:23,  1.63it/s]
 46%|████▋     | 6161/13317 [1:03:30<1:13:21,  1.63it/s]
                                                        
{'loss': 0.2188, 'grad_norm': 0.5409558415412903, 'learning_rate': 6.5013934728217864e-06, 'epoch': 1.39}

 46%|████▋     | 6161/13317 [1:03:30<1:13:21,  1.63it/s]
 46%|████▋     | 6162/13317 [1:03:30<1:13:19,  1.63it/s]
                                                        
{'loss': 0.1707, 'grad_norm': 0.4701395034790039, 'learning_rate': 6.500143269857507e-06, 'epoch': 1.39}

 46%|████▋     | 6162/13317 [1:03:30<1:13:19,  1.63it/s]
 46%|████▋     | 6163/13317 [1:03:31<1:13:19,  1.63it/s]
                                                        
{'loss': 0.1732, 'grad_norm': 0.4532282054424286, 'learning_rate': 6.498892963817499e-06, 'epoch': 1.39}

 46%|████▋     | 6163/13317 [1:03:31<1:13:19,  1.63it/s]
 46%|████▋     | 6164/13317 [1:03:32<1:13:22,  1.62it/s]
                                                        
{'loss': 0.1712, 'grad_norm': 0.5806214213371277, 'learning_rate': 6.4976425547876765e-06, 'epoch': 1.39}

 46%|████▋     | 6164/13317 [1:03:32<1:13:22,  1.62it/s]
 46%|████▋     | 6165/13317 [1:03:32<1:13:16,  1.63it/s]
                                                        
{'loss': 0.1823, 'grad_norm': 0.47021037340164185, 'learning_rate': 6.49639204285395e-06, 'epoch': 1.39}

 46%|████▋     | 6165/13317 [1:03:32<1:13:16,  1.63it/s]
 46%|████▋     | 6166/13317 [1:03:33<1:13:15,  1.63it/s]
                                                        
{'loss': 0.2937, 'grad_norm': 0.6729739308357239, 'learning_rate': 6.4951414281022485e-06, 'epoch': 1.39}

 46%|████▋     | 6166/13317 [1:03:33<1:13:15,  1.63it/s]
 46%|████▋     | 6167/13317 [1:03:33<1:13:12,  1.63it/s]
                                                        
{'loss': 0.2581, 'grad_norm': 0.5275590419769287, 'learning_rate': 6.4938907106184985e-06, 'epoch': 1.39}

 46%|████▋     | 6167/13317 [1:03:34<1:13:12,  1.63it/s]
 46%|████▋     | 6168/13317 [1:03:34<1:13:14,  1.63it/s]
                                                        
{'loss': 0.1469, 'grad_norm': 0.42427948117256165, 'learning_rate': 6.492639890488641e-06, 'epoch': 1.39}

 46%|████▋     | 6168/13317 [1:03:34<1:13:14,  1.63it/s]
 46%|████▋     | 6169/13317 [1:03:35<1:13:16,  1.63it/s]
                                                        
{'loss': 0.18, 'grad_norm': 0.4896298050880432, 'learning_rate': 6.491388967798617e-06, 'epoch': 1.39}

 46%|████▋     | 6169/13317 [1:03:35<1:13:16,  1.63it/s]
 46%|████▋     | 6170/13317 [1:03:35<1:13:14,  1.63it/s]
                                                        
{'loss': 0.1978, 'grad_norm': 0.5183624625205994, 'learning_rate': 6.49013794263438e-06, 'epoch': 1.39}

 46%|████▋     | 6170/13317 [1:03:35<1:13:14,  1.63it/s]
 46%|████▋     | 6171/13317 [1:03:36<1:13:08,  1.63it/s]
                                                        
{'loss': 0.2704, 'grad_norm': 0.6589681506156921, 'learning_rate': 6.488886815081889e-06, 'epoch': 1.39}

 46%|████▋     | 6171/13317 [1:03:36<1:13:08,  1.63it/s]
 46%|████▋     | 6172/13317 [1:03:36<1:13:07,  1.63it/s]
                                                        
{'loss': 0.1772, 'grad_norm': 0.46323609352111816, 'learning_rate': 6.48763558522711e-06, 'epoch': 1.39}

 46%|████▋     | 6172/13317 [1:03:37<1:13:07,  1.63it/s]
 46%|████▋     | 6173/13317 [1:03:37<1:13:08,  1.63it/s]
                                                        
{'loss': 0.2108, 'grad_norm': 0.571451723575592, 'learning_rate': 6.486384253156014e-06, 'epoch': 1.39}

 46%|████▋     | 6173/13317 [1:03:37<1:13:08,  1.63it/s]
 46%|████▋     | 6174/13317 [1:03:38<1:13:06,  1.63it/s]
                                                        
{'loss': 0.1589, 'grad_norm': 0.45009180903434753, 'learning_rate': 6.485132818954583e-06, 'epoch': 1.39}

 46%|████▋     | 6174/13317 [1:03:38<1:13:06,  1.63it/s]
 46%|████▋     | 6175/13317 [1:03:38<1:13:04,  1.63it/s]
                                                        
{'loss': 0.1909, 'grad_norm': 0.543372392654419, 'learning_rate': 6.4838812827088e-06, 'epoch': 1.39}

 46%|████▋     | 6175/13317 [1:03:38<1:13:04,  1.63it/s]
 46%|████▋     | 6176/13317 [1:03:39<1:13:03,  1.63it/s]
                                                        
{'loss': 0.2162, 'grad_norm': 0.5376476049423218, 'learning_rate': 6.482629644504663e-06, 'epoch': 1.39}

 46%|████▋     | 6176/13317 [1:03:39<1:13:03,  1.63it/s]
 46%|████▋     | 6177/13317 [1:03:40<1:13:03,  1.63it/s]
                                                        
{'loss': 0.2373, 'grad_norm': 0.6079692840576172, 'learning_rate': 6.481377904428171e-06, 'epoch': 1.39}

 46%|████▋     | 6177/13317 [1:03:40<1:13:03,  1.63it/s]
 46%|████▋     | 6178/13317 [1:03:40<1:13:08,  1.63it/s]
                                                        
{'loss': 0.1696, 'grad_norm': 0.4381043612957001, 'learning_rate': 6.480126062565331e-06, 'epoch': 1.39}

 46%|████▋     | 6178/13317 [1:03:40<1:13:08,  1.63it/s]
 46%|████▋     | 6179/13317 [1:03:41<1:13:10,  1.63it/s]
                                                        
{'loss': 0.2505, 'grad_norm': 0.6725428700447083, 'learning_rate': 6.47887411900216e-06, 'epoch': 1.39}

 46%|████▋     | 6179/13317 [1:03:41<1:13:10,  1.63it/s]
 46%|████▋     | 6180/13317 [1:03:41<1:13:06,  1.63it/s]
                                                        
{'loss': 0.2866, 'grad_norm': 0.7066778540611267, 'learning_rate': 6.477622073824678e-06, 'epoch': 1.39}

 46%|████▋     | 6180/13317 [1:03:42<1:13:06,  1.63it/s]
 46%|████▋     | 6181/13317 [1:03:42<1:13:03,  1.63it/s]
                                                        
{'loss': 0.2153, 'grad_norm': 0.5454913377761841, 'learning_rate': 6.476369927118913e-06, 'epoch': 1.39}

 46%|████▋     | 6181/13317 [1:03:42<1:13:03,  1.63it/s]
 46%|████▋     | 6182/13317 [1:03:43<1:13:02,  1.63it/s]
                                                        
{'loss': 0.1839, 'grad_norm': 0.4763433039188385, 'learning_rate': 6.475117678970904e-06, 'epoch': 1.39}

 46%|████▋     | 6182/13317 [1:03:43<1:13:02,  1.63it/s]
 46%|████▋     | 6183/13317 [1:03:43<1:13:05,  1.63it/s]
                                                        
{'loss': 0.2314, 'grad_norm': 0.5703057646751404, 'learning_rate': 6.47386532946669e-06, 'epoch': 1.39}

 46%|████▋     | 6183/13317 [1:03:43<1:13:05,  1.63it/s]
 46%|████▋     | 6184/13317 [1:03:44<1:13:06,  1.63it/s]
                                                        
{'loss': 0.2063, 'grad_norm': 0.5666283369064331, 'learning_rate': 6.472612878692324e-06, 'epoch': 1.39}

 46%|████▋     | 6184/13317 [1:03:44<1:13:06,  1.63it/s]
 46%|████▋     | 6185/13317 [1:03:44<1:13:02,  1.63it/s]
                                                        
{'loss': 0.2081, 'grad_norm': 0.4657284915447235, 'learning_rate': 6.471360326733857e-06, 'epoch': 1.39}

 46%|████▋     | 6185/13317 [1:03:45<1:13:02,  1.63it/s]
 46%|████▋     | 6186/13317 [1:03:45<1:12:59,  1.63it/s]
                                                        
{'loss': 0.2164, 'grad_norm': 0.5339168906211853, 'learning_rate': 6.4701076736773615e-06, 'epoch': 1.39}

 46%|████▋     | 6186/13317 [1:03:45<1:12:59,  1.63it/s]
 46%|████▋     | 6187/13317 [1:03:46<1:12:56,  1.63it/s]
                                                        
{'loss': 0.1857, 'grad_norm': 0.5063907504081726, 'learning_rate': 6.4688549196089e-06, 'epoch': 1.39}

 46%|████▋     | 6187/13317 [1:03:46<1:12:56,  1.63it/s]
 46%|████▋     | 6188/13317 [1:03:46<1:13:02,  1.63it/s]
                                                        
{'loss': 0.2334, 'grad_norm': 0.5156970620155334, 'learning_rate': 6.467602064614554e-06, 'epoch': 1.39}

 46%|████▋     | 6188/13317 [1:03:46<1:13:02,  1.63it/s]
 46%|████▋     | 6189/13317 [1:03:47<1:13:09,  1.62it/s]
                                                        
{'loss': 0.1579, 'grad_norm': 0.41933345794677734, 'learning_rate': 6.466349108780408e-06, 'epoch': 1.39}

 46%|████▋     | 6189/13317 [1:03:47<1:13:09,  1.62it/s]
 46%|████▋     | 6190/13317 [1:03:48<1:13:02,  1.63it/s]
                                                        
{'loss': 0.1922, 'grad_norm': 0.5149126052856445, 'learning_rate': 6.46509605219255e-06, 'epoch': 1.39}

 46%|████▋     | 6190/13317 [1:03:48<1:13:02,  1.63it/s]
 46%|████▋     | 6191/13317 [1:03:48<1:12:58,  1.63it/s]
                                                        
{'loss': 0.2134, 'grad_norm': 0.5535634756088257, 'learning_rate': 6.4638428949370825e-06, 'epoch': 1.39}

 46%|████▋     | 6191/13317 [1:03:48<1:12:58,  1.63it/s]
 46%|████▋     | 6192/13317 [1:03:49<1:12:54,  1.63it/s]
                                                        
{'loss': 0.2161, 'grad_norm': 0.5231108665466309, 'learning_rate': 6.462589637100108e-06, 'epoch': 1.39}

 46%|████▋     | 6192/13317 [1:03:49<1:12:54,  1.63it/s]
 47%|████▋     | 6193/13317 [1:03:50<1:16:40,  1.55it/s]
                                                        
{'loss': 0.2124, 'grad_norm': 0.5426623821258545, 'learning_rate': 6.461336278767739e-06, 'epoch': 1.4}

 47%|████▋     | 6193/13317 [1:03:50<1:16:40,  1.55it/s]
 47%|████▋     | 6194/13317 [1:03:50<1:16:04,  1.56it/s]
                                                        
{'loss': 0.1994, 'grad_norm': 0.48436036705970764, 'learning_rate': 6.460082820026095e-06, 'epoch': 1.4}

 47%|████▋     | 6194/13317 [1:03:50<1:16:04,  1.56it/s]
 47%|████▋     | 6195/13317 [1:03:51<1:15:14,  1.58it/s]
                                                        
{'loss': 0.1638, 'grad_norm': 0.5537126660346985, 'learning_rate': 6.458829260961302e-06, 'epoch': 1.4}

 47%|████▋     | 6195/13317 [1:03:51<1:15:14,  1.58it/s]
 47%|████▋     | 6196/13317 [1:03:51<1:14:32,  1.59it/s]
                                                        
{'loss': 0.1753, 'grad_norm': 0.5310096144676208, 'learning_rate': 6.457575601659491e-06, 'epoch': 1.4}

 47%|████▋     | 6196/13317 [1:03:51<1:14:32,  1.59it/s]
 47%|████▋     | 6197/13317 [1:03:52<1:14:02,  1.60it/s]
                                                        
{'loss': 0.2123, 'grad_norm': 0.4474889934062958, 'learning_rate': 6.456321842206805e-06, 'epoch': 1.4}

 47%|████▋     | 6197/13317 [1:03:52<1:14:02,  1.60it/s]
 47%|████▋     | 6198/13317 [1:03:53<1:13:44,  1.61it/s]
                                                        
{'loss': 0.2217, 'grad_norm': 0.5616071820259094, 'learning_rate': 6.455067982689388e-06, 'epoch': 1.4}

 47%|████▋     | 6198/13317 [1:03:53<1:13:44,  1.61it/s]
 47%|████▋     | 6199/13317 [1:03:53<1:17:08,  1.54it/s]
                                                        
{'loss': 0.188, 'grad_norm': 0.5385044813156128, 'learning_rate': 6.453814023193393e-06, 'epoch': 1.4}

 47%|████▋     | 6199/13317 [1:03:53<1:17:08,  1.54it/s]
 47%|████▋     | 6200/13317 [1:03:54<1:15:50,  1.56it/s]
                                                        
{'loss': 0.1883, 'grad_norm': 0.5040098428726196, 'learning_rate': 6.452559963804983e-06, 'epoch': 1.4}

 47%|████▋     | 6200/13317 [1:03:54<1:15:50,  1.56it/s]
 47%|████▋     | 6201/13317 [1:03:55<1:14:55,  1.58it/s]
                                                        
{'loss': 0.2024, 'grad_norm': 0.44053682684898376, 'learning_rate': 6.451305804610323e-06, 'epoch': 1.4}

 47%|████▋     | 6201/13317 [1:03:55<1:14:55,  1.58it/s]
 47%|████▋     | 6202/13317 [1:03:55<1:14:20,  1.60it/s]
                                                        
{'loss': 0.1879, 'grad_norm': 0.5089235901832581, 'learning_rate': 6.450051545695587e-06, 'epoch': 1.4}

 47%|████▋     | 6202/13317 [1:03:55<1:14:20,  1.60it/s]
 47%|████▋     | 6203/13317 [1:03:56<1:13:53,  1.60it/s]
                                                        
{'loss': 0.2256, 'grad_norm': 0.609637439250946, 'learning_rate': 6.448797187146957e-06, 'epoch': 1.4}

 47%|████▋     | 6203/13317 [1:03:56<1:13:53,  1.60it/s]
 47%|████▋     | 6204/13317 [1:03:56<1:13:31,  1.61it/s]
                                                        
{'loss': 0.1239, 'grad_norm': 0.3659555912017822, 'learning_rate': 6.4475427290506195e-06, 'epoch': 1.4}

 47%|████▋     | 6204/13317 [1:03:57<1:13:31,  1.61it/s]
 47%|████▋     | 6205/13317 [1:03:57<1:16:56,  1.54it/s]
                                                        
{'loss': 0.231, 'grad_norm': 0.5737963914871216, 'learning_rate': 6.446288171492771e-06, 'epoch': 1.4}

 47%|████▋     | 6205/13317 [1:03:57<1:16:56,  1.54it/s]
 47%|████▋     | 6206/13317 [1:03:58<1:15:46,  1.56it/s]
                                                        
{'loss': 0.2329, 'grad_norm': 0.6521291732788086, 'learning_rate': 6.445033514559611e-06, 'epoch': 1.4}

 47%|████▋     | 6206/13317 [1:03:58<1:15:46,  1.56it/s]
 47%|████▋     | 6207/13317 [1:03:58<1:14:50,  1.58it/s]
                                                        
{'loss': 0.209, 'grad_norm': 0.5343982577323914, 'learning_rate': 6.443778758337348e-06, 'epoch': 1.4}

 47%|████▋     | 6207/13317 [1:03:58<1:14:50,  1.58it/s]
 47%|████▋     | 6208/13317 [1:03:59<1:14:09,  1.60it/s]
                                                        
{'loss': 0.1943, 'grad_norm': 0.5659050941467285, 'learning_rate': 6.442523902912196e-06, 'epoch': 1.4}

 47%|████▋     | 6208/13317 [1:03:59<1:14:09,  1.60it/s]
 47%|████▋     | 6209/13317 [1:04:00<1:13:43,  1.61it/s]
                                                        
{'loss': 0.2237, 'grad_norm': 0.5305304527282715, 'learning_rate': 6.4412689483703805e-06, 'epoch': 1.4}

 47%|████▋     | 6209/13317 [1:04:00<1:13:43,  1.61it/s]
 47%|████▋     | 6210/13317 [1:04:00<1:13:23,  1.61it/s]
                                                        
{'loss': 0.2272, 'grad_norm': 0.5571351647377014, 'learning_rate': 6.440013894798127e-06, 'epoch': 1.4}

 47%|████▋     | 6210/13317 [1:04:00<1:13:23,  1.61it/s]
 47%|████▋     | 6211/13317 [1:04:01<1:13:14,  1.62it/s]
                                                        
{'loss': 0.2262, 'grad_norm': 0.5502349138259888, 'learning_rate': 6.438758742281672e-06, 'epoch': 1.4}

 47%|████▋     | 6211/13317 [1:04:01<1:13:14,  1.62it/s]
 47%|████▋     | 6212/13317 [1:04:01<1:13:03,  1.62it/s]
                                                        
{'loss': 0.2288, 'grad_norm': 0.5558403134346008, 'learning_rate': 6.4375034909072565e-06, 'epoch': 1.4}

 47%|████▋     | 6212/13317 [1:04:02<1:13:03,  1.62it/s]
 47%|████▋     | 6213/13317 [1:04:02<1:12:55,  1.62it/s]
                                                        
{'loss': 0.1776, 'grad_norm': 0.5098369717597961, 'learning_rate': 6.436248140761132e-06, 'epoch': 1.4}

 47%|████▋     | 6213/13317 [1:04:02<1:12:55,  1.62it/s]
 47%|████▋     | 6214/13317 [1:04:03<1:12:48,  1.63it/s]
                                                        
{'loss': 0.198, 'grad_norm': 0.501220166683197, 'learning_rate': 6.434992691929553e-06, 'epoch': 1.4}

 47%|████▋     | 6214/13317 [1:04:03<1:12:48,  1.63it/s]
 47%|████▋     | 6215/13317 [1:04:03<1:12:45,  1.63it/s]
                                                        
{'loss': 0.1849, 'grad_norm': 0.5154014825820923, 'learning_rate': 6.43373714449878e-06, 'epoch': 1.4}

 47%|████▋     | 6215/13317 [1:04:03<1:12:45,  1.63it/s]
 47%|████▋     | 6216/13317 [1:04:04<1:12:44,  1.63it/s]
                                                        
{'loss': 0.1903, 'grad_norm': 0.5141704678535461, 'learning_rate': 6.432481498555086e-06, 'epoch': 1.4}

 47%|████▋     | 6216/13317 [1:04:04<1:12:44,  1.63it/s]
 47%|████▋     | 6217/13317 [1:04:04<1:12:39,  1.63it/s]
                                                        
{'loss': 0.1744, 'grad_norm': 0.45304784178733826, 'learning_rate': 6.431225754184747e-06, 'epoch': 1.4}

 47%|████▋     | 6217/13317 [1:04:05<1:12:39,  1.63it/s]
 47%|████▋     | 6218/13317 [1:04:05<1:12:39,  1.63it/s]
                                                        
{'loss': 0.1858, 'grad_norm': 0.4606352150440216, 'learning_rate': 6.429969911474043e-06, 'epoch': 1.4}

 47%|████▋     | 6218/13317 [1:04:05<1:12:39,  1.63it/s]
 47%|████▋     | 6219/13317 [1:04:06<1:12:37,  1.63it/s]
                                                        
{'loss': 0.1869, 'grad_norm': 0.5801331996917725, 'learning_rate': 6.428713970509265e-06, 'epoch': 1.4}

 47%|████▋     | 6219/13317 [1:04:06<1:12:37,  1.63it/s]
 47%|████▋     | 6220/13317 [1:04:06<1:12:35,  1.63it/s]
                                                        
{'loss': 0.2059, 'grad_norm': 0.5312549471855164, 'learning_rate': 6.427457931376712e-06, 'epoch': 1.4}

 47%|████▋     | 6220/13317 [1:04:06<1:12:35,  1.63it/s]
 47%|████▋     | 6221/13317 [1:04:07<1:12:33,  1.63it/s]
                                                        
{'loss': 0.1792, 'grad_norm': 0.4642898440361023, 'learning_rate': 6.426201794162681e-06, 'epoch': 1.4}

 47%|████▋     | 6221/13317 [1:04:07<1:12:33,  1.63it/s]
 47%|████▋     | 6222/13317 [1:04:08<1:12:33,  1.63it/s]
                                                        
{'loss': 0.2146, 'grad_norm': 0.4921635687351227, 'learning_rate': 6.424945558953489e-06, 'epoch': 1.4}

 47%|████▋     | 6222/13317 [1:04:08<1:12:33,  1.63it/s]
 47%|████▋     | 6223/13317 [1:04:08<1:12:33,  1.63it/s]
                                                        
{'loss': 0.2999, 'grad_norm': 0.6311932802200317, 'learning_rate': 6.42368922583545e-06, 'epoch': 1.4}

 47%|████▋     | 6223/13317 [1:04:08<1:12:33,  1.63it/s]
 47%|████▋     | 6224/13317 [1:04:09<1:12:42,  1.63it/s]
                                                        
{'loss': 0.1951, 'grad_norm': 0.5758881568908691, 'learning_rate': 6.422432794894885e-06, 'epoch': 1.4}

 47%|████▋     | 6224/13317 [1:04:09<1:12:42,  1.63it/s]
 47%|████▋     | 6225/13317 [1:04:09<1:12:44,  1.63it/s]
                                                        
{'loss': 0.1481, 'grad_norm': 0.41567108035087585, 'learning_rate': 6.421176266218128e-06, 'epoch': 1.4}

 47%|████▋     | 6225/13317 [1:04:10<1:12:44,  1.63it/s]
 47%|████▋     | 6226/13317 [1:04:10<1:12:43,  1.62it/s]
                                                        
{'loss': 0.2415, 'grad_norm': 0.5653963088989258, 'learning_rate': 6.4199196398915106e-06, 'epoch': 1.4}

 47%|████▋     | 6226/13317 [1:04:10<1:12:43,  1.62it/s]
 47%|████▋     | 6227/13317 [1:04:11<1:12:59,  1.62it/s]
                                                        
{'loss': 0.1987, 'grad_norm': 0.45966118574142456, 'learning_rate': 6.418662916001382e-06, 'epoch': 1.4}

 47%|████▋     | 6227/13317 [1:04:11<1:12:59,  1.62it/s]
 47%|████▋     | 6228/13317 [1:04:11<1:12:49,  1.62it/s]
                                                        
{'loss': 0.2189, 'grad_norm': 0.5424683690071106, 'learning_rate': 6.41740609463409e-06, 'epoch': 1.4}

 47%|████▋     | 6228/13317 [1:04:11<1:12:49,  1.62it/s]
 47%|████▋     | 6229/13317 [1:04:12<1:12:47,  1.62it/s]
                                                        
{'loss': 0.3053, 'grad_norm': 0.47814393043518066, 'learning_rate': 6.416149175875992e-06, 'epoch': 1.4}

 47%|████▋     | 6229/13317 [1:04:12<1:12:47,  1.62it/s]
 47%|████▋     | 6230/13317 [1:04:12<1:12:45,  1.62it/s]
                                                        
{'loss': 0.2123, 'grad_norm': 0.5600505471229553, 'learning_rate': 6.414892159813449e-06, 'epoch': 1.4}

 47%|████▋     | 6230/13317 [1:04:13<1:12:45,  1.62it/s]
 47%|████▋     | 6231/13317 [1:04:13<1:13:06,  1.62it/s]
                                                        
{'loss': 0.1911, 'grad_norm': 0.5227470397949219, 'learning_rate': 6.4136350465328356e-06, 'epoch': 1.4}

 47%|████▋     | 6231/13317 [1:04:13<1:13:06,  1.62it/s]
 47%|████▋     | 6232/13317 [1:04:14<1:12:55,  1.62it/s]
                                                        
{'loss': 0.2217, 'grad_norm': 0.553460955619812, 'learning_rate': 6.4123778361205255e-06, 'epoch': 1.4}

 47%|████▋     | 6232/13317 [1:04:14<1:12:55,  1.62it/s]
 47%|████▋     | 6233/13317 [1:04:14<1:12:46,  1.62it/s]
                                                        
{'loss': 0.2058, 'grad_norm': 0.5323300361633301, 'learning_rate': 6.411120528662905e-06, 'epoch': 1.4}

 47%|████▋     | 6233/13317 [1:04:14<1:12:46,  1.62it/s]
 47%|████▋     | 6234/13317 [1:04:15<1:12:39,  1.62it/s]
                                                        
{'loss': 0.2444, 'grad_norm': 0.5981665253639221, 'learning_rate': 6.409863124246362e-06, 'epoch': 1.4}

 47%|████▋     | 6234/13317 [1:04:15<1:12:39,  1.62it/s]
 47%|████▋     | 6235/13317 [1:04:16<1:12:40,  1.62it/s]
                                                        
{'loss': 0.1757, 'grad_norm': 0.4628134071826935, 'learning_rate': 6.408605622957293e-06, 'epoch': 1.4}

 47%|████▋     | 6235/13317 [1:04:16<1:12:40,  1.62it/s]
 47%|████▋     | 6236/13317 [1:04:16<1:12:42,  1.62it/s]
                                                        
{'loss': 0.1679, 'grad_norm': 0.4729425311088562, 'learning_rate': 6.407348024882105e-06, 'epoch': 1.4}

 47%|████▋     | 6236/13317 [1:04:16<1:12:42,  1.62it/s]
 47%|████▋     | 6237/13317 [1:04:17<1:12:35,  1.63it/s]
                                                        
{'loss': 0.1792, 'grad_norm': 0.48425164818763733, 'learning_rate': 6.406090330107206e-06, 'epoch': 1.41}

 47%|████▋     | 6237/13317 [1:04:17<1:12:35,  1.63it/s]
 47%|████▋     | 6238/13317 [1:04:17<1:12:32,  1.63it/s]
                                                        
{'loss': 0.1862, 'grad_norm': 0.46407514810562134, 'learning_rate': 6.404832538719015e-06, 'epoch': 1.41}

 47%|████▋     | 6238/13317 [1:04:18<1:12:32,  1.63it/s]
 47%|████▋     | 6239/13317 [1:04:18<1:12:29,  1.63it/s]
                                                        
{'loss': 0.1827, 'grad_norm': 0.4774889647960663, 'learning_rate': 6.403574650803952e-06, 'epoch': 1.41}

 47%|████▋     | 6239/13317 [1:04:18<1:12:29,  1.63it/s]
 47%|████▋     | 6240/13317 [1:04:19<1:12:24,  1.63it/s]
                                                        
{'loss': 0.1697, 'grad_norm': 0.4871351718902588, 'learning_rate': 6.40231666644845e-06, 'epoch': 1.41}

 47%|████▋     | 6240/13317 [1:04:19<1:12:24,  1.63it/s]
 47%|████▋     | 6241/13317 [1:04:19<1:12:28,  1.63it/s]
                                                        
{'loss': 0.1831, 'grad_norm': 0.47966793179512024, 'learning_rate': 6.401058585738944e-06, 'epoch': 1.41}

 47%|████▋     | 6241/13317 [1:04:19<1:12:28,  1.63it/s]
 47%|████▋     | 6242/13317 [1:04:20<1:12:28,  1.63it/s]
                                                        
{'loss': 0.1879, 'grad_norm': 0.519503116607666, 'learning_rate': 6.399800408761881e-06, 'epoch': 1.41}

 47%|████▋     | 6242/13317 [1:04:20<1:12:28,  1.63it/s]
 47%|████▋     | 6243/13317 [1:04:20<1:12:26,  1.63it/s]
                                                        
{'loss': 0.3271, 'grad_norm': 0.5704113841056824, 'learning_rate': 6.398542135603706e-06, 'epoch': 1.41}

 47%|████▋     | 6243/13317 [1:04:21<1:12:26,  1.63it/s]
 47%|████▋     | 6244/13317 [1:04:21<1:12:26,  1.63it/s]
                                                        
{'loss': 0.2074, 'grad_norm': 0.5289967656135559, 'learning_rate': 6.3972837663508795e-06, 'epoch': 1.41}

 47%|████▋     | 6244/13317 [1:04:21<1:12:26,  1.63it/s]
 47%|████▋     | 6245/13317 [1:04:22<1:12:25,  1.63it/s]
                                                        
{'loss': 0.1819, 'grad_norm': 0.43163737654685974, 'learning_rate': 6.396025301089863e-06, 'epoch': 1.41}

 47%|████▋     | 6245/13317 [1:04:22<1:12:25,  1.63it/s]
 47%|████▋     | 6246/13317 [1:04:22<1:12:32,  1.62it/s]
                                                        
{'loss': 0.194, 'grad_norm': 0.5666007995605469, 'learning_rate': 6.394766739907126e-06, 'epoch': 1.41}

 47%|████▋     | 6246/13317 [1:04:22<1:12:32,  1.62it/s]
 47%|████▋     | 6247/13317 [1:04:23<1:12:28,  1.63it/s]
                                                        
{'loss': 0.2538, 'grad_norm': 0.5943984389305115, 'learning_rate': 6.3935080828891484e-06, 'epoch': 1.41}

 47%|████▋     | 6247/13317 [1:04:23<1:12:28,  1.63it/s]
 47%|████▋     | 6248/13317 [1:04:24<1:12:24,  1.63it/s]
                                                        
{'loss': 0.1976, 'grad_norm': 0.5792598128318787, 'learning_rate': 6.3922493301224085e-06, 'epoch': 1.41}

 47%|████▋     | 6248/13317 [1:04:24<1:12:24,  1.63it/s]
 47%|████▋     | 6249/13317 [1:04:24<1:12:22,  1.63it/s]
                                                        
{'loss': 0.1718, 'grad_norm': 0.4627407491207123, 'learning_rate': 6.3909904816934e-06, 'epoch': 1.41}

 47%|████▋     | 6249/13317 [1:04:24<1:12:22,  1.63it/s]
 47%|████▋     | 6250/13317 [1:04:25<1:12:21,  1.63it/s]
                                                        
{'loss': 0.2003, 'grad_norm': 0.5240364074707031, 'learning_rate': 6.389731537688616e-06, 'epoch': 1.41}

 47%|████▋     | 6250/13317 [1:04:25<1:12:21,  1.63it/s]
 47%|████▋     | 6251/13317 [1:04:25<1:12:39,  1.62it/s]
                                                        
{'loss': 0.2121, 'grad_norm': 0.6628172993659973, 'learning_rate': 6.388472498194561e-06, 'epoch': 1.41}

 47%|████▋     | 6251/13317 [1:04:26<1:12:39,  1.62it/s]
 47%|████▋     | 6252/13317 [1:04:26<1:12:31,  1.62it/s]
                                                        
{'loss': 0.1635, 'grad_norm': 0.5227164030075073, 'learning_rate': 6.387213363297746e-06, 'epoch': 1.41}

 47%|████▋     | 6252/13317 [1:04:26<1:12:31,  1.62it/s]
 47%|████▋     | 6253/13317 [1:04:27<1:12:26,  1.63it/s]
                                                        
{'loss': 0.2067, 'grad_norm': 0.5205746293067932, 'learning_rate': 6.385954133084682e-06, 'epoch': 1.41}

 47%|████▋     | 6253/13317 [1:04:27<1:12:26,  1.63it/s]
 47%|████▋     | 6254/13317 [1:04:27<1:12:23,  1.63it/s]
                                                        
{'loss': 0.2343, 'grad_norm': 0.5783048272132874, 'learning_rate': 6.3846948076418955e-06, 'epoch': 1.41}

 47%|████▋     | 6254/13317 [1:04:27<1:12:23,  1.63it/s]
 47%|████▋     | 6255/13317 [1:04:28<1:12:19,  1.63it/s]
                                                        
{'loss': 0.2244, 'grad_norm': 0.5412430763244629, 'learning_rate': 6.383435387055914e-06, 'epoch': 1.41}

 47%|████▋     | 6255/13317 [1:04:28<1:12:19,  1.63it/s]
 47%|████▋     | 6256/13317 [1:04:28<1:12:28,  1.62it/s]
                                                        
{'loss': 0.186, 'grad_norm': 0.48044779896736145, 'learning_rate': 6.382175871413274e-06, 'epoch': 1.41}

 47%|████▋     | 6256/13317 [1:04:29<1:12:28,  1.62it/s]
 47%|████▋     | 6257/13317 [1:04:29<1:12:22,  1.63it/s]
                                                        
{'loss': 0.2164, 'grad_norm': 0.6127757430076599, 'learning_rate': 6.380916260800516e-06, 'epoch': 1.41}

 47%|████▋     | 6257/13317 [1:04:29<1:12:22,  1.63it/s]
 47%|████▋     | 6258/13317 [1:04:30<1:12:17,  1.63it/s]
                                                        
{'loss': 0.1936, 'grad_norm': 0.48054078221321106, 'learning_rate': 6.37965655530419e-06, 'epoch': 1.41}

 47%|████▋     | 6258/13317 [1:04:30<1:12:17,  1.63it/s]
 47%|████▋     | 6259/13317 [1:04:30<1:12:16,  1.63it/s]
                                                        
{'loss': 0.1596, 'grad_norm': 0.3884889483451843, 'learning_rate': 6.378396755010851e-06, 'epoch': 1.41}

 47%|████▋     | 6259/13317 [1:04:30<1:12:16,  1.63it/s]
 47%|████▋     | 6260/13317 [1:04:31<1:12:17,  1.63it/s]
                                                        
{'loss': 0.2357, 'grad_norm': 0.5561718940734863, 'learning_rate': 6.377136860007061e-06, 'epoch': 1.41}

 47%|████▋     | 6260/13317 [1:04:31<1:12:17,  1.63it/s]
 47%|████▋     | 6261/13317 [1:04:32<1:12:21,  1.63it/s]
                                                        
{'loss': 0.1999, 'grad_norm': 0.47295066714286804, 'learning_rate': 6.3758768703793865e-06, 'epoch': 1.41}

 47%|████▋     | 6261/13317 [1:04:32<1:12:21,  1.63it/s]
 47%|████▋     | 6262/13317 [1:04:32<1:12:17,  1.63it/s]
                                                        
{'loss': 0.18, 'grad_norm': 0.49488428235054016, 'learning_rate': 6.374616786214402e-06, 'epoch': 1.41}

 47%|████▋     | 6262/13317 [1:04:32<1:12:17,  1.63it/s]
 47%|████▋     | 6263/13317 [1:04:33<1:12:13,  1.63it/s]
                                                        
{'loss': 0.208, 'grad_norm': 0.5319304466247559, 'learning_rate': 6.373356607598691e-06, 'epoch': 1.41}

 47%|████▋     | 6263/13317 [1:04:33<1:12:13,  1.63it/s]
 47%|████▋     | 6264/13317 [1:04:33<1:12:10,  1.63it/s]
                                                        
{'loss': 0.1562, 'grad_norm': 0.4151991009712219, 'learning_rate': 6.372096334618839e-06, 'epoch': 1.41}

 47%|████▋     | 6264/13317 [1:04:34<1:12:10,  1.63it/s]
 47%|████▋     | 6265/13317 [1:04:34<1:12:08,  1.63it/s]
                                                        
{'loss': 0.2393, 'grad_norm': 0.533207893371582, 'learning_rate': 6.370835967361441e-06, 'epoch': 1.41}

 47%|████▋     | 6265/13317 [1:04:34<1:12:08,  1.63it/s]
 47%|████▋     | 6266/13317 [1:04:35<1:12:10,  1.63it/s]
                                                        
{'loss': 0.181, 'grad_norm': 0.4734135866165161, 'learning_rate': 6.369575505913097e-06, 'epoch': 1.41}

 47%|████▋     | 6266/13317 [1:04:35<1:12:10,  1.63it/s]
 47%|████▋     | 6267/13317 [1:04:35<1:12:08,  1.63it/s]
                                                        
{'loss': 0.1754, 'grad_norm': 0.45648834109306335, 'learning_rate': 6.368314950360416e-06, 'epoch': 1.41}

 47%|████▋     | 6267/13317 [1:04:35<1:12:08,  1.63it/s]
 47%|████▋     | 6268/13317 [1:04:36<1:12:07,  1.63it/s]
                                                        
{'loss': 0.1889, 'grad_norm': 0.4780542254447937, 'learning_rate': 6.3670543007900075e-06, 'epoch': 1.41}

 47%|████▋     | 6268/13317 [1:04:36<1:12:07,  1.63it/s]
 47%|████▋     | 6269/13317 [1:04:36<1:12:05,  1.63it/s]
                                                        
{'loss': 0.1522, 'grad_norm': 0.39077240228652954, 'learning_rate': 6.3657935572884955e-06, 'epoch': 1.41}

 47%|████▋     | 6269/13317 [1:04:37<1:12:05,  1.63it/s]
 47%|████▋     | 6270/13317 [1:04:37<1:12:05,  1.63it/s]
                                                        
{'loss': 0.1868, 'grad_norm': 0.4842846095561981, 'learning_rate': 6.364532719942504e-06, 'epoch': 1.41}

 47%|████▋     | 6270/13317 [1:04:37<1:12:05,  1.63it/s]
 47%|████▋     | 6271/13317 [1:04:38<1:12:10,  1.63it/s]
                                                        
{'loss': 0.1784, 'grad_norm': 0.46457111835479736, 'learning_rate': 6.363271788838667e-06, 'epoch': 1.41}

 47%|████▋     | 6271/13317 [1:04:38<1:12:10,  1.63it/s]
 47%|████▋     | 6272/13317 [1:04:38<1:12:09,  1.63it/s]
                                                        
{'loss': 0.2291, 'grad_norm': 0.7573326230049133, 'learning_rate': 6.362010764063626e-06, 'epoch': 1.41}

 47%|████▋     | 6272/13317 [1:04:38<1:12:09,  1.63it/s]
 47%|████▋     | 6273/13317 [1:04:39<1:12:06,  1.63it/s]
                                                        
{'loss': 0.2053, 'grad_norm': 0.5723636150360107, 'learning_rate': 6.360749645704021e-06, 'epoch': 1.41}

 47%|████▋     | 6273/13317 [1:04:39<1:12:06,  1.63it/s]
 47%|████▋     | 6274/13317 [1:04:40<1:12:05,  1.63it/s]
                                                        
{'loss': 0.1987, 'grad_norm': 0.5378570556640625, 'learning_rate': 6.359488433846509e-06, 'epoch': 1.41}

 47%|████▋     | 6274/13317 [1:04:40<1:12:05,  1.63it/s]
 47%|████▋     | 6275/13317 [1:04:40<1:12:05,  1.63it/s]
                                                        
{'loss': 0.1563, 'grad_norm': 0.41064053773880005, 'learning_rate': 6.358227128577746e-06, 'epoch': 1.41}

 47%|████▋     | 6275/13317 [1:04:40<1:12:05,  1.63it/s]
 47%|████▋     | 6276/13317 [1:04:41<1:12:10,  1.63it/s]
                                                        
{'loss': 0.1592, 'grad_norm': 0.4249448776245117, 'learning_rate': 6.356965729984399e-06, 'epoch': 1.41}

 47%|████▋     | 6276/13317 [1:04:41<1:12:10,  1.63it/s]
 47%|████▋     | 6277/13317 [1:04:41<1:12:06,  1.63it/s]
                                                        
{'loss': 0.1734, 'grad_norm': 0.49060365557670593, 'learning_rate': 6.355704238153139e-06, 'epoch': 1.41}

 47%|████▋     | 6277/13317 [1:04:41<1:12:06,  1.63it/s]
 47%|████▋     | 6278/13317 [1:04:42<1:12:03,  1.63it/s]
                                                        
{'loss': 0.1788, 'grad_norm': 0.4609062969684601, 'learning_rate': 6.3544426531706425e-06, 'epoch': 1.41}

 47%|████▋     | 6278/13317 [1:04:42<1:12:03,  1.63it/s]
 47%|████▋     | 6279/13317 [1:04:43<1:11:59,  1.63it/s]
                                                        
{'loss': 0.2012, 'grad_norm': 0.5433367490768433, 'learning_rate': 6.353180975123595e-06, 'epoch': 1.41}

 47%|████▋     | 6279/13317 [1:04:43<1:11:59,  1.63it/s]
 47%|████▋     | 6280/13317 [1:04:43<1:11:58,  1.63it/s]
                                                        
{'loss': 0.2744, 'grad_norm': 0.6046909689903259, 'learning_rate': 6.351919204098687e-06, 'epoch': 1.41}

 47%|████▋     | 6280/13317 [1:04:43<1:11:58,  1.63it/s]
 47%|████▋     | 6281/13317 [1:04:44<1:12:02,  1.63it/s]
                                                        
{'loss': 0.2078, 'grad_norm': 0.5305758714675903, 'learning_rate': 6.350657340182615e-06, 'epoch': 1.41}

 47%|████▋     | 6281/13317 [1:04:44<1:12:02,  1.63it/s]
 47%|████▋     | 6282/13317 [1:04:44<1:12:01,  1.63it/s]
                                                        
{'loss': 0.1726, 'grad_norm': 0.4797252118587494, 'learning_rate': 6.349395383462084e-06, 'epoch': 1.42}

 47%|████▋     | 6282/13317 [1:04:45<1:12:01,  1.63it/s]
 47%|████▋     | 6283/13317 [1:04:45<1:11:59,  1.63it/s]
                                                        
{'loss': 0.1666, 'grad_norm': 0.4257938861846924, 'learning_rate': 6.3481333340238e-06, 'epoch': 1.42}

 47%|████▋     | 6283/13317 [1:04:45<1:11:59,  1.63it/s]
 47%|████▋     | 6284/13317 [1:04:46<1:11:58,  1.63it/s]
                                                        
{'loss': 0.1931, 'grad_norm': 0.4789767563343048, 'learning_rate': 6.346871191954484e-06, 'epoch': 1.42}

 47%|████▋     | 6284/13317 [1:04:46<1:11:58,  1.63it/s]
 47%|████▋     | 6285/13317 [1:04:46<1:11:56,  1.63it/s]
                                                        
{'loss': 0.2129, 'grad_norm': 0.6272796392440796, 'learning_rate': 6.3456089573408554e-06, 'epoch': 1.42}

 47%|████▋     | 6285/13317 [1:04:46<1:11:56,  1.63it/s]
 47%|████▋     | 6286/13317 [1:04:47<1:12:03,  1.63it/s]
                                                        
{'loss': 0.1646, 'grad_norm': 0.4079553782939911, 'learning_rate': 6.344346630269643e-06, 'epoch': 1.42}

 47%|████▋     | 6286/13317 [1:04:47<1:12:03,  1.63it/s]
 47%|████▋     | 6287/13317 [1:04:48<1:11:58,  1.63it/s]
                                                        
{'loss': 0.2388, 'grad_norm': 0.6574755907058716, 'learning_rate': 6.343084210827584e-06, 'epoch': 1.42}

 47%|████▋     | 6287/13317 [1:04:48<1:11:58,  1.63it/s]
 47%|████▋     | 6288/13317 [1:04:48<1:11:58,  1.63it/s]
                                                        
{'loss': 0.168, 'grad_norm': 0.4862302839756012, 'learning_rate': 6.3418216991014195e-06, 'epoch': 1.42}

 47%|████▋     | 6288/13317 [1:04:48<1:11:58,  1.63it/s]
 47%|████▋     | 6289/13317 [1:04:49<1:11:57,  1.63it/s]
                                                        
{'loss': 0.1976, 'grad_norm': 0.5353889465332031, 'learning_rate': 6.340559095177896e-06, 'epoch': 1.42}

 47%|████▋     | 6289/13317 [1:04:49<1:11:57,  1.63it/s]
 47%|████▋     | 6290/13317 [1:04:49<1:11:54,  1.63it/s]
                                                        
{'loss': 0.1938, 'grad_norm': 0.4702991545200348, 'learning_rate': 6.339296399143768e-06, 'epoch': 1.42}

 47%|████▋     | 6290/13317 [1:04:49<1:11:54,  1.63it/s]
 47%|████▋     | 6291/13317 [1:04:50<1:11:54,  1.63it/s]
                                                        
{'loss': 0.1978, 'grad_norm': 0.4963800311088562, 'learning_rate': 6.338033611085798e-06, 'epoch': 1.42}

 47%|████▋     | 6291/13317 [1:04:50<1:11:54,  1.63it/s]
 47%|████▋     | 6292/13317 [1:04:51<1:11:51,  1.63it/s]
                                                        
{'loss': 0.1828, 'grad_norm': 0.4803071618080139, 'learning_rate': 6.33677073109075e-06, 'epoch': 1.42}

 47%|████▋     | 6292/13317 [1:04:51<1:11:51,  1.63it/s]
 47%|████▋     | 6293/13317 [1:04:51<1:11:51,  1.63it/s]
                                                        
{'loss': 0.1775, 'grad_norm': 0.49015334248542786, 'learning_rate': 6.3355077592453995e-06, 'epoch': 1.42}

 47%|████▋     | 6293/13317 [1:04:51<1:11:51,  1.63it/s]
 47%|████▋     | 6294/13317 [1:04:52<1:11:50,  1.63it/s]
                                                        
{'loss': 0.184, 'grad_norm': 0.502303421497345, 'learning_rate': 6.334244695636527e-06, 'epoch': 1.42}

 47%|████▋     | 6294/13317 [1:04:52<1:11:50,  1.63it/s]
 47%|████▋     | 6295/13317 [1:04:52<1:11:49,  1.63it/s]
                                                        
{'loss': 0.1418, 'grad_norm': 0.3848872780799866, 'learning_rate': 6.332981540350913e-06, 'epoch': 1.42}

 47%|████▋     | 6295/13317 [1:04:53<1:11:49,  1.63it/s]
 47%|████▋     | 6296/13317 [1:04:53<1:11:53,  1.63it/s]
                                                        
{'loss': 0.2162, 'grad_norm': 0.5322238206863403, 'learning_rate': 6.3317182934753575e-06, 'epoch': 1.42}

 47%|████▋     | 6296/13317 [1:04:53<1:11:53,  1.63it/s]
 47%|████▋     | 6297/13317 [1:04:54<1:11:48,  1.63it/s]
                                                        
{'loss': 0.2124, 'grad_norm': 0.563052773475647, 'learning_rate': 6.330454955096652e-06, 'epoch': 1.42}

 47%|████▋     | 6297/13317 [1:04:54<1:11:48,  1.63it/s]
 47%|████▋     | 6298/13317 [1:04:54<1:11:50,  1.63it/s]
                                                        
{'loss': 0.204, 'grad_norm': 0.5264509916305542, 'learning_rate': 6.3291915253016055e-06, 'epoch': 1.42}

 47%|████▋     | 6298/13317 [1:04:54<1:11:50,  1.63it/s]
 47%|████▋     | 6299/13317 [1:04:55<1:11:48,  1.63it/s]
                                                        
{'loss': 0.1882, 'grad_norm': 0.4925963580608368, 'learning_rate': 6.327928004177028e-06, 'epoch': 1.42}

 47%|████▋     | 6299/13317 [1:04:55<1:11:48,  1.63it/s]
 47%|████▋     | 6300/13317 [1:04:55<1:11:47,  1.63it/s]
                                                        
{'loss': 0.2121, 'grad_norm': 0.5615347027778625, 'learning_rate': 6.326664391809734e-06, 'epoch': 1.42}

 47%|████▋     | 6300/13317 [1:04:56<1:11:47,  1.63it/s]
 47%|████▋     | 6301/13317 [1:04:56<1:11:45,  1.63it/s]
                                                        
{'loss': 0.221, 'grad_norm': 0.5166786313056946, 'learning_rate': 6.325400688286551e-06, 'epoch': 1.42}

 47%|████▋     | 6301/13317 [1:04:56<1:11:45,  1.63it/s]
 47%|████▋     | 6302/13317 [1:04:57<1:11:46,  1.63it/s]
                                                        
{'loss': 0.1664, 'grad_norm': 0.43303728103637695, 'learning_rate': 6.3241368936943045e-06, 'epoch': 1.42}

 47%|████▋     | 6302/13317 [1:04:57<1:11:46,  1.63it/s]
 47%|████▋     | 6303/13317 [1:04:57<1:11:44,  1.63it/s]
                                                        
{'loss': 0.2224, 'grad_norm': 0.5715180039405823, 'learning_rate': 6.322873008119836e-06, 'epoch': 1.42}

 47%|████▋     | 6303/13317 [1:04:57<1:11:44,  1.63it/s]
 47%|████▋     | 6304/13317 [1:04:58<1:11:42,  1.63it/s]
                                                        
{'loss': 0.144, 'grad_norm': 0.42213064432144165, 'learning_rate': 6.321609031649984e-06, 'epoch': 1.42}

 47%|████▋     | 6304/13317 [1:04:58<1:11:42,  1.63it/s]
 47%|████▋     | 6305/13317 [1:04:59<1:11:42,  1.63it/s]
                                                        
{'loss': 0.2198, 'grad_norm': 0.4926610291004181, 'learning_rate': 6.3203449643716e-06, 'epoch': 1.42}

 47%|████▋     | 6305/13317 [1:04:59<1:11:42,  1.63it/s]
 47%|████▋     | 6306/13317 [1:04:59<1:11:42,  1.63it/s]
                                                        
{'loss': 0.2159, 'grad_norm': 0.5390424132347107, 'learning_rate': 6.319080806371535e-06, 'epoch': 1.42}

 47%|████▋     | 6306/13317 [1:04:59<1:11:42,  1.63it/s]
 47%|████▋     | 6307/13317 [1:05:00<1:11:39,  1.63it/s]
                                                        
{'loss': 0.2054, 'grad_norm': 0.4938482344150543, 'learning_rate': 6.317816557736652e-06, 'epoch': 1.42}

 47%|████▋     | 6307/13317 [1:05:00<1:11:39,  1.63it/s]
 47%|████▋     | 6308/13317 [1:05:00<1:11:37,  1.63it/s]
                                                        
{'loss': 0.2471, 'grad_norm': 0.5863372087478638, 'learning_rate': 6.31655221855382e-06, 'epoch': 1.42}

 47%|████▋     | 6308/13317 [1:05:01<1:11:37,  1.63it/s]
 47%|████▋     | 6309/13317 [1:05:01<1:11:36,  1.63it/s]
                                                        
{'loss': 0.163, 'grad_norm': 0.4586273431777954, 'learning_rate': 6.315287788909907e-06, 'epoch': 1.42}

 47%|████▋     | 6309/13317 [1:05:01<1:11:36,  1.63it/s]
 47%|████▋     | 6310/13317 [1:05:02<1:11:35,  1.63it/s]
                                                        
{'loss': 0.2315, 'grad_norm': 0.5534414052963257, 'learning_rate': 6.3140232688918005e-06, 'epoch': 1.42}

 47%|████▋     | 6310/13317 [1:05:02<1:11:35,  1.63it/s]
 47%|████▋     | 6311/13317 [1:05:02<1:11:37,  1.63it/s]
                                                        
{'loss': 0.2044, 'grad_norm': 0.502237856388092, 'learning_rate': 6.312758658586379e-06, 'epoch': 1.42}

 47%|████▋     | 6311/13317 [1:05:02<1:11:37,  1.63it/s]
 47%|████▋     | 6312/13317 [1:05:03<1:11:38,  1.63it/s]
                                                        
{'loss': 0.1896, 'grad_norm': 0.5311893820762634, 'learning_rate': 6.31149395808054e-06, 'epoch': 1.42}

 47%|████▋     | 6312/13317 [1:05:03<1:11:38,  1.63it/s]
 47%|████▋     | 6313/13317 [1:05:03<1:11:36,  1.63it/s]
                                                        
{'loss': 0.1607, 'grad_norm': 0.4708309471607208, 'learning_rate': 6.3102291674611796e-06, 'epoch': 1.42}

 47%|████▋     | 6313/13317 [1:05:04<1:11:36,  1.63it/s]
 47%|████▋     | 6314/13317 [1:05:04<1:11:36,  1.63it/s]
                                                        
{'loss': 0.1936, 'grad_norm': 0.5288280248641968, 'learning_rate': 6.308964286815203e-06, 'epoch': 1.42}

 47%|████▋     | 6314/13317 [1:05:04<1:11:36,  1.63it/s]
 47%|████▋     | 6315/13317 [1:05:05<1:11:34,  1.63it/s]
                                                        
{'loss': 0.2129, 'grad_norm': 0.536774218082428, 'learning_rate': 6.307699316229518e-06, 'epoch': 1.42}

 47%|████▋     | 6315/13317 [1:05:05<1:11:34,  1.63it/s]
 47%|████▋     | 6316/13317 [1:05:05<1:11:35,  1.63it/s]
                                                        
{'loss': 0.1805, 'grad_norm': 0.5631079077720642, 'learning_rate': 6.306434255791046e-06, 'epoch': 1.42}

 47%|████▋     | 6316/13317 [1:05:05<1:11:35,  1.63it/s]
 47%|████▋     | 6317/13317 [1:05:06<1:11:35,  1.63it/s]
                                                        
{'loss': 0.2358, 'grad_norm': 0.5642367601394653, 'learning_rate': 6.305169105586707e-06, 'epoch': 1.42}

 47%|████▋     | 6317/13317 [1:05:06<1:11:35,  1.63it/s]
 47%|████▋     | 6318/13317 [1:05:07<1:11:35,  1.63it/s]
                                                        
{'loss': 0.2406, 'grad_norm': 0.625156044960022, 'learning_rate': 6.303903865703431e-06, 'epoch': 1.42}

 47%|████▋     | 6318/13317 [1:05:07<1:11:35,  1.63it/s]
 47%|████▋     | 6319/13317 [1:05:07<1:11:35,  1.63it/s]
                                                        
{'loss': 0.2139, 'grad_norm': 0.5348389744758606, 'learning_rate': 6.302638536228154e-06, 'epoch': 1.42}

 47%|████▋     | 6319/13317 [1:05:07<1:11:35,  1.63it/s]
 47%|████▋     | 6320/13317 [1:05:08<1:11:34,  1.63it/s]
                                                        
{'loss': 0.2001, 'grad_norm': 0.5412058234214783, 'learning_rate': 6.3013731172478166e-06, 'epoch': 1.42}

 47%|████▋     | 6320/13317 [1:05:08<1:11:34,  1.63it/s]
 47%|████▋     | 6321/13317 [1:05:08<1:11:34,  1.63it/s]
                                                        
{'loss': 0.1771, 'grad_norm': 0.5008030533790588, 'learning_rate': 6.300107608849369e-06, 'epoch': 1.42}

 47%|████▋     | 6321/13317 [1:05:09<1:11:34,  1.63it/s]
 47%|████▋     | 6322/13317 [1:05:09<1:11:32,  1.63it/s]
                                                        
{'loss': 0.1544, 'grad_norm': 0.4233092665672302, 'learning_rate': 6.298842011119759e-06, 'epoch': 1.42}

 47%|████▋     | 6322/13317 [1:05:09<1:11:32,  1.63it/s]
 47%|████▋     | 6323/13317 [1:05:10<1:11:32,  1.63it/s]
                                                        
{'loss': 0.2108, 'grad_norm': 0.5960335731506348, 'learning_rate': 6.297576324145953e-06, 'epoch': 1.42}

 47%|████▋     | 6323/13317 [1:05:10<1:11:32,  1.63it/s]
 47%|████▋     | 6324/13317 [1:05:10<1:11:34,  1.63it/s]
                                                        
{'loss': 0.2064, 'grad_norm': 0.5117935538291931, 'learning_rate': 6.2963105480149145e-06, 'epoch': 1.42}

 47%|████▋     | 6324/13317 [1:05:10<1:11:34,  1.63it/s]
 47%|████▋     | 6325/13317 [1:05:11<1:11:31,  1.63it/s]
                                                        
{'loss': 0.1995, 'grad_norm': 0.5480663776397705, 'learning_rate': 6.295044682813615e-06, 'epoch': 1.42}

 47%|████▋     | 6325/13317 [1:05:11<1:11:31,  1.63it/s]
 48%|████▊     | 6326/13317 [1:05:11<1:11:29,  1.63it/s]
                                                        
{'loss': 0.1268, 'grad_norm': 0.34533432126045227, 'learning_rate': 6.2937787286290335e-06, 'epoch': 1.43}

 48%|████▊     | 6326/13317 [1:05:12<1:11:29,  1.63it/s]
 48%|████▊     | 6327/13317 [1:05:12<1:11:32,  1.63it/s]
                                                        
{'loss': 0.1821, 'grad_norm': 0.48893874883651733, 'learning_rate': 6.2925126855481545e-06, 'epoch': 1.43}

 48%|████▊     | 6327/13317 [1:05:12<1:11:32,  1.63it/s]
 48%|████▊     | 6328/13317 [1:05:13<1:11:30,  1.63it/s]
                                                        
{'loss': 0.2892, 'grad_norm': 0.5834617018699646, 'learning_rate': 6.291246553657971e-06, 'epoch': 1.43}

 48%|████▊     | 6328/13317 [1:05:13<1:11:30,  1.63it/s]
 48%|████▊     | 6329/13317 [1:05:13<1:11:27,  1.63it/s]
                                                        
{'loss': 0.1914, 'grad_norm': 0.4384794235229492, 'learning_rate': 6.2899803330454734e-06, 'epoch': 1.43}

 48%|████▊     | 6329/13317 [1:05:13<1:11:27,  1.63it/s]
 48%|████▊     | 6330/13317 [1:05:14<1:11:25,  1.63it/s]
                                                        
{'loss': 0.2171, 'grad_norm': 0.510819673538208, 'learning_rate': 6.2887140237976714e-06, 'epoch': 1.43}

 48%|████▊     | 6330/13317 [1:05:14<1:11:25,  1.63it/s]
 48%|████▊     | 6331/13317 [1:05:15<1:11:27,  1.63it/s]
                                                        
{'loss': 0.1586, 'grad_norm': 0.48098036646842957, 'learning_rate': 6.28744762600157e-06, 'epoch': 1.43}

 48%|████▊     | 6331/13317 [1:05:15<1:11:27,  1.63it/s]
 48%|████▊     | 6332/13317 [1:05:15<1:11:28,  1.63it/s]
                                                        
{'loss': 0.2666, 'grad_norm': 0.560812771320343, 'learning_rate': 6.286181139744185e-06, 'epoch': 1.43}

 48%|████▊     | 6332/13317 [1:05:15<1:11:28,  1.63it/s]
 48%|████▊     | 6333/13317 [1:05:16<1:11:29,  1.63it/s]
                                                        
{'loss': 0.1858, 'grad_norm': 0.451817125082016, 'learning_rate': 6.284914565112537e-06, 'epoch': 1.43}

 48%|████▊     | 6333/13317 [1:05:16<1:11:29,  1.63it/s]
 48%|████▊     | 6334/13317 [1:05:16<1:11:27,  1.63it/s]
                                                        
{'loss': 0.1806, 'grad_norm': 0.4863416850566864, 'learning_rate': 6.2836479021936535e-06, 'epoch': 1.43}

 48%|████▊     | 6334/13317 [1:05:16<1:11:27,  1.63it/s]
 48%|████▊     | 6335/13317 [1:05:17<1:11:25,  1.63it/s]
                                                        
{'loss': 0.1752, 'grad_norm': 0.4870012700557709, 'learning_rate': 6.2823811510745695e-06, 'epoch': 1.43}

 48%|████▊     | 6335/13317 [1:05:17<1:11:25,  1.63it/s]
 48%|████▊     | 6336/13317 [1:05:18<1:11:25,  1.63it/s]
                                                        
{'loss': 0.1678, 'grad_norm': 0.5313838124275208, 'learning_rate': 6.281114311842319e-06, 'epoch': 1.43}

 48%|████▊     | 6336/13317 [1:05:18<1:11:25,  1.63it/s]
 48%|████▊     | 6337/13317 [1:05:18<1:11:23,  1.63it/s]
                                                        
{'loss': 0.1395, 'grad_norm': 0.43688565492630005, 'learning_rate': 6.279847384583953e-06, 'epoch': 1.43}

 48%|████▊     | 6337/13317 [1:05:18<1:11:23,  1.63it/s]
 48%|████▊     | 6338/13317 [1:05:19<1:11:20,  1.63it/s]
                                                        
{'loss': 0.2046, 'grad_norm': 0.5397348999977112, 'learning_rate': 6.27858036938652e-06, 'epoch': 1.43}

 48%|████▊     | 6338/13317 [1:05:19<1:11:20,  1.63it/s]
 48%|████▊     | 6339/13317 [1:05:19<1:11:22,  1.63it/s]
                                                        
{'loss': 0.1936, 'grad_norm': 0.5516456961631775, 'learning_rate': 6.277313266337077e-06, 'epoch': 1.43}

 48%|████▊     | 6339/13317 [1:05:20<1:11:22,  1.63it/s]
 48%|████▊     | 6340/13317 [1:05:20<1:11:21,  1.63it/s]
                                                        
{'loss': 0.2117, 'grad_norm': 0.5924142599105835, 'learning_rate': 6.276046075522689e-06, 'epoch': 1.43}

 48%|████▊     | 6340/13317 [1:05:20<1:11:21,  1.63it/s]
 48%|████▊     | 6341/13317 [1:05:21<1:11:20,  1.63it/s]
                                                        
{'loss': 0.2617, 'grad_norm': 0.6905509233474731, 'learning_rate': 6.274778797030424e-06, 'epoch': 1.43}

 48%|████▊     | 6341/13317 [1:05:21<1:11:20,  1.63it/s]
 48%|████▊     | 6342/13317 [1:05:21<1:11:19,  1.63it/s]
                                                        
{'loss': 0.1812, 'grad_norm': 0.5072506070137024, 'learning_rate': 6.27351143094736e-06, 'epoch': 1.43}

 48%|████▊     | 6342/13317 [1:05:21<1:11:19,  1.63it/s]
 48%|████▊     | 6343/13317 [1:05:22<1:11:16,  1.63it/s]
                                                        
{'loss': 0.1808, 'grad_norm': 0.5606675744056702, 'learning_rate': 6.272243977360575e-06, 'epoch': 1.43}

 48%|████▊     | 6343/13317 [1:05:22<1:11:16,  1.63it/s]
 48%|████▊     | 6344/13317 [1:05:22<1:11:17,  1.63it/s]
                                                        
{'loss': 0.2423, 'grad_norm': 0.6233714818954468, 'learning_rate': 6.270976436357158e-06, 'epoch': 1.43}

 48%|████▊     | 6344/13317 [1:05:23<1:11:17,  1.63it/s]
 48%|████▊     | 6345/13317 [1:05:23<1:11:15,  1.63it/s]
                                                        
{'loss': 0.1948, 'grad_norm': 0.520324170589447, 'learning_rate': 6.2697088080242015e-06, 'epoch': 1.43}

 48%|████▊     | 6345/13317 [1:05:23<1:11:15,  1.63it/s]
 48%|████▊     | 6346/13317 [1:05:24<1:11:15,  1.63it/s]
                                                        
{'loss': 0.2113, 'grad_norm': 0.5677404999732971, 'learning_rate': 6.268441092448807e-06, 'epoch': 1.43}

 48%|████▊     | 6346/13317 [1:05:24<1:11:15,  1.63it/s]
 48%|████▊     | 6347/13317 [1:05:24<1:11:14,  1.63it/s]
                                                        
{'loss': 0.1954, 'grad_norm': 0.5364057421684265, 'learning_rate': 6.267173289718079e-06, 'epoch': 1.43}

 48%|████▊     | 6347/13317 [1:05:24<1:11:14,  1.63it/s]
 48%|████▊     | 6348/13317 [1:05:25<1:11:15,  1.63it/s]
                                                        
{'loss': 0.1891, 'grad_norm': 0.4903767704963684, 'learning_rate': 6.26590539991913e-06, 'epoch': 1.43}

 48%|████▊     | 6348/13317 [1:05:25<1:11:15,  1.63it/s]
 48%|████▊     | 6349/13317 [1:05:26<1:11:15,  1.63it/s]
                                                        
{'loss': 0.2207, 'grad_norm': 0.5881667733192444, 'learning_rate': 6.264637423139075e-06, 'epoch': 1.43}

 48%|████▊     | 6349/13317 [1:05:26<1:11:15,  1.63it/s]
 48%|████▊     | 6350/13317 [1:05:26<1:11:15,  1.63it/s]
                                                        
{'loss': 0.1836, 'grad_norm': 0.4724927544593811, 'learning_rate': 6.263369359465039e-06, 'epoch': 1.43}

 48%|████▊     | 6350/13317 [1:05:26<1:11:15,  1.63it/s]
 48%|████▊     | 6351/13317 [1:05:27<1:11:16,  1.63it/s]
                                                        
{'loss': 0.1839, 'grad_norm': 0.56012362241745, 'learning_rate': 6.262101208984153e-06, 'epoch': 1.43}

 48%|████▊     | 6351/13317 [1:05:27<1:11:16,  1.63it/s]
 48%|████▊     | 6352/13317 [1:05:27<1:11:17,  1.63it/s]
                                                        
{'loss': 0.2425, 'grad_norm': 0.5432025194168091, 'learning_rate': 6.260832971783548e-06, 'epoch': 1.43}

 48%|████▊     | 6352/13317 [1:05:28<1:11:17,  1.63it/s]
 48%|████▊     | 6353/13317 [1:05:28<1:11:13,  1.63it/s]
                                                        
{'loss': 0.1861, 'grad_norm': 0.481161892414093, 'learning_rate': 6.259564647950369e-06, 'epoch': 1.43}

 48%|████▊     | 6353/13317 [1:05:28<1:11:13,  1.63it/s]
 48%|████▊     | 6354/13317 [1:05:29<1:11:12,  1.63it/s]
                                                        
{'loss': 0.2049, 'grad_norm': 0.5634961724281311, 'learning_rate': 6.258296237571761e-06, 'epoch': 1.43}

 48%|████▊     | 6354/13317 [1:05:29<1:11:12,  1.63it/s]
 48%|████▊     | 6355/13317 [1:05:29<1:11:12,  1.63it/s]
                                                        
{'loss': 0.2464, 'grad_norm': 0.629857063293457, 'learning_rate': 6.257027740734881e-06, 'epoch': 1.43}

 48%|████▊     | 6355/13317 [1:05:29<1:11:12,  1.63it/s]
 48%|████▊     | 6356/13317 [1:05:30<1:11:10,  1.63it/s]
                                                        
{'loss': 0.1842, 'grad_norm': 0.5183464288711548, 'learning_rate': 6.2557591575268836e-06, 'epoch': 1.43}

 48%|████▊     | 6356/13317 [1:05:30<1:11:10,  1.63it/s]
 48%|████▊     | 6357/13317 [1:05:30<1:11:11,  1.63it/s]
                                                        
{'loss': 0.2026, 'grad_norm': 0.556102991104126, 'learning_rate': 6.254490488034938e-06, 'epoch': 1.43}

 48%|████▊     | 6357/13317 [1:05:31<1:11:11,  1.63it/s]
 48%|████▊     | 6358/13317 [1:05:31<1:11:10,  1.63it/s]
                                                        
{'loss': 0.1796, 'grad_norm': 0.45272403955459595, 'learning_rate': 6.25322173234621e-06, 'epoch': 1.43}

 48%|████▊     | 6358/13317 [1:05:31<1:11:10,  1.63it/s]
 48%|████▊     | 6359/13317 [1:05:32<1:11:09,  1.63it/s]
                                                        
{'loss': 0.1836, 'grad_norm': 0.4788232147693634, 'learning_rate': 6.2519528905478825e-06, 'epoch': 1.43}

 48%|████▊     | 6359/13317 [1:05:32<1:11:09,  1.63it/s]
 48%|████▊     | 6360/13317 [1:05:32<1:11:09,  1.63it/s]
                                                        
{'loss': 0.2266, 'grad_norm': 0.64285808801651, 'learning_rate': 6.250683962727133e-06, 'epoch': 1.43}

 48%|████▊     | 6360/13317 [1:05:32<1:11:09,  1.63it/s]
 48%|████▊     | 6361/13317 [1:05:33<1:11:08,  1.63it/s]
                                                        
{'loss': 0.1966, 'grad_norm': 0.4942858815193176, 'learning_rate': 6.249414948971154e-06, 'epoch': 1.43}

 48%|████▊     | 6361/13317 [1:05:33<1:11:08,  1.63it/s]
 48%|████▊     | 6362/13317 [1:05:34<1:11:08,  1.63it/s]
                                                        
{'loss': 0.1751, 'grad_norm': 0.4792081415653229, 'learning_rate': 6.248145849367141e-06, 'epoch': 1.43}

 48%|████▊     | 6362/13317 [1:05:34<1:11:08,  1.63it/s]
 48%|████▊     | 6363/13317 [1:05:34<1:11:05,  1.63it/s]
                                                        
{'loss': 0.1587, 'grad_norm': 0.44530850648880005, 'learning_rate': 6.246876664002289e-06, 'epoch': 1.43}

 48%|████▊     | 6363/13317 [1:05:34<1:11:05,  1.63it/s]
 48%|████▊     | 6364/13317 [1:05:35<1:11:06,  1.63it/s]
                                                        
{'loss': 0.2199, 'grad_norm': 0.5516749620437622, 'learning_rate': 6.2456073929638105e-06, 'epoch': 1.43}

 48%|████▊     | 6364/13317 [1:05:35<1:11:06,  1.63it/s]
 48%|████▊     | 6365/13317 [1:05:35<1:11:03,  1.63it/s]
                                                        
{'loss': 0.1939, 'grad_norm': 0.42481258511543274, 'learning_rate': 6.244338036338915e-06, 'epoch': 1.43}

 48%|████▊     | 6365/13317 [1:05:36<1:11:03,  1.63it/s]
 48%|████▊     | 6366/13317 [1:05:36<1:11:03,  1.63it/s]
                                                        
{'loss': 0.2194, 'grad_norm': 0.5576515793800354, 'learning_rate': 6.243068594214821e-06, 'epoch': 1.43}

 48%|████▊     | 6366/13317 [1:05:36<1:11:03,  1.63it/s]
 48%|████▊     | 6367/13317 [1:05:37<1:11:05,  1.63it/s]
                                                        
{'loss': 0.1766, 'grad_norm': 0.4786950647830963, 'learning_rate': 6.241799066678752e-06, 'epoch': 1.43}

 48%|████▊     | 6367/13317 [1:05:37<1:11:05,  1.63it/s]
 48%|████▊     | 6368/13317 [1:05:37<1:11:02,  1.63it/s]
                                                        
{'loss': 0.1637, 'grad_norm': 0.44018274545669556, 'learning_rate': 6.240529453817938e-06, 'epoch': 1.43}

 48%|████▊     | 6368/13317 [1:05:37<1:11:02,  1.63it/s]
 48%|████▊     | 6369/13317 [1:05:38<1:11:00,  1.63it/s]
                                                        
{'loss': 0.1681, 'grad_norm': 0.4584895968437195, 'learning_rate': 6.2392597557196175e-06, 'epoch': 1.43}

 48%|████▊     | 6369/13317 [1:05:38<1:11:00,  1.63it/s]
 48%|████▊     | 6370/13317 [1:05:38<1:11:02,  1.63it/s]
                                                        
{'loss': 0.1779, 'grad_norm': 0.5076431632041931, 'learning_rate': 6.237989972471029e-06, 'epoch': 1.44}

 48%|████▊     | 6370/13317 [1:05:39<1:11:02,  1.63it/s]
 48%|████▊     | 6371/13317 [1:05:39<1:11:01,  1.63it/s]
                                                        
{'loss': 0.2215, 'grad_norm': 0.5611907243728638, 'learning_rate': 6.2367201041594235e-06, 'epoch': 1.44}

 48%|████▊     | 6371/13317 [1:05:39<1:11:01,  1.63it/s]
 48%|████▊     | 6372/13317 [1:05:40<1:10:59,  1.63it/s]
                                                        
{'loss': 0.1961, 'grad_norm': 0.4969497323036194, 'learning_rate': 6.235450150872051e-06, 'epoch': 1.44}

 48%|████▊     | 6372/13317 [1:05:40<1:10:59,  1.63it/s]
 48%|████▊     | 6373/13317 [1:05:40<1:10:59,  1.63it/s]
                                                        
{'loss': 0.2019, 'grad_norm': 0.5424797534942627, 'learning_rate': 6.234180112696171e-06, 'epoch': 1.44}

 48%|████▊     | 6373/13317 [1:05:40<1:10:59,  1.63it/s]
 48%|████▊     | 6374/13317 [1:05:41<1:10:59,  1.63it/s]
                                                        
{'loss': 0.1929, 'grad_norm': 0.4844137132167816, 'learning_rate': 6.23290998971905e-06, 'epoch': 1.44}

 48%|████▊     | 6374/13317 [1:05:41<1:10:59,  1.63it/s]
 48%|████▊     | 6375/13317 [1:05:42<1:11:00,  1.63it/s]
                                                        
{'loss': 0.2407, 'grad_norm': 0.5940674543380737, 'learning_rate': 6.231639782027958e-06, 'epoch': 1.44}

 48%|████▊     | 6375/13317 [1:05:42<1:11:00,  1.63it/s]
 48%|████▊     | 6376/13317 [1:05:42<1:11:02,  1.63it/s]
                                                        
{'loss': 0.2206, 'grad_norm': 0.6040075421333313, 'learning_rate': 6.230369489710173e-06, 'epoch': 1.44}

 48%|████▊     | 6376/13317 [1:05:42<1:11:02,  1.63it/s]
 48%|████▊     | 6377/13317 [1:05:43<1:11:02,  1.63it/s]
                                                        
{'loss': 0.1628, 'grad_norm': 0.46183985471725464, 'learning_rate': 6.229099112852978e-06, 'epoch': 1.44}

 48%|████▊     | 6377/13317 [1:05:43<1:11:02,  1.63it/s]
 48%|████▊     | 6378/13317 [1:05:43<1:11:00,  1.63it/s]
                                                        
{'loss': 0.2262, 'grad_norm': 0.5775881409645081, 'learning_rate': 6.2278286515436585e-06, 'epoch': 1.44}

 48%|████▊     | 6378/13317 [1:05:43<1:11:00,  1.63it/s]
 48%|████▊     | 6379/13317 [1:05:44<1:10:57,  1.63it/s]
                                                        
{'loss': 0.1686, 'grad_norm': 0.5232610106468201, 'learning_rate': 6.2265581058695104e-06, 'epoch': 1.44}

 48%|████▊     | 6379/13317 [1:05:44<1:10:57,  1.63it/s]
 48%|████▊     | 6380/13317 [1:05:45<1:11:01,  1.63it/s]
                                                        
{'loss': 0.2281, 'grad_norm': 0.5552584528923035, 'learning_rate': 6.225287475917833e-06, 'epoch': 1.44}

 48%|████▊     | 6380/13317 [1:05:45<1:11:01,  1.63it/s]
 48%|████▊     | 6381/13317 [1:05:45<1:11:00,  1.63it/s]
                                                        
{'loss': 0.1462, 'grad_norm': 0.3954775929450989, 'learning_rate': 6.224016761775933e-06, 'epoch': 1.44}

 48%|████▊     | 6381/13317 [1:05:45<1:11:00,  1.63it/s]
 48%|████▊     | 6382/13317 [1:05:46<1:10:58,  1.63it/s]
                                                        
{'loss': 0.2213, 'grad_norm': 0.5181097388267517, 'learning_rate': 6.2227459635311215e-06, 'epoch': 1.44}

 48%|████▊     | 6382/13317 [1:05:46<1:10:58,  1.63it/s]
 48%|████▊     | 6383/13317 [1:05:46<1:10:57,  1.63it/s]
                                                        
{'loss': 0.168, 'grad_norm': 0.4822598397731781, 'learning_rate': 6.221475081270714e-06, 'epoch': 1.44}

 48%|████▊     | 6383/13317 [1:05:47<1:10:57,  1.63it/s]
 48%|████▊     | 6384/13317 [1:05:47<1:10:55,  1.63it/s]
                                                        
{'loss': 0.1785, 'grad_norm': 0.4893747866153717, 'learning_rate': 6.220204115082037e-06, 'epoch': 1.44}

 48%|████▊     | 6384/13317 [1:05:47<1:10:55,  1.63it/s]
 48%|████▊     | 6385/13317 [1:05:48<1:10:54,  1.63it/s]
                                                        
{'loss': 0.1863, 'grad_norm': 0.4888295531272888, 'learning_rate': 6.218933065052416e-06, 'epoch': 1.44}

 48%|████▊     | 6385/13317 [1:05:48<1:10:54,  1.63it/s]
 48%|████▊     | 6386/13317 [1:05:48<1:10:53,  1.63it/s]
                                                        
{'loss': 0.1979, 'grad_norm': 0.5212591290473938, 'learning_rate': 6.21766193126919e-06, 'epoch': 1.44}

 48%|████▊     | 6386/13317 [1:05:48<1:10:53,  1.63it/s]
 48%|████▊     | 6387/13317 [1:05:49<1:10:54,  1.63it/s]
                                                        
{'loss': 0.2717, 'grad_norm': 0.593563437461853, 'learning_rate': 6.216390713819694e-06, 'epoch': 1.44}

 48%|████▊     | 6387/13317 [1:05:49<1:10:54,  1.63it/s]
 48%|████▊     | 6388/13317 [1:05:49<1:10:53,  1.63it/s]
                                                        
{'loss': 0.174, 'grad_norm': 0.44103550910949707, 'learning_rate': 6.2151194127912775e-06, 'epoch': 1.44}

 48%|████▊     | 6388/13317 [1:05:50<1:10:53,  1.63it/s]
 48%|████▊     | 6389/13317 [1:05:50<1:10:51,  1.63it/s]
                                                        
{'loss': 0.1879, 'grad_norm': 0.4600614309310913, 'learning_rate': 6.213848028271292e-06, 'epoch': 1.44}

 48%|████▊     | 6389/13317 [1:05:50<1:10:51,  1.63it/s]
 48%|████▊     | 6390/13317 [1:05:51<1:10:51,  1.63it/s]
                                                        
{'loss': 0.2357, 'grad_norm': 0.6154506206512451, 'learning_rate': 6.212576560347092e-06, 'epoch': 1.44}

 48%|████▊     | 6390/13317 [1:05:51<1:10:51,  1.63it/s]
 48%|████▊     | 6391/13317 [1:05:51<1:10:51,  1.63it/s]
                                                        
{'loss': 0.173, 'grad_norm': 0.45571810007095337, 'learning_rate': 6.211305009106047e-06, 'epoch': 1.44}

 48%|████▊     | 6391/13317 [1:05:51<1:10:51,  1.63it/s]
 48%|████▊     | 6392/13317 [1:05:52<1:10:52,  1.63it/s]
                                                        
{'loss': 0.1739, 'grad_norm': 0.4834192395210266, 'learning_rate': 6.210033374635519e-06, 'epoch': 1.44}

 48%|████▊     | 6392/13317 [1:05:52<1:10:52,  1.63it/s]
 48%|████▊     | 6393/13317 [1:05:53<1:10:49,  1.63it/s]
                                                        
{'loss': 0.191, 'grad_norm': 0.5183947682380676, 'learning_rate': 6.208761657022888e-06, 'epoch': 1.44}

 48%|████▊     | 6393/13317 [1:05:53<1:10:49,  1.63it/s]
 48%|████▊     | 6394/13317 [1:05:53<1:10:49,  1.63it/s]
                                                        
{'loss': 0.1774, 'grad_norm': 0.4619204103946686, 'learning_rate': 6.207489856355533e-06, 'epoch': 1.44}

 48%|████▊     | 6394/13317 [1:05:53<1:10:49,  1.63it/s]
 48%|████▊     | 6395/13317 [1:05:54<1:10:52,  1.63it/s]
                                                        
{'loss': 0.2075, 'grad_norm': 0.5373156070709229, 'learning_rate': 6.206217972720838e-06, 'epoch': 1.44}

 48%|████▊     | 6395/13317 [1:05:54<1:10:52,  1.63it/s]
 48%|████▊     | 6396/13317 [1:05:54<1:10:48,  1.63it/s]
                                                        
{'loss': 0.174, 'grad_norm': 0.44578009843826294, 'learning_rate': 6.204946006206198e-06, 'epoch': 1.44}

 48%|████▊     | 6396/13317 [1:05:55<1:10:48,  1.63it/s]
 48%|████▊     | 6397/13317 [1:05:55<1:10:47,  1.63it/s]
                                                        
{'loss': 0.224, 'grad_norm': 0.5588594675064087, 'learning_rate': 6.203673956899009e-06, 'epoch': 1.44}

 48%|████▊     | 6397/13317 [1:05:55<1:10:47,  1.63it/s]
 48%|████▊     | 6398/13317 [1:05:56<1:10:46,  1.63it/s]
                                                        
{'loss': 0.1803, 'grad_norm': 0.5502091646194458, 'learning_rate': 6.202401824886675e-06, 'epoch': 1.44}

 48%|████▊     | 6398/13317 [1:05:56<1:10:46,  1.63it/s]
 48%|████▊     | 6399/13317 [1:05:56<1:10:44,  1.63it/s]
                                                        
{'loss': 0.1574, 'grad_norm': 0.44322270154953003, 'learning_rate': 6.201129610256603e-06, 'epoch': 1.44}

 48%|████▊     | 6399/13317 [1:05:56<1:10:44,  1.63it/s]
 48%|████▊     | 6400/13317 [1:05:57<1:10:43,  1.63it/s]
                                                        
{'loss': 0.2152, 'grad_norm': 0.4971807599067688, 'learning_rate': 6.1998573130962104e-06, 'epoch': 1.44}

 48%|████▊     | 6400/13317 [1:05:57<1:10:43,  1.63it/s]
 48%|████▊     | 6401/13317 [1:05:57<1:10:42,  1.63it/s]
                                                        
{'loss': 0.145, 'grad_norm': 0.42989060282707214, 'learning_rate': 6.198584933492915e-06, 'epoch': 1.44}

 48%|████▊     | 6401/13317 [1:05:58<1:10:42,  1.63it/s]
 48%|████▊     | 6402/13317 [1:05:58<1:10:40,  1.63it/s]
                                                        
{'loss': 0.2744, 'grad_norm': 0.5546313524246216, 'learning_rate': 6.197312471534144e-06, 'epoch': 1.44}

 48%|████▊     | 6402/13317 [1:05:58<1:10:40,  1.63it/s]
 48%|████▊     | 6403/13317 [1:05:59<1:10:40,  1.63it/s]
                                                        
{'loss': 0.1874, 'grad_norm': 0.5015431642532349, 'learning_rate': 6.196039927307329e-06, 'epoch': 1.44}

 48%|████▊     | 6403/13317 [1:05:59<1:10:40,  1.63it/s]
 48%|████▊     | 6404/13317 [1:05:59<1:10:38,  1.63it/s]
                                                        
{'loss': 0.2037, 'grad_norm': 0.5125389695167542, 'learning_rate': 6.194767300899909e-06, 'epoch': 1.44}

 48%|████▊     | 6404/13317 [1:05:59<1:10:38,  1.63it/s]
 48%|████▊     | 6405/13317 [1:06:00<1:10:40,  1.63it/s]
                                                        
{'loss': 0.2287, 'grad_norm': 0.5743062496185303, 'learning_rate': 6.193494592399323e-06, 'epoch': 1.44}

 48%|████▊     | 6405/13317 [1:06:00<1:10:40,  1.63it/s]
 48%|████▊     | 6406/13317 [1:06:01<1:10:39,  1.63it/s]
                                                        
{'loss': 0.1613, 'grad_norm': 0.4290110468864441, 'learning_rate': 6.192221801893021e-06, 'epoch': 1.44}

 48%|████▊     | 6406/13317 [1:06:01<1:10:39,  1.63it/s]
 48%|████▊     | 6407/13317 [1:06:01<1:10:39,  1.63it/s]
                                                        
{'loss': 0.2124, 'grad_norm': 0.5851753354072571, 'learning_rate': 6.190948929468459e-06, 'epoch': 1.44}

 48%|████▊     | 6407/13317 [1:06:01<1:10:39,  1.63it/s]
 48%|████▊     | 6408/13317 [1:06:02<1:10:38,  1.63it/s]
                                                        
{'loss': 0.2269, 'grad_norm': 0.5640058517456055, 'learning_rate': 6.189675975213094e-06, 'epoch': 1.44}

 48%|████▊     | 6408/13317 [1:06:02<1:10:38,  1.63it/s]
 48%|████▊     | 6409/13317 [1:06:02<1:10:36,  1.63it/s]
                                                        
{'loss': 0.1699, 'grad_norm': 0.5247061252593994, 'learning_rate': 6.1884029392143956e-06, 'epoch': 1.44}

 48%|████▊     | 6409/13317 [1:06:03<1:10:36,  1.63it/s]
 48%|████▊     | 6410/13317 [1:06:03<1:10:36,  1.63it/s]
                                                        
{'loss': 0.1946, 'grad_norm': 0.60521000623703, 'learning_rate': 6.18712982155983e-06, 'epoch': 1.44}

 48%|████▊     | 6410/13317 [1:06:03<1:10:36,  1.63it/s]
 48%|████▊     | 6411/13317 [1:06:04<1:10:36,  1.63it/s]
                                                        
{'loss': 0.1982, 'grad_norm': 0.5427032709121704, 'learning_rate': 6.185856622336878e-06, 'epoch': 1.44}

 48%|████▊     | 6411/13317 [1:06:04<1:10:36,  1.63it/s]
 48%|████▊     | 6412/13317 [1:06:04<1:10:36,  1.63it/s]
                                                        
{'loss': 0.1652, 'grad_norm': 0.43433865904808044, 'learning_rate': 6.184583341633017e-06, 'epoch': 1.44}

 48%|████▊     | 6412/13317 [1:06:04<1:10:36,  1.63it/s]
 48%|████▊     | 6413/13317 [1:06:05<1:10:36,  1.63it/s]
                                                        
{'loss': 0.1849, 'grad_norm': 0.48008790612220764, 'learning_rate': 6.183309979535741e-06, 'epoch': 1.44}

 48%|████▊     | 6413/13317 [1:06:05<1:10:36,  1.63it/s]
 48%|████▊     | 6414/13317 [1:06:05<1:10:34,  1.63it/s]
                                                        
{'loss': 0.1815, 'grad_norm': 0.5857475996017456, 'learning_rate': 6.182036536132539e-06, 'epoch': 1.44}

 48%|████▊     | 6414/13317 [1:06:06<1:10:34,  1.63it/s]
 48%|████▊     | 6415/13317 [1:06:06<1:10:34,  1.63it/s]
                                                        
{'loss': 0.2033, 'grad_norm': 0.5315467119216919, 'learning_rate': 6.180763011510912e-06, 'epoch': 1.45}

 48%|████▊     | 6415/13317 [1:06:06<1:10:34,  1.63it/s]
 48%|████▊     | 6416/13317 [1:06:07<1:10:34,  1.63it/s]
                                                        
{'loss': 0.2117, 'grad_norm': 0.5739481449127197, 'learning_rate': 6.179489405758364e-06, 'epoch': 1.45}

 48%|████▊     | 6416/13317 [1:06:07<1:10:34,  1.63it/s]
 48%|████▊     | 6417/13317 [1:06:07<1:10:35,  1.63it/s]
                                                        
{'loss': 0.2192, 'grad_norm': 0.5462766289710999, 'learning_rate': 6.178215718962404e-06, 'epoch': 1.45}

 48%|████▊     | 6417/13317 [1:06:07<1:10:35,  1.63it/s]
 48%|████▊     | 6418/13317 [1:06:08<1:10:35,  1.63it/s]
                                                        
{'loss': 0.1254, 'grad_norm': 0.3632100224494934, 'learning_rate': 6.17694195121055e-06, 'epoch': 1.45}

 48%|████▊     | 6418/13317 [1:06:08<1:10:35,  1.63it/s]
 48%|████▊     | 6419/13317 [1:06:09<1:10:33,  1.63it/s]
                                                        
{'loss': 0.3061, 'grad_norm': 0.625757098197937, 'learning_rate': 6.175668102590321e-06, 'epoch': 1.45}

 48%|████▊     | 6419/13317 [1:06:09<1:10:33,  1.63it/s]
 48%|████▊     | 6420/13317 [1:06:09<1:10:36,  1.63it/s]
                                                        
{'loss': 0.242, 'grad_norm': 0.6090017557144165, 'learning_rate': 6.174394173189246e-06, 'epoch': 1.45}

 48%|████▊     | 6420/13317 [1:06:09<1:10:36,  1.63it/s]
 48%|████▊     | 6421/13317 [1:06:10<1:10:35,  1.63it/s]
                                                        
{'loss': 0.2162, 'grad_norm': 0.5550070405006409, 'learning_rate': 6.173120163094858e-06, 'epoch': 1.45}

 48%|████▊     | 6421/13317 [1:06:10<1:10:35,  1.63it/s]
 48%|████▊     | 6422/13317 [1:06:10<1:10:31,  1.63it/s]
                                                        
{'loss': 0.1651, 'grad_norm': 0.44683900475502014, 'learning_rate': 6.171846072394691e-06, 'epoch': 1.45}

 48%|████▊     | 6422/13317 [1:06:10<1:10:31,  1.63it/s]
 48%|████▊     | 6423/13317 [1:06:11<1:10:32,  1.63it/s]
                                                        
{'loss': 0.1596, 'grad_norm': 0.4519789516925812, 'learning_rate': 6.170571901176293e-06, 'epoch': 1.45}

 48%|████▊     | 6423/13317 [1:06:11<1:10:32,  1.63it/s]
 48%|████▊     | 6424/13317 [1:06:12<1:10:30,  1.63it/s]
                                                        
{'loss': 0.2536, 'grad_norm': 0.5899850726127625, 'learning_rate': 6.169297649527211e-06, 'epoch': 1.45}

 48%|████▊     | 6424/13317 [1:06:12<1:10:30,  1.63it/s]
 48%|████▊     | 6425/13317 [1:06:12<1:10:28,  1.63it/s]
                                                        
{'loss': 0.1721, 'grad_norm': 0.46217554807662964, 'learning_rate': 6.168023317535e-06, 'epoch': 1.45}

 48%|████▊     | 6425/13317 [1:06:12<1:10:28,  1.63it/s]
 48%|████▊     | 6426/13317 [1:06:13<1:10:27,  1.63it/s]
                                                        
{'loss': 0.1443, 'grad_norm': 0.419450044631958, 'learning_rate': 6.166748905287218e-06, 'epoch': 1.45}

 48%|████▊     | 6426/13317 [1:06:13<1:10:27,  1.63it/s]
 48%|████▊     | 6427/13317 [1:06:13<1:10:25,  1.63it/s]
                                                        
{'loss': 0.1715, 'grad_norm': 0.4997049570083618, 'learning_rate': 6.165474412871435e-06, 'epoch': 1.45}

 48%|████▊     | 6427/13317 [1:06:14<1:10:25,  1.63it/s]
 48%|████▊     | 6428/13317 [1:06:14<1:10:24,  1.63it/s]
                                                        
{'loss': 0.1645, 'grad_norm': 0.4370724558830261, 'learning_rate': 6.164199840375218e-06, 'epoch': 1.45}

 48%|████▊     | 6428/13317 [1:06:14<1:10:24,  1.63it/s]
 48%|████▊     | 6429/13317 [1:06:15<1:10:24,  1.63it/s]
                                                        
{'loss': 0.2156, 'grad_norm': 0.6002161502838135, 'learning_rate': 6.1629251878861454e-06, 'epoch': 1.45}

 48%|████▊     | 6429/13317 [1:06:15<1:10:24,  1.63it/s]
 48%|████▊     | 6430/13317 [1:06:15<1:10:24,  1.63it/s]
                                                        
{'loss': 0.2547, 'grad_norm': 0.5946900248527527, 'learning_rate': 6.1616504554917985e-06, 'epoch': 1.45}

 48%|████▊     | 6430/13317 [1:06:15<1:10:24,  1.63it/s]
 48%|████▊     | 6431/13317 [1:06:16<1:10:23,  1.63it/s]
                                                        
{'loss': 0.1855, 'grad_norm': 0.4833572208881378, 'learning_rate': 6.160375643279767e-06, 'epoch': 1.45}

 48%|████▊     | 6431/13317 [1:06:16<1:10:23,  1.63it/s]
 48%|████▊     | 6432/13317 [1:06:16<1:10:22,  1.63it/s]
                                                        
{'loss': 0.1915, 'grad_norm': 0.6155687570571899, 'learning_rate': 6.1591007513376425e-06, 'epoch': 1.45}

 48%|████▊     | 6432/13317 [1:06:17<1:10:22,  1.63it/s]
 48%|████▊     | 6433/13317 [1:06:17<1:10:20,  1.63it/s]
                                                        
{'loss': 0.2348, 'grad_norm': 0.5816057324409485, 'learning_rate': 6.157825779753023e-06, 'epoch': 1.45}

 48%|████▊     | 6433/13317 [1:06:17<1:10:20,  1.63it/s]
 48%|████▊     | 6434/13317 [1:06:18<1:10:20,  1.63it/s]
                                                        
{'loss': 0.2582, 'grad_norm': 0.7066422700881958, 'learning_rate': 6.156550728613513e-06, 'epoch': 1.45}

 48%|████▊     | 6434/13317 [1:06:18<1:10:20,  1.63it/s]
 48%|████▊     | 6435/13317 [1:06:18<1:10:20,  1.63it/s]
                                                        
{'loss': 0.175, 'grad_norm': 0.46382221579551697, 'learning_rate': 6.155275598006723e-06, 'epoch': 1.45}

 48%|████▊     | 6435/13317 [1:06:18<1:10:20,  1.63it/s]
 48%|████▊     | 6436/13317 [1:06:19<1:10:20,  1.63it/s]
                                                        
{'loss': 0.1708, 'grad_norm': 0.469110369682312, 'learning_rate': 6.154000388020266e-06, 'epoch': 1.45}

 48%|████▊     | 6436/13317 [1:06:19<1:10:20,  1.63it/s]
 48%|████▊     | 6437/13317 [1:06:20<1:10:19,  1.63it/s]
                                                        
{'loss': 0.1866, 'grad_norm': 0.5071525573730469, 'learning_rate': 6.152725098741764e-06, 'epoch': 1.45}

 48%|████▊     | 6437/13317 [1:06:20<1:10:19,  1.63it/s]
 48%|████▊     | 6438/13317 [1:06:20<1:10:19,  1.63it/s]
                                                        
{'loss': 0.2114, 'grad_norm': 0.5300816893577576, 'learning_rate': 6.151449730258843e-06, 'epoch': 1.45}

 48%|████▊     | 6438/13317 [1:06:20<1:10:19,  1.63it/s]
 48%|████▊     | 6439/13317 [1:06:21<1:10:17,  1.63it/s]
                                                        
{'loss': 0.2044, 'grad_norm': 0.47706398367881775, 'learning_rate': 6.150174282659133e-06, 'epoch': 1.45}

 48%|████▊     | 6439/13317 [1:06:21<1:10:17,  1.63it/s]
 48%|████▊     | 6440/13317 [1:06:21<1:10:16,  1.63it/s]
                                                        
{'loss': 0.219, 'grad_norm': 0.6628315448760986, 'learning_rate': 6.148898756030273e-06, 'epoch': 1.45}

 48%|████▊     | 6440/13317 [1:06:22<1:10:16,  1.63it/s]
 48%|████▊     | 6441/13317 [1:06:22<1:10:16,  1.63it/s]
                                                        
{'loss': 0.2085, 'grad_norm': 0.5373318791389465, 'learning_rate': 6.147623150459901e-06, 'epoch': 1.45}

 48%|████▊     | 6441/13317 [1:06:22<1:10:16,  1.63it/s]
 48%|████▊     | 6442/13317 [1:06:23<1:10:18,  1.63it/s]
                                                        
{'loss': 0.1782, 'grad_norm': 0.5020185112953186, 'learning_rate': 6.146347466035669e-06, 'epoch': 1.45}

 48%|████▊     | 6442/13317 [1:06:23<1:10:18,  1.63it/s]
 48%|████▊     | 6443/13317 [1:06:23<1:10:16,  1.63it/s]
                                                        
{'loss': 0.1689, 'grad_norm': 0.4801888167858124, 'learning_rate': 6.14507170284523e-06, 'epoch': 1.45}

 48%|████▊     | 6443/13317 [1:06:23<1:10:16,  1.63it/s]
 48%|████▊     | 6444/13317 [1:06:24<1:10:14,  1.63it/s]
                                                        
{'loss': 0.1382, 'grad_norm': 0.41784223914146423, 'learning_rate': 6.143795860976239e-06, 'epoch': 1.45}

 48%|████▊     | 6444/13317 [1:06:24<1:10:14,  1.63it/s]
 48%|████▊     | 6445/13317 [1:06:24<1:10:15,  1.63it/s]
                                                        
{'loss': 0.1697, 'grad_norm': 0.4460342228412628, 'learning_rate': 6.142519940516361e-06, 'epoch': 1.45}

 48%|████▊     | 6445/13317 [1:06:25<1:10:15,  1.63it/s]
 48%|████▊     | 6446/13317 [1:06:25<1:10:15,  1.63it/s]
                                                        
{'loss': 0.2218, 'grad_norm': 0.5663385987281799, 'learning_rate': 6.1412439415532665e-06, 'epoch': 1.45}

 48%|████▊     | 6446/13317 [1:06:25<1:10:15,  1.63it/s]
 48%|████▊     | 6447/13317 [1:06:26<1:10:15,  1.63it/s]
                                                        
{'loss': 0.2441, 'grad_norm': 0.534221887588501, 'learning_rate': 6.139967864174631e-06, 'epoch': 1.45}

 48%|████▊     | 6447/13317 [1:06:26<1:10:15,  1.63it/s]
 48%|████▊     | 6448/13317 [1:06:26<1:10:15,  1.63it/s]
                                                        
{'loss': 0.3199, 'grad_norm': 0.6764276027679443, 'learning_rate': 6.138691708468131e-06, 'epoch': 1.45}

 48%|████▊     | 6448/13317 [1:06:26<1:10:15,  1.63it/s]
 48%|████▊     | 6449/13317 [1:06:27<1:10:13,  1.63it/s]
                                                        
{'loss': 0.1503, 'grad_norm': 0.4347374439239502, 'learning_rate': 6.1374154745214546e-06, 'epoch': 1.45}

 48%|████▊     | 6449/13317 [1:06:27<1:10:13,  1.63it/s]
 48%|████▊     | 6450/13317 [1:06:28<1:10:13,  1.63it/s]
                                                        
{'loss': 0.2247, 'grad_norm': 0.5468183755874634, 'learning_rate': 6.1361391624222925e-06, 'epoch': 1.45}

 48%|████▊     | 6450/13317 [1:06:28<1:10:13,  1.63it/s]
 48%|████▊     | 6451/13317 [1:06:28<1:10:12,  1.63it/s]
                                                        
{'loss': 0.1953, 'grad_norm': 0.5354989767074585, 'learning_rate': 6.134862772258339e-06, 'epoch': 1.45}

 48%|████▊     | 6451/13317 [1:06:28<1:10:12,  1.63it/s]
 48%|████▊     | 6452/13317 [1:06:29<1:10:12,  1.63it/s]
                                                        
{'loss': 0.2519, 'grad_norm': 0.5287705063819885, 'learning_rate': 6.133586304117299e-06, 'epoch': 1.45}

 48%|████▊     | 6452/13317 [1:06:29<1:10:12,  1.63it/s]
 48%|████▊     | 6453/13317 [1:06:29<1:10:09,  1.63it/s]
                                                        
{'loss': 0.184, 'grad_norm': 0.5610770583152771, 'learning_rate': 6.132309758086875e-06, 'epoch': 1.45}

 48%|████▊     | 6453/13317 [1:06:29<1:10:09,  1.63it/s]
 48%|████▊     | 6454/13317 [1:06:30<1:10:09,  1.63it/s]
                                                        
{'loss': 0.2144, 'grad_norm': 0.5780531764030457, 'learning_rate': 6.131033134254784e-06, 'epoch': 1.45}

 48%|████▊     | 6454/13317 [1:06:30<1:10:09,  1.63it/s]
 48%|████▊     | 6455/13317 [1:06:31<1:10:08,  1.63it/s]
                                                        
{'loss': 0.2145, 'grad_norm': 0.4849793016910553, 'learning_rate': 6.129756432708739e-06, 'epoch': 1.45}

 48%|████▊     | 6455/13317 [1:06:31<1:10:08,  1.63it/s]
 48%|████▊     | 6456/13317 [1:06:31<1:10:06,  1.63it/s]
                                                        
{'loss': 0.2322, 'grad_norm': 0.5932474732398987, 'learning_rate': 6.128479653536465e-06, 'epoch': 1.45}

 48%|████▊     | 6456/13317 [1:06:31<1:10:06,  1.63it/s]
 48%|████▊     | 6457/13317 [1:06:32<1:10:19,  1.63it/s]
                                                        
{'loss': 0.1935, 'grad_norm': 0.4723108410835266, 'learning_rate': 6.127202796825692e-06, 'epoch': 1.45}

 48%|████▊     | 6457/13317 [1:06:32<1:10:19,  1.63it/s]
 48%|████▊     | 6458/13317 [1:06:32<1:10:15,  1.63it/s]
                                                        
{'loss': 0.2266, 'grad_norm': 0.7468894720077515, 'learning_rate': 6.12592586266415e-06, 'epoch': 1.45}

 48%|████▊     | 6458/13317 [1:06:33<1:10:15,  1.63it/s]
 49%|████▊     | 6459/13317 [1:06:33<1:10:11,  1.63it/s]
                                                        
{'loss': 0.2008, 'grad_norm': 0.49488914012908936, 'learning_rate': 6.124648851139581e-06, 'epoch': 1.46}

 49%|████▊     | 6459/13317 [1:06:33<1:10:11,  1.63it/s]
 49%|████▊     | 6460/13317 [1:06:34<1:10:12,  1.63it/s]
                                                        
{'loss': 0.1567, 'grad_norm': 0.5078750252723694, 'learning_rate': 6.123371762339728e-06, 'epoch': 1.46}

 49%|████▊     | 6460/13317 [1:06:34<1:10:12,  1.63it/s]
 49%|████▊     | 6461/13317 [1:06:34<1:10:11,  1.63it/s]
                                                        
{'loss': 0.1616, 'grad_norm': 0.47639456391334534, 'learning_rate': 6.12209459635234e-06, 'epoch': 1.46}

 49%|████▊     | 6461/13317 [1:06:34<1:10:11,  1.63it/s]
 49%|████▊     | 6462/13317 [1:06:35<1:10:11,  1.63it/s]
                                                        
{'loss': 0.2214, 'grad_norm': 0.48809105157852173, 'learning_rate': 6.120817353265173e-06, 'epoch': 1.46}

 49%|████▊     | 6462/13317 [1:06:35<1:10:11,  1.63it/s]
 49%|████▊     | 6463/13317 [1:06:36<1:10:12,  1.63it/s]
                                                        
{'loss': 0.2186, 'grad_norm': 0.5035330057144165, 'learning_rate': 6.119540033165985e-06, 'epoch': 1.46}

 49%|████▊     | 6463/13317 [1:06:36<1:10:12,  1.63it/s]
 49%|████▊     | 6464/13317 [1:06:36<1:10:08,  1.63it/s]
                                                        
{'loss': 0.2002, 'grad_norm': 0.5489224791526794, 'learning_rate': 6.118262636142544e-06, 'epoch': 1.46}

 49%|████▊     | 6464/13317 [1:06:36<1:10:08,  1.63it/s]
 49%|████▊     | 6465/13317 [1:06:37<1:10:09,  1.63it/s]
                                                        
{'loss': 0.196, 'grad_norm': 0.5383415222167969, 'learning_rate': 6.11698516228262e-06, 'epoch': 1.46}

 49%|████▊     | 6465/13317 [1:06:37<1:10:09,  1.63it/s]
 49%|████▊     | 6466/13317 [1:06:37<1:10:05,  1.63it/s]
                                                        
{'loss': 0.2223, 'grad_norm': 0.5392643809318542, 'learning_rate': 6.115707611673987e-06, 'epoch': 1.46}

 49%|████▊     | 6466/13317 [1:06:37<1:10:05,  1.63it/s]
 49%|████▊     | 6467/13317 [1:06:38<1:10:04,  1.63it/s]
                                                        
{'loss': 0.2924, 'grad_norm': 0.586269736289978, 'learning_rate': 6.114429984404428e-06, 'epoch': 1.46}

 49%|████▊     | 6467/13317 [1:06:38<1:10:04,  1.63it/s]
 49%|████▊     | 6468/13317 [1:06:39<1:10:03,  1.63it/s]
                                                        
{'loss': 0.2437, 'grad_norm': 0.582368791103363, 'learning_rate': 6.11315228056173e-06, 'epoch': 1.46}

 49%|████▊     | 6468/13317 [1:06:39<1:10:03,  1.63it/s]
 49%|████▊     | 6469/13317 [1:06:39<1:10:01,  1.63it/s]
                                                        
{'loss': 0.1723, 'grad_norm': 0.6539122462272644, 'learning_rate': 6.111874500233685e-06, 'epoch': 1.46}

 49%|████▊     | 6469/13317 [1:06:39<1:10:01,  1.63it/s]
 49%|████▊     | 6470/13317 [1:06:40<1:10:06,  1.63it/s]
                                                        
{'loss': 0.2458, 'grad_norm': 0.5524458885192871, 'learning_rate': 6.110596643508088e-06, 'epoch': 1.46}

 49%|████▊     | 6470/13317 [1:06:40<1:10:06,  1.63it/s]
 49%|████▊     | 6471/13317 [1:06:40<1:10:01,  1.63it/s]
                                                        
{'loss': 0.2373, 'grad_norm': 0.6474801301956177, 'learning_rate': 6.109318710472741e-06, 'epoch': 1.46}

 49%|████▊     | 6471/13317 [1:06:41<1:10:01,  1.63it/s]
 49%|████▊     | 6472/13317 [1:06:41<1:10:01,  1.63it/s]
                                                        
{'loss': 0.2433, 'grad_norm': 0.6220043897628784, 'learning_rate': 6.108040701215456e-06, 'epoch': 1.46}

 49%|████▊     | 6472/13317 [1:06:41<1:10:01,  1.63it/s]
 49%|████▊     | 6473/13317 [1:06:42<1:10:01,  1.63it/s]
                                                        
{'loss': 0.1807, 'grad_norm': 0.4709665775299072, 'learning_rate': 6.10676261582404e-06, 'epoch': 1.46}

 49%|████▊     | 6473/13317 [1:06:42<1:10:01,  1.63it/s]
 49%|████▊     | 6474/13317 [1:06:42<1:09:59,  1.63it/s]
                                                        
{'loss': 0.1653, 'grad_norm': 0.4390797019004822, 'learning_rate': 6.105484454386316e-06, 'epoch': 1.46}

 49%|████▊     | 6474/13317 [1:06:42<1:09:59,  1.63it/s]
 49%|████▊     | 6475/13317 [1:06:43<1:09:58,  1.63it/s]
                                                        
{'loss': 0.1518, 'grad_norm': 0.4051465690135956, 'learning_rate': 6.1042062169901015e-06, 'epoch': 1.46}

 49%|████▊     | 6475/13317 [1:06:43<1:09:58,  1.63it/s]
 49%|████▊     | 6476/13317 [1:06:43<1:09:58,  1.63it/s]
                                                        
{'loss': 0.1717, 'grad_norm': 0.5263600945472717, 'learning_rate': 6.102927903723231e-06, 'epoch': 1.46}

 49%|████▊     | 6476/13317 [1:06:44<1:09:58,  1.63it/s]
 49%|████▊     | 6477/13317 [1:06:44<1:09:59,  1.63it/s]
                                                        
{'loss': 0.2441, 'grad_norm': 0.5950492024421692, 'learning_rate': 6.101649514673535e-06, 'epoch': 1.46}

 49%|████▊     | 6477/13317 [1:06:44<1:09:59,  1.63it/s]
 49%|████▊     | 6478/13317 [1:06:45<1:09:58,  1.63it/s]
                                                        
{'loss': 0.1859, 'grad_norm': 0.4690324068069458, 'learning_rate': 6.100371049928853e-06, 'epoch': 1.46}

 49%|████▊     | 6478/13317 [1:06:45<1:09:58,  1.63it/s]
 49%|████▊     | 6479/13317 [1:06:45<1:09:56,  1.63it/s]
                                                        
{'loss': 0.1765, 'grad_norm': 0.49651840329170227, 'learning_rate': 6.099092509577031e-06, 'epoch': 1.46}

 49%|████▊     | 6479/13317 [1:06:45<1:09:56,  1.63it/s]
 49%|████▊     | 6480/13317 [1:06:46<1:09:54,  1.63it/s]
                                                        
{'loss': 0.1908, 'grad_norm': 0.4933200478553772, 'learning_rate': 6.097813893705912e-06, 'epoch': 1.46}

 49%|████▊     | 6480/13317 [1:06:46<1:09:54,  1.63it/s]
 49%|████▊     | 6481/13317 [1:06:47<1:09:54,  1.63it/s]
                                                        
{'loss': 0.218, 'grad_norm': 0.5168551802635193, 'learning_rate': 6.0965352024033575e-06, 'epoch': 1.46}

 49%|████▊     | 6481/13317 [1:06:47<1:09:54,  1.63it/s]
 49%|████▊     | 6482/13317 [1:06:47<1:09:53,  1.63it/s]
                                                        
{'loss': 0.1783, 'grad_norm': 0.5022075772285461, 'learning_rate': 6.095256435757224e-06, 'epoch': 1.46}

 49%|████▊     | 6482/13317 [1:06:47<1:09:53,  1.63it/s]
 49%|████▊     | 6483/13317 [1:06:48<1:09:54,  1.63it/s]
                                                        
{'loss': 0.1781, 'grad_norm': 0.4582172632217407, 'learning_rate': 6.093977593855376e-06, 'epoch': 1.46}

 49%|████▊     | 6483/13317 [1:06:48<1:09:54,  1.63it/s]
 49%|████▊     | 6484/13317 [1:06:48<1:09:51,  1.63it/s]
                                                        
{'loss': 0.1844, 'grad_norm': 0.4307849109172821, 'learning_rate': 6.092698676785684e-06, 'epoch': 1.46}

 49%|████▊     | 6484/13317 [1:06:49<1:09:51,  1.63it/s]
 49%|████▊     | 6485/13317 [1:06:49<1:09:49,  1.63it/s]
                                                        
{'loss': 0.2206, 'grad_norm': 0.557482123374939, 'learning_rate': 6.091419684636023e-06, 'epoch': 1.46}

 49%|████▊     | 6485/13317 [1:06:49<1:09:49,  1.63it/s]
 49%|████▊     | 6486/13317 [1:06:50<1:09:48,  1.63it/s]
                                                        
{'loss': 0.2469, 'grad_norm': 0.6324387192726135, 'learning_rate': 6.090140617494275e-06, 'epoch': 1.46}

 49%|████▊     | 6486/13317 [1:06:50<1:09:48,  1.63it/s]
 49%|████▊     | 6487/13317 [1:06:50<1:09:47,  1.63it/s]
                                                        
{'loss': 0.2214, 'grad_norm': 0.505693793296814, 'learning_rate': 6.088861475448323e-06, 'epoch': 1.46}

 49%|████▊     | 6487/13317 [1:06:50<1:09:47,  1.63it/s]
 49%|████▊     | 6488/13317 [1:06:51<1:09:46,  1.63it/s]
                                                        
{'loss': 0.1987, 'grad_norm': 0.5303196907043457, 'learning_rate': 6.0875822585860585e-06, 'epoch': 1.46}

 49%|████▊     | 6488/13317 [1:06:51<1:09:46,  1.63it/s]
 49%|████▊     | 6489/13317 [1:06:51<1:09:47,  1.63it/s]
                                                        
{'loss': 0.4295, 'grad_norm': 0.585081934928894, 'learning_rate': 6.086302966995378e-06, 'epoch': 1.46}

 49%|████▊     | 6489/13317 [1:06:52<1:09:47,  1.63it/s]
 49%|████▊     | 6490/13317 [1:06:52<1:09:48,  1.63it/s]
                                                        
{'loss': 0.2274, 'grad_norm': 0.6939131021499634, 'learning_rate': 6.085023600764181e-06, 'epoch': 1.46}

 49%|████▊     | 6490/13317 [1:06:52<1:09:48,  1.63it/s]
 49%|████▊     | 6491/13317 [1:06:53<1:09:48,  1.63it/s]
                                                        
{'loss': 0.2694, 'grad_norm': 0.7033320665359497, 'learning_rate': 6.083744159980373e-06, 'epoch': 1.46}

 49%|████▊     | 6491/13317 [1:06:53<1:09:48,  1.63it/s]
 49%|████▊     | 6492/13317 [1:06:53<1:09:46,  1.63it/s]
                                                        
{'loss': 0.1685, 'grad_norm': 0.5492423176765442, 'learning_rate': 6.082464644731868e-06, 'epoch': 1.46}

 49%|████▊     | 6492/13317 [1:06:53<1:09:46,  1.63it/s]
 49%|████▉     | 6493/13317 [1:06:54<1:09:44,  1.63it/s]
                                                        
{'loss': 0.2047, 'grad_norm': 0.5747434496879578, 'learning_rate': 6.08118505510658e-06, 'epoch': 1.46}

 49%|████▉     | 6493/13317 [1:06:54<1:09:44,  1.63it/s]
 49%|████▉     | 6494/13317 [1:06:55<1:09:45,  1.63it/s]
                                                        
{'loss': 0.2169, 'grad_norm': 0.6553597450256348, 'learning_rate': 6.079905391192432e-06, 'epoch': 1.46}

 49%|████▉     | 6494/13317 [1:06:55<1:09:45,  1.63it/s]
 49%|████▉     | 6495/13317 [1:06:55<1:09:43,  1.63it/s]
                                                        
{'loss': 0.2094, 'grad_norm': 0.583979606628418, 'learning_rate': 6.078625653077349e-06, 'epoch': 1.46}

 49%|████▉     | 6495/13317 [1:06:55<1:09:43,  1.63it/s]
 49%|████▉     | 6496/13317 [1:06:56<1:09:43,  1.63it/s]
                                                        
{'loss': 0.2783, 'grad_norm': 0.5220510959625244, 'learning_rate': 6.077345840849263e-06, 'epoch': 1.46}

 49%|████▉     | 6496/13317 [1:06:56<1:09:43,  1.63it/s]
 49%|████▉     | 6497/13317 [1:06:56<1:09:43,  1.63it/s]
                                                        
{'loss': 0.2075, 'grad_norm': 0.5272742509841919, 'learning_rate': 6.076065954596109e-06, 'epoch': 1.46}

 49%|████▉     | 6497/13317 [1:06:57<1:09:43,  1.63it/s]
 49%|████▉     | 6498/13317 [1:06:57<1:09:43,  1.63it/s]
                                                        
{'loss': 0.1879, 'grad_norm': 0.48566046357154846, 'learning_rate': 6.074785994405832e-06, 'epoch': 1.46}

 49%|████▉     | 6498/13317 [1:06:57<1:09:43,  1.63it/s]
 49%|████▉     | 6499/13317 [1:06:58<1:09:42,  1.63it/s]
                                                        
{'loss': 0.2185, 'grad_norm': 0.5905725359916687, 'learning_rate': 6.073505960366377e-06, 'epoch': 1.46}

 49%|████▉     | 6499/13317 [1:06:58<1:09:42,  1.63it/s]
 49%|████▉     | 6500/13317 [1:06:58<1:09:40,  1.63it/s]
                                                        
{'loss': 0.2752, 'grad_norm': 0.5844641327857971, 'learning_rate': 6.0722258525656954e-06, 'epoch': 1.46}

 49%|████▉     | 6500/13317 [1:06:58<1:09:40,  1.63it/s]

  0%|          | 0/13 [00:00<?, ?it/s]

 15%|█▌        | 2/13 [00:00<00:00, 13.37it/s]

 31%|███       | 4/13 [00:00<00:01,  8.40it/s]

 38%|███▊      | 5/13 [00:00<00:01,  7.80it/s]

 46%|████▌     | 6/13 [00:00<00:00,  7.43it/s]

 54%|█████▍    | 7/13 [00:00<00:00,  7.19it/s]

 62%|██████▏   | 8/13 [00:01<00:00,  7.02it/s]

 69%|██████▉   | 9/13 [00:01<00:00,  6.92it/s]

 77%|███████▋  | 10/13 [00:01<00:00,  6.84it/s]

 85%|████████▍ | 11/13 [00:01<00:00,  6.78it/s]

 92%|█████████▏| 12/13 [00:01<00:00,  6.74it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]
                                                        

                                               
{'eval_loss': 0.576500654220581, 'eval_runtime': 1.9518, 'eval_samples_per_second': 102.469, 'eval_steps_per_second': 6.66, 'epoch': 1.46}

 49%|████▉     | 6500/13317 [1:07:00<1:09:40,  1.63it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]

                                               
 49%|████▉     | 6501/13317 [1:07:01<2:16:25,  1.20s/it]
                                                        
{'loss': 0.2304, 'grad_norm': 0.6999367475509644, 'learning_rate': 6.0709456710917455e-06, 'epoch': 1.46}

 49%|████▉     | 6501/13317 [1:07:01<2:16:25,  1.20s/it]
 49%|████▉     | 6502/13317 [1:07:01<1:56:25,  1.02s/it]
                                                        
{'loss': 0.2039, 'grad_norm': 0.49373459815979004, 'learning_rate': 6.0696654160324875e-06, 'epoch': 1.46}

 49%|████▉     | 6502/13317 [1:07:02<1:56:25,  1.02s/it]
 49%|████▉     | 6503/13317 [1:07:02<1:42:22,  1.11it/s]
                                                        
{'loss': 0.1697, 'grad_norm': 0.5644209980964661, 'learning_rate': 6.068385087475891e-06, 'epoch': 1.46}

 49%|████▉     | 6503/13317 [1:07:02<1:42:22,  1.11it/s]
 49%|████▉     | 6504/13317 [1:07:03<1:32:33,  1.23it/s]
                                                        
{'loss': 0.1901, 'grad_norm': 0.5384250283241272, 'learning_rate': 6.0671046855099255e-06, 'epoch': 1.47}

 49%|████▉     | 6504/13317 [1:07:03<1:32:33,  1.23it/s]
 49%|████▉     | 6505/13317 [1:07:03<1:25:39,  1.33it/s]
                                                        
{'loss': 0.186, 'grad_norm': 0.4845925569534302, 'learning_rate': 6.0658242102225706e-06, 'epoch': 1.47}

 49%|████▉     | 6505/13317 [1:07:03<1:25:39,  1.33it/s]
 49%|████▉     | 6506/13317 [1:07:04<1:20:48,  1.40it/s]
                                                        
{'loss': 0.1809, 'grad_norm': 0.5193185210227966, 'learning_rate': 6.0645436617018075e-06, 'epoch': 1.47}

 49%|████▉     | 6506/13317 [1:07:04<1:20:48,  1.40it/s]
 49%|████▉     | 6507/13317 [1:07:04<1:17:26,  1.47it/s]
                                                        
{'loss': 0.1698, 'grad_norm': 0.4630511701107025, 'learning_rate': 6.063263040035622e-06, 'epoch': 1.47}

 49%|████▉     | 6507/13317 [1:07:05<1:17:26,  1.47it/s]
 49%|████▉     | 6508/13317 [1:07:05<1:15:05,  1.51it/s]
                                                        
{'loss': 0.1837, 'grad_norm': 0.5179226398468018, 'learning_rate': 6.061982345312008e-06, 'epoch': 1.47}

 49%|████▉     | 6508/13317 [1:07:05<1:15:05,  1.51it/s]
 49%|████▉     | 6509/13317 [1:07:06<1:13:27,  1.54it/s]
                                                        
{'loss': 0.2091, 'grad_norm': 0.5419445633888245, 'learning_rate': 6.060701577618964e-06, 'epoch': 1.47}

 49%|████▉     | 6509/13317 [1:07:06<1:13:27,  1.54it/s]
 49%|████▉     | 6510/13317 [1:07:06<1:12:18,  1.57it/s]
                                                        
{'loss': 0.1945, 'grad_norm': 0.4782806932926178, 'learning_rate': 6.059420737044489e-06, 'epoch': 1.47}

 49%|████▉     | 6510/13317 [1:07:06<1:12:18,  1.57it/s]
 49%|████▉     | 6511/13317 [1:07:07<1:11:27,  1.59it/s]
                                                        
{'loss': 0.2284, 'grad_norm': 0.597383439540863, 'learning_rate': 6.058139823676592e-06, 'epoch': 1.47}

 49%|████▉     | 6511/13317 [1:07:07<1:11:27,  1.59it/s]
 49%|████▉     | 6512/13317 [1:07:08<1:10:53,  1.60it/s]
                                                        
{'loss': 0.1597, 'grad_norm': 0.5037989616394043, 'learning_rate': 6.056858837603286e-06, 'epoch': 1.47}

 49%|████▉     | 6512/13317 [1:07:08<1:10:53,  1.60it/s]
 49%|████▉     | 6513/13317 [1:07:08<1:10:28,  1.61it/s]
                                                        
{'loss': 0.1989, 'grad_norm': 0.5175331234931946, 'learning_rate': 6.055577778912587e-06, 'epoch': 1.47}

 49%|████▉     | 6513/13317 [1:07:08<1:10:28,  1.61it/s]
 49%|████▉     | 6514/13317 [1:07:09<1:10:10,  1.62it/s]
                                                        
{'loss': 0.1898, 'grad_norm': 0.49723830819129944, 'learning_rate': 6.054296647692518e-06, 'epoch': 1.47}

 49%|████▉     | 6514/13317 [1:07:09<1:10:10,  1.62it/s]
 49%|████▉     | 6515/13317 [1:07:09<1:09:57,  1.62it/s]
                                                        
{'loss': 0.1655, 'grad_norm': 0.4581829011440277, 'learning_rate': 6.053015444031107e-06, 'epoch': 1.47}

 49%|████▉     | 6515/13317 [1:07:10<1:09:57,  1.62it/s]
 49%|████▉     | 6516/13317 [1:07:10<1:09:53,  1.62it/s]
                                                        
{'loss': 0.1785, 'grad_norm': 0.5252153873443604, 'learning_rate': 6.051734168016385e-06, 'epoch': 1.47}

 49%|████▉     | 6516/13317 [1:07:10<1:09:53,  1.62it/s]
 49%|████▉     | 6517/13317 [1:07:11<1:11:39,  1.58it/s]
                                                        
{'loss': 0.1834, 'grad_norm': 0.4656916856765747, 'learning_rate': 6.05045281973639e-06, 'epoch': 1.47}

 49%|████▉     | 6517/13317 [1:07:11<1:11:39,  1.58it/s]
 49%|████▉     | 6518/13317 [1:07:11<1:10:58,  1.60it/s]
                                                        
{'loss': 0.1981, 'grad_norm': 0.5587313175201416, 'learning_rate': 6.049171399279165e-06, 'epoch': 1.47}

 49%|████▉     | 6518/13317 [1:07:11<1:10:58,  1.60it/s]
 49%|████▉     | 6519/13317 [1:07:12<1:10:33,  1.61it/s]
                                                        
{'loss': 0.1998, 'grad_norm': 0.5239765048027039, 'learning_rate': 6.0478899067327535e-06, 'epoch': 1.47}

 49%|████▉     | 6519/13317 [1:07:12<1:10:33,  1.61it/s]
 49%|████▉     | 6520/13317 [1:07:13<1:10:31,  1.61it/s]
                                                        
{'loss': 0.2503, 'grad_norm': 0.5687004327774048, 'learning_rate': 6.046608342185212e-06, 'epoch': 1.47}

 49%|████▉     | 6520/13317 [1:07:13<1:10:31,  1.61it/s]
 49%|████▉     | 6521/13317 [1:07:13<1:10:21,  1.61it/s]
                                                        
{'loss': 0.2234, 'grad_norm': 0.5924002528190613, 'learning_rate': 6.045326705724595e-06, 'epoch': 1.47}

 49%|████▉     | 6521/13317 [1:07:13<1:10:21,  1.61it/s]
 49%|████▉     | 6522/13317 [1:07:14<1:10:05,  1.62it/s]
                                                        
{'loss': 0.2427, 'grad_norm': 0.5889170169830322, 'learning_rate': 6.044044997438964e-06, 'epoch': 1.47}

 49%|████▉     | 6522/13317 [1:07:14<1:10:05,  1.62it/s]
 49%|████▉     | 6523/13317 [1:07:14<1:09:55,  1.62it/s]
                                                        
{'loss': 0.1712, 'grad_norm': 0.5486186742782593, 'learning_rate': 6.042763217416391e-06, 'epoch': 1.47}

 49%|████▉     | 6523/13317 [1:07:14<1:09:55,  1.62it/s]
 49%|████▉     | 6524/13317 [1:07:15<1:09:46,  1.62it/s]
                                                        
{'loss': 0.2064, 'grad_norm': 0.568105936050415, 'learning_rate': 6.0414813657449415e-06, 'epoch': 1.47}

 49%|████▉     | 6524/13317 [1:07:15<1:09:46,  1.62it/s]
 49%|████▉     | 6525/13317 [1:07:16<1:09:40,  1.62it/s]
                                                        
{'loss': 0.1779, 'grad_norm': 0.5070335268974304, 'learning_rate': 6.040199442512694e-06, 'epoch': 1.47}

 49%|████▉     | 6525/13317 [1:07:16<1:09:40,  1.62it/s]
 49%|████▉     | 6526/13317 [1:07:16<1:09:40,  1.62it/s]
                                                        
{'loss': 0.1507, 'grad_norm': 0.45432454347610474, 'learning_rate': 6.0389174478077325e-06, 'epoch': 1.47}

 49%|████▉     | 6526/13317 [1:07:16<1:09:40,  1.62it/s]
 49%|████▉     | 6527/13317 [1:07:17<1:09:37,  1.63it/s]
                                                        
{'loss': 0.2285, 'grad_norm': 0.5499832630157471, 'learning_rate': 6.037635381718141e-06, 'epoch': 1.47}

 49%|████▉     | 6527/13317 [1:07:17<1:09:37,  1.63it/s]
 49%|████▉     | 6528/13317 [1:07:17<1:09:35,  1.63it/s]
                                                        
{'loss': 0.2314, 'grad_norm': 0.5644673109054565, 'learning_rate': 6.036353244332014e-06, 'epoch': 1.47}

 49%|████▉     | 6528/13317 [1:07:18<1:09:35,  1.63it/s]
 49%|████▉     | 6529/13317 [1:07:18<1:09:28,  1.63it/s]
                                                        
{'loss': 0.1709, 'grad_norm': 0.48438897728919983, 'learning_rate': 6.035071035737443e-06, 'epoch': 1.47}

 49%|████▉     | 6529/13317 [1:07:18<1:09:28,  1.63it/s]
 49%|████▉     | 6530/13317 [1:07:19<1:09:28,  1.63it/s]
                                                        
{'loss': 0.1694, 'grad_norm': 0.44049957394599915, 'learning_rate': 6.033788756022535e-06, 'epoch': 1.47}

 49%|████▉     | 6530/13317 [1:07:19<1:09:28,  1.63it/s]
 49%|████▉     | 6531/13317 [1:07:19<1:09:34,  1.63it/s]
                                                        
{'loss': 0.1345, 'grad_norm': 0.4214302897453308, 'learning_rate': 6.032506405275392e-06, 'epoch': 1.47}

 49%|████▉     | 6531/13317 [1:07:19<1:09:34,  1.63it/s]
 49%|████▉     | 6532/13317 [1:07:20<1:09:29,  1.63it/s]
                                                        
{'loss': 0.2178, 'grad_norm': 0.5507137179374695, 'learning_rate': 6.031223983584127e-06, 'epoch': 1.47}

 49%|████▉     | 6532/13317 [1:07:20<1:09:29,  1.63it/s]
 49%|████▉     | 6533/13317 [1:07:20<1:09:27,  1.63it/s]
                                                        
{'loss': 0.2061, 'grad_norm': 0.5783699750900269, 'learning_rate': 6.029941491036856e-06, 'epoch': 1.47}

 49%|████▉     | 6533/13317 [1:07:21<1:09:27,  1.63it/s]
 49%|████▉     | 6534/13317 [1:07:21<1:09:24,  1.63it/s]
                                                        
{'loss': 0.1962, 'grad_norm': 0.5079081654548645, 'learning_rate': 6.028658927721698e-06, 'epoch': 1.47}

 49%|████▉     | 6534/13317 [1:07:21<1:09:24,  1.63it/s]
 49%|████▉     | 6535/13317 [1:07:22<1:09:22,  1.63it/s]
                                                        
{'loss': 0.1893, 'grad_norm': 0.46989744901657104, 'learning_rate': 6.027376293726782e-06, 'epoch': 1.47}

 49%|████▉     | 6535/13317 [1:07:22<1:09:22,  1.63it/s]
 49%|████▉     | 6536/13317 [1:07:22<1:09:24,  1.63it/s]
                                                        
{'loss': 0.1507, 'grad_norm': 0.4371132552623749, 'learning_rate': 6.026093589140235e-06, 'epoch': 1.47}

 49%|████▉     | 6536/13317 [1:07:22<1:09:24,  1.63it/s]
 49%|████▉     | 6537/13317 [1:07:23<1:09:22,  1.63it/s]
                                                        
{'loss': 0.2028, 'grad_norm': 0.5740913152694702, 'learning_rate': 6.024810814050195e-06, 'epoch': 1.47}

 49%|████▉     | 6537/13317 [1:07:23<1:09:22,  1.63it/s]
 49%|████▉     | 6538/13317 [1:07:24<1:09:22,  1.63it/s]
                                                        
{'loss': 0.2173, 'grad_norm': 0.6059438586235046, 'learning_rate': 6.023527968544801e-06, 'epoch': 1.47}

 49%|████▉     | 6538/13317 [1:07:24<1:09:22,  1.63it/s]
 49%|████▉     | 6539/13317 [1:07:24<1:09:19,  1.63it/s]
                                                        
{'loss': 0.2033, 'grad_norm': 0.48561587929725647, 'learning_rate': 6.022245052712198e-06, 'epoch': 1.47}

 49%|████▉     | 6539/13317 [1:07:24<1:09:19,  1.63it/s]
 49%|████▉     | 6540/13317 [1:07:25<1:09:18,  1.63it/s]
                                                        
{'loss': 0.2125, 'grad_norm': 0.5429365038871765, 'learning_rate': 6.020962066640536e-06, 'epoch': 1.47}

 49%|████▉     | 6540/13317 [1:07:25<1:09:18,  1.63it/s]
 49%|████▉     | 6541/13317 [1:07:25<1:09:25,  1.63it/s]
                                                        
{'loss': 0.1554, 'grad_norm': 0.4409153461456299, 'learning_rate': 6.01967901041797e-06, 'epoch': 1.47}

 49%|████▉     | 6541/13317 [1:07:26<1:09:25,  1.63it/s]
 49%|████▉     | 6542/13317 [1:07:26<1:09:22,  1.63it/s]
                                                        
{'loss': 0.1817, 'grad_norm': 0.5008338689804077, 'learning_rate': 6.01839588413266e-06, 'epoch': 1.47}

 49%|████▉     | 6542/13317 [1:07:26<1:09:22,  1.63it/s]
 49%|████▉     | 6543/13317 [1:07:27<1:09:20,  1.63it/s]
                                                        
{'loss': 0.1964, 'grad_norm': 0.4757034480571747, 'learning_rate': 6.01711268787277e-06, 'epoch': 1.47}

 49%|████▉     | 6543/13317 [1:07:27<1:09:20,  1.63it/s]
 49%|████▉     | 6544/13317 [1:07:27<1:09:19,  1.63it/s]
                                                        
{'loss': 0.2126, 'grad_norm': 0.543981671333313, 'learning_rate': 6.015829421726469e-06, 'epoch': 1.47}

 49%|████▉     | 6544/13317 [1:07:27<1:09:19,  1.63it/s]
 49%|████▉     | 6545/13317 [1:07:28<1:09:19,  1.63it/s]
                                                        
{'loss': 0.188, 'grad_norm': 0.49184176325798035, 'learning_rate': 6.014546085781929e-06, 'epoch': 1.47}

 49%|████▉     | 6545/13317 [1:07:28<1:09:19,  1.63it/s]
 49%|████▉     | 6546/13317 [1:07:28<1:09:26,  1.63it/s]
                                                        
{'loss': 0.1878, 'grad_norm': 0.5208247900009155, 'learning_rate': 6.0132626801273334e-06, 'epoch': 1.47}

 49%|████▉     | 6546/13317 [1:07:29<1:09:26,  1.63it/s]
 49%|████▉     | 6547/13317 [1:07:29<1:09:21,  1.63it/s]
                                                        
{'loss': 0.2162, 'grad_norm': 0.5139176845550537, 'learning_rate': 6.01197920485086e-06, 'epoch': 1.47}

 49%|████▉     | 6547/13317 [1:07:29<1:09:21,  1.63it/s]
 49%|████▉     | 6548/13317 [1:07:30<1:09:17,  1.63it/s]
                                                        
{'loss': 0.1816, 'grad_norm': 0.4805299639701843, 'learning_rate': 6.010695660040704e-06, 'epoch': 1.48}

 49%|████▉     | 6548/13317 [1:07:30<1:09:17,  1.63it/s]
 49%|████▉     | 6549/13317 [1:07:30<1:09:15,  1.63it/s]
                                                        
{'loss': 0.1764, 'grad_norm': 0.5082164406776428, 'learning_rate': 6.009412045785051e-06, 'epoch': 1.48}

 49%|████▉     | 6549/13317 [1:07:30<1:09:15,  1.63it/s]
 49%|████▉     | 6550/13317 [1:07:31<1:09:13,  1.63it/s]
                                                        
{'loss': 0.1729, 'grad_norm': 0.4967433512210846, 'learning_rate': 6.0081283621721055e-06, 'epoch': 1.48}

 49%|████▉     | 6550/13317 [1:07:31<1:09:13,  1.63it/s]
 49%|████▉     | 6551/13317 [1:07:32<1:09:16,  1.63it/s]
                                                        
{'loss': 0.2414, 'grad_norm': 0.5480334162712097, 'learning_rate': 6.006844609290066e-06, 'epoch': 1.48}

 49%|████▉     | 6551/13317 [1:07:32<1:09:16,  1.63it/s]
 49%|████▉     | 6552/13317 [1:07:32<1:09:15,  1.63it/s]
                                                        
{'loss': 0.2028, 'grad_norm': 0.560914158821106, 'learning_rate': 6.005560787227142e-06, 'epoch': 1.48}

 49%|████▉     | 6552/13317 [1:07:32<1:09:15,  1.63it/s]
 49%|████▉     | 6553/13317 [1:07:33<1:09:13,  1.63it/s]
                                                        
{'loss': 0.1923, 'grad_norm': 0.5036630034446716, 'learning_rate': 6.004276896071545e-06, 'epoch': 1.48}

 49%|████▉     | 6553/13317 [1:07:33<1:09:13,  1.63it/s]
 49%|████▉     | 6554/13317 [1:07:33<1:09:10,  1.63it/s]
                                                        
{'loss': 0.1775, 'grad_norm': 0.49291637539863586, 'learning_rate': 6.002992935911489e-06, 'epoch': 1.48}

 49%|████▉     | 6554/13317 [1:07:34<1:09:10,  1.63it/s]
 49%|████▉     | 6555/13317 [1:07:34<1:09:08,  1.63it/s]
                                                        
{'loss': 0.2179, 'grad_norm': 0.5550397038459778, 'learning_rate': 6.001708906835202e-06, 'epoch': 1.48}

 49%|████▉     | 6555/13317 [1:07:34<1:09:08,  1.63it/s]
 49%|████▉     | 6556/13317 [1:07:35<1:09:18,  1.63it/s]
                                                        
{'loss': 0.216, 'grad_norm': 0.5476694107055664, 'learning_rate': 6.000424808930904e-06, 'epoch': 1.48}

 49%|████▉     | 6556/13317 [1:07:35<1:09:18,  1.63it/s]
 49%|████▉     | 6557/13317 [1:07:35<1:09:14,  1.63it/s]
                                                        
{'loss': 0.1531, 'grad_norm': 0.4285644292831421, 'learning_rate': 5.999140642286831e-06, 'epoch': 1.48}

 49%|████▉     | 6557/13317 [1:07:35<1:09:14,  1.63it/s]
 49%|████▉     | 6558/13317 [1:07:36<1:09:10,  1.63it/s]
                                                        
{'loss': 0.1736, 'grad_norm': 0.5119601488113403, 'learning_rate': 5.997856406991216e-06, 'epoch': 1.48}

 49%|████▉     | 6558/13317 [1:07:36<1:09:10,  1.63it/s]
 49%|████▉     | 6559/13317 [1:07:36<1:09:11,  1.63it/s]
                                                        
{'loss': 0.2183, 'grad_norm': 0.5701276063919067, 'learning_rate': 5.9965721031323e-06, 'epoch': 1.48}

 49%|████▉     | 6559/13317 [1:07:37<1:09:11,  1.63it/s]
 49%|████▉     | 6560/13317 [1:07:37<1:09:12,  1.63it/s]
                                                        
{'loss': 0.2386, 'grad_norm': 0.5203710794448853, 'learning_rate': 5.99528773079833e-06, 'epoch': 1.48}

 49%|████▉     | 6560/13317 [1:07:37<1:09:12,  1.63it/s]
 49%|████▉     | 6561/13317 [1:07:38<1:09:16,  1.63it/s]
                                                        
{'loss': 0.2481, 'grad_norm': 0.6109244227409363, 'learning_rate': 5.994003290077552e-06, 'epoch': 1.48}

 49%|████▉     | 6561/13317 [1:07:38<1:09:16,  1.63it/s]
 49%|████▉     | 6562/13317 [1:07:38<1:09:12,  1.63it/s]
                                                        
{'loss': 0.1961, 'grad_norm': 0.5199070572853088, 'learning_rate': 5.992718781058226e-06, 'epoch': 1.48}

 49%|████▉     | 6562/13317 [1:07:38<1:09:12,  1.63it/s]
 49%|████▉     | 6563/13317 [1:07:39<1:09:09,  1.63it/s]
                                                        
{'loss': 0.1579, 'grad_norm': 0.4416359066963196, 'learning_rate': 5.991434203828607e-06, 'epoch': 1.48}

 49%|████▉     | 6563/13317 [1:07:39<1:09:09,  1.63it/s]
 49%|████▉     | 6564/13317 [1:07:40<1:09:05,  1.63it/s]
                                                        
{'loss': 0.2444, 'grad_norm': 0.7719336152076721, 'learning_rate': 5.990149558476962e-06, 'epoch': 1.48}

 49%|████▉     | 6564/13317 [1:07:40<1:09:05,  1.63it/s]
 49%|████▉     | 6565/13317 [1:07:40<1:09:05,  1.63it/s]
                                                        
{'loss': 0.14, 'grad_norm': 0.39019501209259033, 'learning_rate': 5.988864845091557e-06, 'epoch': 1.48}

 49%|████▉     | 6565/13317 [1:07:40<1:09:05,  1.63it/s]
 49%|████▉     | 6566/13317 [1:07:41<1:09:09,  1.63it/s]
                                                        
{'loss': 0.1596, 'grad_norm': 0.4416877329349518, 'learning_rate': 5.987580063760669e-06, 'epoch': 1.48}

 49%|████▉     | 6566/13317 [1:07:41<1:09:09,  1.63it/s]
 49%|████▉     | 6567/13317 [1:07:41<1:09:06,  1.63it/s]
                                                        
{'loss': 0.245, 'grad_norm': 0.586362898349762, 'learning_rate': 5.986295214572572e-06, 'epoch': 1.48}

 49%|████▉     | 6567/13317 [1:07:42<1:09:06,  1.63it/s]
 49%|████▉     | 6568/13317 [1:07:42<1:09:04,  1.63it/s]
                                                        
{'loss': 0.1874, 'grad_norm': 0.5246663689613342, 'learning_rate': 5.985010297615551e-06, 'epoch': 1.48}

 49%|████▉     | 6568/13317 [1:07:42<1:09:04,  1.63it/s]
 49%|████▉     | 6569/13317 [1:07:43<1:09:05,  1.63it/s]
                                                        
{'loss': 0.1855, 'grad_norm': 0.48999956250190735, 'learning_rate': 5.983725312977895e-06, 'epoch': 1.48}

 49%|████▉     | 6569/13317 [1:07:43<1:09:05,  1.63it/s]
 49%|████▉     | 6570/13317 [1:07:43<1:09:03,  1.63it/s]
                                                        
{'loss': 0.1723, 'grad_norm': 0.4386381506919861, 'learning_rate': 5.9824402607478924e-06, 'epoch': 1.48}

 49%|████▉     | 6570/13317 [1:07:43<1:09:03,  1.63it/s]
 49%|████▉     | 6571/13317 [1:07:44<1:09:04,  1.63it/s]
                                                        
{'loss': 0.1663, 'grad_norm': 0.41791966557502747, 'learning_rate': 5.981155141013842e-06, 'epoch': 1.48}

 49%|████▉     | 6571/13317 [1:07:44<1:09:04,  1.63it/s]
 49%|████▉     | 6572/13317 [1:07:44<1:09:02,  1.63it/s]
                                                        
{'loss': 0.2099, 'grad_norm': 0.575911819934845, 'learning_rate': 5.979869953864044e-06, 'epoch': 1.48}

 49%|████▉     | 6572/13317 [1:07:45<1:09:02,  1.63it/s]
 49%|████▉     | 6573/13317 [1:07:45<1:08:59,  1.63it/s]
                                                        
{'loss': 0.156, 'grad_norm': 0.4725673496723175, 'learning_rate': 5.9785846993868055e-06, 'epoch': 1.48}

 49%|████▉     | 6573/13317 [1:07:45<1:08:59,  1.63it/s]
 49%|████▉     | 6574/13317 [1:07:46<1:08:59,  1.63it/s]
                                                        
{'loss': 0.171, 'grad_norm': 0.46058449149131775, 'learning_rate': 5.977299377670437e-06, 'epoch': 1.48}

 49%|████▉     | 6574/13317 [1:07:46<1:08:59,  1.63it/s]
 49%|████▉     | 6575/13317 [1:07:46<1:08:56,  1.63it/s]
                                                        
{'loss': 0.1858, 'grad_norm': 0.4911617934703827, 'learning_rate': 5.976013988803253e-06, 'epoch': 1.48}

 49%|████▉     | 6575/13317 [1:07:46<1:08:56,  1.63it/s]
 49%|████▉     | 6576/13317 [1:07:47<1:09:00,  1.63it/s]
                                                        
{'loss': 0.2265, 'grad_norm': 0.6099352836608887, 'learning_rate': 5.974728532873574e-06, 'epoch': 1.48}

 49%|████▉     | 6576/13317 [1:07:47<1:09:00,  1.63it/s]
 49%|████▉     | 6577/13317 [1:07:48<1:08:59,  1.63it/s]
                                                        
{'loss': 0.2233, 'grad_norm': 0.6374669075012207, 'learning_rate': 5.973443009969726e-06, 'epoch': 1.48}

 49%|████▉     | 6577/13317 [1:07:48<1:08:59,  1.63it/s]
 49%|████▉     | 6578/13317 [1:07:48<1:08:58,  1.63it/s]
                                                        
{'loss': 0.1997, 'grad_norm': 0.496786892414093, 'learning_rate': 5.972157420180034e-06, 'epoch': 1.48}

 49%|████▉     | 6578/13317 [1:07:48<1:08:58,  1.63it/s]
 49%|████▉     | 6579/13317 [1:07:49<1:08:55,  1.63it/s]
                                                        
{'loss': 0.2092, 'grad_norm': 0.4955011308193207, 'learning_rate': 5.970871763592835e-06, 'epoch': 1.48}

 49%|████▉     | 6579/13317 [1:07:49<1:08:55,  1.63it/s]
 49%|████▉     | 6580/13317 [1:07:49<1:08:53,  1.63it/s]
                                                        
{'loss': 0.2001, 'grad_norm': 0.5732947587966919, 'learning_rate': 5.969586040296465e-06, 'epoch': 1.48}

 49%|████▉     | 6580/13317 [1:07:49<1:08:53,  1.63it/s]
 49%|████▉     | 6581/13317 [1:07:50<1:09:00,  1.63it/s]
                                                        
{'loss': 0.166, 'grad_norm': 0.5064725279808044, 'learning_rate': 5.968300250379268e-06, 'epoch': 1.48}

 49%|████▉     | 6581/13317 [1:07:50<1:09:00,  1.63it/s]
 49%|████▉     | 6582/13317 [1:07:51<1:08:56,  1.63it/s]
                                                        
{'loss': 0.1671, 'grad_norm': 0.4649851322174072, 'learning_rate': 5.967014393929593e-06, 'epoch': 1.48}

 49%|████▉     | 6582/13317 [1:07:51<1:08:56,  1.63it/s]
 49%|████▉     | 6583/13317 [1:07:51<1:08:54,  1.63it/s]
                                                        
{'loss': 0.1629, 'grad_norm': 0.44901490211486816, 'learning_rate': 5.965728471035788e-06, 'epoch': 1.48}

 49%|████▉     | 6583/13317 [1:07:51<1:08:54,  1.63it/s]
 49%|████▉     | 6584/13317 [1:07:52<1:08:54,  1.63it/s]
                                                        
{'loss': 0.173, 'grad_norm': 0.4447803199291229, 'learning_rate': 5.964442481786214e-06, 'epoch': 1.48}

 49%|████▉     | 6584/13317 [1:07:52<1:08:54,  1.63it/s]
 49%|████▉     | 6585/13317 [1:07:52<1:08:52,  1.63it/s]
                                                        
{'loss': 0.1938, 'grad_norm': 0.5554113388061523, 'learning_rate': 5.963156426269228e-06, 'epoch': 1.48}

 49%|████▉     | 6585/13317 [1:07:53<1:08:52,  1.63it/s]
 49%|████▉     | 6586/13317 [1:07:53<1:08:59,  1.63it/s]
                                                        
{'loss': 0.1914, 'grad_norm': 0.5102429986000061, 'learning_rate': 5.961870304573198e-06, 'epoch': 1.48}

 49%|████▉     | 6586/13317 [1:07:53<1:08:59,  1.63it/s]
 49%|████▉     | 6587/13317 [1:07:54<1:08:57,  1.63it/s]
                                                        
{'loss': 0.1825, 'grad_norm': 0.4660092890262604, 'learning_rate': 5.960584116786493e-06, 'epoch': 1.48}

 49%|████▉     | 6587/13317 [1:07:54<1:08:57,  1.63it/s]
 49%|████▉     | 6588/13317 [1:07:54<1:08:56,  1.63it/s]
                                                        
{'loss': 0.1717, 'grad_norm': 0.5024987459182739, 'learning_rate': 5.959297862997489e-06, 'epoch': 1.48}

 49%|████▉     | 6588/13317 [1:07:54<1:08:56,  1.63it/s]
 49%|████▉     | 6589/13317 [1:07:55<1:08:55,  1.63it/s]
                                                        
{'loss': 0.1979, 'grad_norm': 0.5660159587860107, 'learning_rate': 5.9580115432945655e-06, 'epoch': 1.48}

 49%|████▉     | 6589/13317 [1:07:55<1:08:55,  1.63it/s]
 49%|████▉     | 6590/13317 [1:07:56<1:08:53,  1.63it/s]
                                                        
{'loss': 0.1915, 'grad_norm': 0.47740206122398376, 'learning_rate': 5.956725157766104e-06, 'epoch': 1.48}

 49%|████▉     | 6590/13317 [1:07:56<1:08:53,  1.63it/s]
 49%|████▉     | 6591/13317 [1:07:56<1:08:55,  1.63it/s]
                                                        
{'loss': 0.1835, 'grad_norm': 0.5003652572631836, 'learning_rate': 5.955438706500495e-06, 'epoch': 1.48}

 49%|████▉     | 6591/13317 [1:07:56<1:08:55,  1.63it/s]
 50%|████▉     | 6592/13317 [1:07:57<1:08:51,  1.63it/s]
                                                        
{'loss': 0.2056, 'grad_norm': 0.4998016655445099, 'learning_rate': 5.95415218958613e-06, 'epoch': 1.49}

 50%|████▉     | 6592/13317 [1:07:57<1:08:51,  1.63it/s]
 50%|████▉     | 6593/13317 [1:07:57<1:08:50,  1.63it/s]
                                                        
{'loss': 0.2623, 'grad_norm': 0.5999817848205566, 'learning_rate': 5.9528656071114065e-06, 'epoch': 1.49}

 50%|████▉     | 6593/13317 [1:07:57<1:08:50,  1.63it/s]
 50%|████▉     | 6594/13317 [1:07:58<1:08:48,  1.63it/s]
                                                        
{'loss': 0.1671, 'grad_norm': 0.44621574878692627, 'learning_rate': 5.951578959164728e-06, 'epoch': 1.49}

 50%|████▉     | 6594/13317 [1:07:58<1:08:48,  1.63it/s]
 50%|████▉     | 6595/13317 [1:07:59<1:08:52,  1.63it/s]
                                                        
{'loss': 0.184, 'grad_norm': 0.46498873829841614, 'learning_rate': 5.950292245834498e-06, 'epoch': 1.49}

 50%|████▉     | 6595/13317 [1:07:59<1:08:52,  1.63it/s]
 50%|████▉     | 6596/13317 [1:07:59<1:08:53,  1.63it/s]
                                                        
{'loss': 0.1909, 'grad_norm': 0.5448911786079407, 'learning_rate': 5.9490054672091305e-06, 'epoch': 1.49}

 50%|████▉     | 6596/13317 [1:07:59<1:08:53,  1.63it/s]
 50%|████▉     | 6597/13317 [1:08:00<1:08:48,  1.63it/s]
                                                        
{'loss': 0.2442, 'grad_norm': 0.5843507647514343, 'learning_rate': 5.9477186233770365e-06, 'epoch': 1.49}

 50%|████▉     | 6597/13317 [1:08:00<1:08:48,  1.63it/s]
 50%|████▉     | 6598/13317 [1:08:00<1:08:48,  1.63it/s]
                                                        
{'loss': 0.2083, 'grad_norm': 0.6196820735931396, 'learning_rate': 5.946431714426641e-06, 'epoch': 1.49}

 50%|████▉     | 6598/13317 [1:08:01<1:08:48,  1.63it/s]
 50%|████▉     | 6599/13317 [1:08:01<1:08:44,  1.63it/s]
                                                        
{'loss': 0.1958, 'grad_norm': 0.5492133498191833, 'learning_rate': 5.9451447404463645e-06, 'epoch': 1.49}

 50%|████▉     | 6599/13317 [1:08:01<1:08:44,  1.63it/s]
 50%|████▉     | 6600/13317 [1:08:02<1:08:46,  1.63it/s]
                                                        
{'loss': 0.1914, 'grad_norm': 0.5102687478065491, 'learning_rate': 5.9438577015246375e-06, 'epoch': 1.49}

 50%|████▉     | 6600/13317 [1:08:02<1:08:46,  1.63it/s]
 50%|████▉     | 6601/13317 [1:08:02<1:08:51,  1.63it/s]
                                                        
{'loss': 0.2267, 'grad_norm': 0.5998116135597229, 'learning_rate': 5.942570597749892e-06, 'epoch': 1.49}

 50%|████▉     | 6601/13317 [1:08:02<1:08:51,  1.63it/s]
 50%|████▉     | 6602/13317 [1:08:03<1:08:48,  1.63it/s]
                                                        
{'loss': 0.2431, 'grad_norm': 0.5193089246749878, 'learning_rate': 5.941283429210568e-06, 'epoch': 1.49}

 50%|████▉     | 6602/13317 [1:08:03<1:08:48,  1.63it/s]
 50%|████▉     | 6603/13317 [1:08:03<1:08:42,  1.63it/s]
                                                        
{'loss': 0.198, 'grad_norm': 0.5452669858932495, 'learning_rate': 5.939996195995105e-06, 'epoch': 1.49}

 50%|████▉     | 6603/13317 [1:08:04<1:08:42,  1.63it/s]
 50%|████▉     | 6604/13317 [1:08:04<1:08:42,  1.63it/s]
                                                        
{'loss': 0.1658, 'grad_norm': 0.416278213262558, 'learning_rate': 5.938708898191951e-06, 'epoch': 1.49}

 50%|████▉     | 6604/13317 [1:08:04<1:08:42,  1.63it/s]
 50%|████▉     | 6605/13317 [1:08:05<1:08:44,  1.63it/s]
                                                        
{'loss': 0.2089, 'grad_norm': 0.5342127084732056, 'learning_rate': 5.937421535889556e-06, 'epoch': 1.49}

 50%|████▉     | 6605/13317 [1:08:05<1:08:44,  1.63it/s]
 50%|████▉     | 6606/13317 [1:08:05<1:08:49,  1.63it/s]
                                                        
{'loss': 0.2423, 'grad_norm': 0.5336310267448425, 'learning_rate': 5.936134109176376e-06, 'epoch': 1.49}

 50%|████▉     | 6606/13317 [1:08:05<1:08:49,  1.63it/s]
 50%|████▉     | 6607/13317 [1:08:06<1:08:43,  1.63it/s]
                                                        
{'loss': 0.2998, 'grad_norm': 0.5918105244636536, 'learning_rate': 5.934846618140871e-06, 'epoch': 1.49}

 50%|████▉     | 6607/13317 [1:08:06<1:08:43,  1.63it/s]
 50%|████▉     | 6608/13317 [1:08:07<1:08:40,  1.63it/s]
                                                        
{'loss': 0.182, 'grad_norm': 0.5087835192680359, 'learning_rate': 5.933559062871503e-06, 'epoch': 1.49}

 50%|████▉     | 6608/13317 [1:08:07<1:08:40,  1.63it/s]
 50%|████▉     | 6609/13317 [1:08:07<1:08:38,  1.63it/s]
                                                        
{'loss': 0.1949, 'grad_norm': 0.5700437426567078, 'learning_rate': 5.932271443456745e-06, 'epoch': 1.49}

 50%|████▉     | 6609/13317 [1:08:07<1:08:38,  1.63it/s]
 50%|████▉     | 6610/13317 [1:08:08<1:08:38,  1.63it/s]
                                                        
{'loss': 0.2023, 'grad_norm': 0.5089877247810364, 'learning_rate': 5.930983759985066e-06, 'epoch': 1.49}

 50%|████▉     | 6610/13317 [1:08:08<1:08:38,  1.63it/s]
 50%|████▉     | 6611/13317 [1:08:08<1:08:44,  1.63it/s]
                                                        
{'loss': 0.1988, 'grad_norm': 0.49788233637809753, 'learning_rate': 5.929696012544948e-06, 'epoch': 1.49}

 50%|████▉     | 6611/13317 [1:08:09<1:08:44,  1.63it/s]
 50%|████▉     | 6612/13317 [1:08:09<1:08:40,  1.63it/s]
                                                        
{'loss': 0.234, 'grad_norm': 0.5852102637290955, 'learning_rate': 5.928408201224866e-06, 'epoch': 1.49}

 50%|████▉     | 6612/13317 [1:08:09<1:08:40,  1.63it/s]
 50%|████▉     | 6613/13317 [1:08:10<1:08:36,  1.63it/s]
                                                        
{'loss': 0.1749, 'grad_norm': 0.4897296130657196, 'learning_rate': 5.927120326113315e-06, 'epoch': 1.49}

 50%|████▉     | 6613/13317 [1:08:10<1:08:36,  1.63it/s]
 50%|████▉     | 6614/13317 [1:08:10<1:08:35,  1.63it/s]
                                                        
{'loss': 0.2199, 'grad_norm': 0.5962900519371033, 'learning_rate': 5.925832387298779e-06, 'epoch': 1.49}

 50%|████▉     | 6614/13317 [1:08:10<1:08:35,  1.63it/s]
 50%|████▉     | 6615/13317 [1:08:11<1:08:36,  1.63it/s]
                                                        
{'loss': 0.2193, 'grad_norm': 0.71827232837677, 'learning_rate': 5.9245443848697544e-06, 'epoch': 1.49}

 50%|████▉     | 6615/13317 [1:08:11<1:08:36,  1.63it/s]
 50%|████▉     | 6616/13317 [1:08:11<1:08:42,  1.63it/s]
                                                        
{'loss': 0.2232, 'grad_norm': 0.5499694347381592, 'learning_rate': 5.923256318914743e-06, 'epoch': 1.49}

 50%|████▉     | 6616/13317 [1:08:12<1:08:42,  1.63it/s]
 50%|████▉     | 6617/13317 [1:08:12<1:08:38,  1.63it/s]
                                                        
{'loss': 0.246, 'grad_norm': 0.6112617254257202, 'learning_rate': 5.921968189522245e-06, 'epoch': 1.49}

 50%|████▉     | 6617/13317 [1:08:12<1:08:38,  1.63it/s]
 50%|████▉     | 6618/13317 [1:08:13<1:08:37,  1.63it/s]
                                                        
{'loss': 0.1773, 'grad_norm': 0.5218244194984436, 'learning_rate': 5.920679996780772e-06, 'epoch': 1.49}

 50%|████▉     | 6618/13317 [1:08:13<1:08:37,  1.63it/s]
 50%|████▉     | 6619/13317 [1:08:13<1:08:32,  1.63it/s]
                                                        
{'loss': 0.1926, 'grad_norm': 0.5306541919708252, 'learning_rate': 5.919391740778833e-06, 'epoch': 1.49}

 50%|████▉     | 6619/13317 [1:08:13<1:08:32,  1.63it/s]
 50%|████▉     | 6620/13317 [1:08:14<1:08:36,  1.63it/s]
                                                        
{'loss': 0.2211, 'grad_norm': 0.5251107811927795, 'learning_rate': 5.918103421604949e-06, 'epoch': 1.49}

 50%|████▉     | 6620/13317 [1:08:14<1:08:36,  1.63it/s]
 50%|████▉     | 6621/13317 [1:08:15<1:09:02,  1.62it/s]
                                                        
{'loss': 0.1428, 'grad_norm': 0.4023112654685974, 'learning_rate': 5.9168150393476374e-06, 'epoch': 1.49}

 50%|████▉     | 6621/13317 [1:08:15<1:09:02,  1.62it/s]
 50%|████▉     | 6622/13317 [1:08:15<1:08:51,  1.62it/s]
                                                        
{'loss': 0.18, 'grad_norm': 0.4940793216228485, 'learning_rate': 5.915526594095427e-06, 'epoch': 1.49}

 50%|████▉     | 6622/13317 [1:08:15<1:08:51,  1.62it/s]
 50%|████▉     | 6623/13317 [1:08:16<1:08:40,  1.62it/s]
                                                        
{'loss': 0.1752, 'grad_norm': 0.48338961601257324, 'learning_rate': 5.914238085936843e-06, 'epoch': 1.49}

 50%|████▉     | 6623/13317 [1:08:16<1:08:40,  1.62it/s]
 50%|████▉     | 6624/13317 [1:08:16<1:08:37,  1.63it/s]
                                                        
{'loss': 0.272, 'grad_norm': 0.7070180773735046, 'learning_rate': 5.912949514960424e-06, 'epoch': 1.49}

 50%|████▉     | 6624/13317 [1:08:17<1:08:37,  1.63it/s]
 50%|████▉     | 6625/13317 [1:08:17<1:08:56,  1.62it/s]
                                                        
{'loss': 0.2217, 'grad_norm': 0.5358129739761353, 'learning_rate': 5.911660881254708e-06, 'epoch': 1.49}

 50%|████▉     | 6625/13317 [1:08:17<1:08:56,  1.62it/s]
 50%|████▉     | 6626/13317 [1:08:18<1:08:52,  1.62it/s]
                                                        
{'loss': 0.1645, 'grad_norm': 0.5047697424888611, 'learning_rate': 5.910372184908235e-06, 'epoch': 1.49}

 50%|████▉     | 6626/13317 [1:08:18<1:08:52,  1.62it/s]
 50%|████▉     | 6627/13317 [1:08:18<1:08:44,  1.62it/s]
                                                        
{'loss': 0.1831, 'grad_norm': 0.5210901498794556, 'learning_rate': 5.909083426009554e-06, 'epoch': 1.49}

 50%|████▉     | 6627/13317 [1:08:18<1:08:44,  1.62it/s]
 50%|████▉     | 6628/13317 [1:08:19<1:08:37,  1.62it/s]
                                                        
{'loss': 0.1899, 'grad_norm': 0.4823595881462097, 'learning_rate': 5.907794604647217e-06, 'epoch': 1.49}

 50%|████▉     | 6628/13317 [1:08:19<1:08:37,  1.62it/s]
 50%|████▉     | 6629/13317 [1:08:19<1:08:35,  1.62it/s]
                                                        
{'loss': 0.2107, 'grad_norm': 0.5797916650772095, 'learning_rate': 5.9065057209097774e-06, 'epoch': 1.49}

 50%|████▉     | 6629/13317 [1:08:20<1:08:35,  1.62it/s]
 50%|████▉     | 6630/13317 [1:08:20<1:08:36,  1.62it/s]
                                                        
{'loss': 0.3106, 'grad_norm': 0.701606273651123, 'learning_rate': 5.905216774885797e-06, 'epoch': 1.49}

 50%|████▉     | 6630/13317 [1:08:20<1:08:36,  1.62it/s]
 50%|████▉     | 6631/13317 [1:08:21<1:08:37,  1.62it/s]
                                                        
{'loss': 0.1721, 'grad_norm': 0.49482178688049316, 'learning_rate': 5.9039277666638405e-06, 'epoch': 1.49}

 50%|████▉     | 6631/13317 [1:08:21<1:08:37,  1.62it/s]
 50%|████▉     | 6632/13317 [1:08:21<1:08:34,  1.62it/s]
                                                        
{'loss': 0.1641, 'grad_norm': 0.44517943263053894, 'learning_rate': 5.902638696332475e-06, 'epoch': 1.49}

 50%|████▉     | 6632/13317 [1:08:21<1:08:34,  1.62it/s]
 50%|████▉     | 6633/13317 [1:08:22<1:08:29,  1.63it/s]
                                                        
{'loss': 0.155, 'grad_norm': 0.497104287147522, 'learning_rate': 5.901349563980273e-06, 'epoch': 1.49}

 50%|████▉     | 6633/13317 [1:08:22<1:08:29,  1.63it/s]
 50%|████▉     | 6634/13317 [1:08:23<1:08:27,  1.63it/s]
                                                        
{'loss': 0.2183, 'grad_norm': 0.5590978264808655, 'learning_rate': 5.900060369695812e-06, 'epoch': 1.49}

 50%|████▉     | 6634/13317 [1:08:23<1:08:27,  1.63it/s]
 50%|████▉     | 6635/13317 [1:08:23<1:08:30,  1.63it/s]
                                                        
{'loss': 0.1903, 'grad_norm': 0.4577925503253937, 'learning_rate': 5.8987711135676754e-06, 'epoch': 1.49}

 50%|████▉     | 6635/13317 [1:08:23<1:08:30,  1.63it/s]
 50%|████▉     | 6636/13317 [1:08:24<1:08:35,  1.62it/s]
                                                        
{'loss': 0.1993, 'grad_norm': 0.6113492250442505, 'learning_rate': 5.897481795684447e-06, 'epoch': 1.49}

 50%|████▉     | 6636/13317 [1:08:24<1:08:35,  1.62it/s]
 50%|████▉     | 6637/13317 [1:08:24<1:08:29,  1.63it/s]
                                                        
{'loss': 0.195, 'grad_norm': 0.46884122490882874, 'learning_rate': 5.896192416134714e-06, 'epoch': 1.5}

 50%|████▉     | 6637/13317 [1:08:25<1:08:29,  1.63it/s]
 50%|████▉     | 6638/13317 [1:08:25<1:08:25,  1.63it/s]
                                                        
{'loss': 0.1915, 'grad_norm': 0.4771753251552582, 'learning_rate': 5.894902975007075e-06, 'epoch': 1.5}

 50%|████▉     | 6638/13317 [1:08:25<1:08:25,  1.63it/s]
 50%|████▉     | 6639/13317 [1:08:26<1:08:20,  1.63it/s]
                                                        
{'loss': 0.267, 'grad_norm': 0.5983549356460571, 'learning_rate': 5.893613472390125e-06, 'epoch': 1.5}

 50%|████▉     | 6639/13317 [1:08:26<1:08:20,  1.63it/s]
 50%|████▉     | 6640/13317 [1:08:26<1:08:19,  1.63it/s]
                                                        
{'loss': 0.2077, 'grad_norm': 0.5775431394577026, 'learning_rate': 5.892323908372469e-06, 'epoch': 1.5}

 50%|████▉     | 6640/13317 [1:08:26<1:08:19,  1.63it/s]
 50%|████▉     | 6641/13317 [1:08:27<1:08:24,  1.63it/s]
                                                        
{'loss': 0.2551, 'grad_norm': 0.66911381483078, 'learning_rate': 5.891034283042712e-06, 'epoch': 1.5}

 50%|████▉     | 6641/13317 [1:08:27<1:08:24,  1.63it/s]
 50%|████▉     | 6642/13317 [1:08:27<1:08:21,  1.63it/s]
                                                        
{'loss': 0.1945, 'grad_norm': 0.6519255042076111, 'learning_rate': 5.889744596489465e-06, 'epoch': 1.5}

 50%|████▉     | 6642/13317 [1:08:28<1:08:21,  1.63it/s]
 50%|████▉     | 6643/13317 [1:08:28<1:08:21,  1.63it/s]
                                                        
{'loss': 0.2325, 'grad_norm': 0.6123174428939819, 'learning_rate': 5.888454848801345e-06, 'epoch': 1.5}

 50%|████▉     | 6643/13317 [1:08:28<1:08:21,  1.63it/s]
 50%|████▉     | 6644/13317 [1:08:29<1:08:17,  1.63it/s]
                                                        
{'loss': 0.2812, 'grad_norm': 0.6282447576522827, 'learning_rate': 5.887165040066967e-06, 'epoch': 1.5}

 50%|████▉     | 6644/13317 [1:08:29<1:08:17,  1.63it/s]
 50%|████▉     | 6645/13317 [1:08:29<1:08:17,  1.63it/s]
                                                        
{'loss': 0.1938, 'grad_norm': 0.5188471674919128, 'learning_rate': 5.88587517037496e-06, 'epoch': 1.5}

 50%|████▉     | 6645/13317 [1:08:29<1:08:17,  1.63it/s]
 50%|████▉     | 6646/13317 [1:08:30<1:08:22,  1.63it/s]
                                                        
{'loss': 0.2173, 'grad_norm': 0.5621569752693176, 'learning_rate': 5.884585239813946e-06, 'epoch': 1.5}

 50%|████▉     | 6646/13317 [1:08:30<1:08:22,  1.63it/s]
 50%|████▉     | 6647/13317 [1:08:31<1:08:20,  1.63it/s]
                                                        
{'loss': 0.203, 'grad_norm': 0.5228132009506226, 'learning_rate': 5.883295248472563e-06, 'epoch': 1.5}

 50%|████▉     | 6647/13317 [1:08:31<1:08:20,  1.63it/s]
 50%|████▉     | 6648/13317 [1:08:31<1:08:17,  1.63it/s]
                                                        
{'loss': 0.1701, 'grad_norm': 0.44034314155578613, 'learning_rate': 5.882005196439442e-06, 'epoch': 1.5}

 50%|████▉     | 6648/13317 [1:08:31<1:08:17,  1.63it/s]
 50%|████▉     | 6649/13317 [1:08:32<1:08:14,  1.63it/s]
                                                        
{'loss': 0.211, 'grad_norm': 0.5204157829284668, 'learning_rate': 5.880715083803225e-06, 'epoch': 1.5}

 50%|████▉     | 6649/13317 [1:08:32<1:08:14,  1.63it/s]
 50%|████▉     | 6650/13317 [1:08:32<1:08:32,  1.62it/s]
                                                        
{'loss': 0.1783, 'grad_norm': 0.46005281805992126, 'learning_rate': 5.879424910652558e-06, 'epoch': 1.5}

 50%|████▉     | 6650/13317 [1:08:33<1:08:32,  1.62it/s]
 50%|████▉     | 6651/13317 [1:08:33<1:08:28,  1.62it/s]
                                                        
{'loss': 0.2265, 'grad_norm': 0.5924535393714905, 'learning_rate': 5.878134677076087e-06, 'epoch': 1.5}

 50%|████▉     | 6651/13317 [1:08:33<1:08:28,  1.62it/s]
 50%|████▉     | 6652/13317 [1:08:34<1:08:21,  1.62it/s]
                                                        
{'loss': 0.2004, 'grad_norm': 0.48736312985420227, 'learning_rate': 5.876844383162468e-06, 'epoch': 1.5}

 50%|████▉     | 6652/13317 [1:08:34<1:08:21,  1.62it/s]
 50%|████▉     | 6653/13317 [1:08:34<1:08:16,  1.63it/s]
                                                        
{'loss': 0.2323, 'grad_norm': 0.6660680174827576, 'learning_rate': 5.875554029000353e-06, 'epoch': 1.5}

 50%|████▉     | 6653/13317 [1:08:34<1:08:16,  1.63it/s]
 50%|████▉     | 6654/13317 [1:08:35<1:08:12,  1.63it/s]
                                                        
{'loss': 0.2536, 'grad_norm': 0.5510594844818115, 'learning_rate': 5.8742636146784075e-06, 'epoch': 1.5}

 50%|████▉     | 6654/13317 [1:08:35<1:08:12,  1.63it/s]
 50%|████▉     | 6655/13317 [1:08:35<1:08:15,  1.63it/s]
                                                        
{'loss': 0.254, 'grad_norm': 0.5763795971870422, 'learning_rate': 5.872973140285293e-06, 'epoch': 1.5}

 50%|████▉     | 6655/13317 [1:08:36<1:08:15,  1.63it/s]
 50%|████▉     | 6656/13317 [1:08:36<1:08:19,  1.62it/s]
                                                        
{'loss': 0.1908, 'grad_norm': 0.4775589406490326, 'learning_rate': 5.871682605909682e-06, 'epoch': 1.5}

 50%|████▉     | 6656/13317 [1:08:36<1:08:19,  1.62it/s]
 50%|████▉     | 6657/13317 [1:08:37<1:08:14,  1.63it/s]
                                                        
{'loss': 0.2078, 'grad_norm': 0.5346893072128296, 'learning_rate': 5.870392011640246e-06, 'epoch': 1.5}

 50%|████▉     | 6657/13317 [1:08:37<1:08:14,  1.63it/s]
 50%|████▉     | 6658/13317 [1:08:37<1:08:12,  1.63it/s]
                                                        
{'loss': 0.2084, 'grad_norm': 0.4714047610759735, 'learning_rate': 5.869101357565663e-06, 'epoch': 1.5}

 50%|████▉     | 6658/13317 [1:08:37<1:08:12,  1.63it/s]
 50%|█████     | 6659/13317 [1:08:38<1:08:08,  1.63it/s]
                                                        
{'loss': 0.2445, 'grad_norm': 0.57242751121521, 'learning_rate': 5.8678106437746156e-06, 'epoch': 1.5}

 50%|█████     | 6659/13317 [1:08:38<1:08:08,  1.63it/s]
 50%|█████     | 6660/13317 [1:08:39<1:08:09,  1.63it/s]
                                                        
{'loss': 0.2162, 'grad_norm': 0.48472410440444946, 'learning_rate': 5.866519870355787e-06, 'epoch': 1.5}

 50%|█████     | 6660/13317 [1:08:39<1:08:09,  1.63it/s]
 50%|█████     | 6661/13317 [1:08:39<1:08:14,  1.63it/s]
                                                        
{'loss': 0.2284, 'grad_norm': 0.4678800404071808, 'learning_rate': 5.865229037397868e-06, 'epoch': 1.5}

 50%|█████     | 6661/13317 [1:08:39<1:08:14,  1.63it/s]
 50%|█████     | 6662/13317 [1:08:40<1:08:10,  1.63it/s]
                                                        
{'loss': 0.1955, 'grad_norm': 0.5334163904190063, 'learning_rate': 5.863938144989554e-06, 'epoch': 1.5}

 50%|█████     | 6662/13317 [1:08:40<1:08:10,  1.63it/s]
 50%|█████     | 6663/13317 [1:08:40<1:08:09,  1.63it/s]
                                                        
{'loss': 0.1944, 'grad_norm': 0.5010931491851807, 'learning_rate': 5.862647193219542e-06, 'epoch': 1.5}

 50%|█████     | 6663/13317 [1:08:41<1:08:09,  1.63it/s]
 50%|█████     | 6664/13317 [1:08:41<1:08:06,  1.63it/s]
                                                        
{'loss': 0.1679, 'grad_norm': 0.43667909502983093, 'learning_rate': 5.861356182176533e-06, 'epoch': 1.5}

 50%|█████     | 6664/13317 [1:08:41<1:08:06,  1.63it/s]
 50%|█████     | 6665/13317 [1:08:42<1:08:10,  1.63it/s]
                                                        
{'loss': 0.1762, 'grad_norm': 0.4714881181716919, 'learning_rate': 5.860065111949236e-06, 'epoch': 1.5}

 50%|█████     | 6665/13317 [1:08:42<1:08:10,  1.63it/s]
 50%|█████     | 6666/13317 [1:08:42<1:08:13,  1.62it/s]
                                                        
{'loss': 0.1935, 'grad_norm': 0.45897161960601807, 'learning_rate': 5.858773982626357e-06, 'epoch': 1.5}

 50%|█████     | 6666/13317 [1:08:42<1:08:13,  1.62it/s]
 50%|█████     | 6667/13317 [1:08:43<1:08:10,  1.63it/s]
                                                        
{'loss': 0.2406, 'grad_norm': 0.5348417162895203, 'learning_rate': 5.857482794296614e-06, 'epoch': 1.5}

 50%|█████     | 6667/13317 [1:08:43<1:08:10,  1.63it/s]
 50%|█████     | 6668/13317 [1:08:43<1:08:04,  1.63it/s]
                                                        
{'loss': 0.1713, 'grad_norm': 0.4263274371623993, 'learning_rate': 5.856191547048723e-06, 'epoch': 1.5}

 50%|█████     | 6668/13317 [1:08:44<1:08:04,  1.63it/s]
 50%|█████     | 6669/13317 [1:08:44<1:08:00,  1.63it/s]
                                                        
{'loss': 0.1566, 'grad_norm': 0.3788835406303406, 'learning_rate': 5.854900240971408e-06, 'epoch': 1.5}

 50%|█████     | 6669/13317 [1:08:44<1:08:00,  1.63it/s]
 50%|█████     | 6670/13317 [1:08:45<1:08:02,  1.63it/s]
                                                        
{'loss': 0.1637, 'grad_norm': 0.4365832805633545, 'learning_rate': 5.853608876153395e-06, 'epoch': 1.5}

 50%|█████     | 6670/13317 [1:08:45<1:08:02,  1.63it/s]
 50%|█████     | 6671/13317 [1:08:45<1:08:23,  1.62it/s]
                                                        
{'loss': 0.2068, 'grad_norm': 0.5688385367393494, 'learning_rate': 5.8523174526834125e-06, 'epoch': 1.5}

 50%|█████     | 6671/13317 [1:08:45<1:08:23,  1.62it/s]
 50%|█████     | 6672/13317 [1:08:46<1:08:15,  1.62it/s]
                                                        
{'loss': 0.2228, 'grad_norm': 0.5535288453102112, 'learning_rate': 5.851025970650198e-06, 'epoch': 1.5}

 50%|█████     | 6672/13317 [1:08:46<1:08:15,  1.62it/s]
 50%|█████     | 6673/13317 [1:08:47<1:08:11,  1.62it/s]
                                                        
{'loss': 0.1894, 'grad_norm': 0.4771668612957001, 'learning_rate': 5.849734430142486e-06, 'epoch': 1.5}

 50%|█████     | 6673/13317 [1:08:47<1:08:11,  1.62it/s]
 50%|█████     | 6674/13317 [1:08:47<1:08:08,  1.62it/s]
                                                        
{'loss': 0.1831, 'grad_norm': 0.4709009826183319, 'learning_rate': 5.848442831249025e-06, 'epoch': 1.5}

 50%|█████     | 6674/13317 [1:08:47<1:08:08,  1.62it/s]
 50%|█████     | 6675/13317 [1:08:48<1:08:06,  1.63it/s]
                                                        
{'loss': 0.2578, 'grad_norm': 0.6500061750411987, 'learning_rate': 5.847151174058556e-06, 'epoch': 1.5}

 50%|█████     | 6675/13317 [1:08:48<1:08:06,  1.63it/s]
 50%|█████     | 6676/13317 [1:08:48<1:08:20,  1.62it/s]
                                                        
{'loss': 0.207, 'grad_norm': 0.5486708283424377, 'learning_rate': 5.845859458659833e-06, 'epoch': 1.5}

 50%|█████     | 6676/13317 [1:08:49<1:08:20,  1.62it/s]
 50%|█████     | 6677/13317 [1:08:49<1:08:11,  1.62it/s]
                                                        
{'loss': 0.1947, 'grad_norm': 0.4656673073768616, 'learning_rate': 5.8445676851416064e-06, 'epoch': 1.5}

 50%|█████     | 6677/13317 [1:08:49<1:08:11,  1.62it/s]
 50%|█████     | 6678/13317 [1:08:50<1:08:05,  1.63it/s]
                                                        
{'loss': 0.2518, 'grad_norm': 0.5925135612487793, 'learning_rate': 5.843275853592639e-06, 'epoch': 1.5}

 50%|█████     | 6678/13317 [1:08:50<1:08:05,  1.63it/s]
 50%|█████     | 6679/13317 [1:08:50<1:08:01,  1.63it/s]
                                                        
{'loss': 0.2314, 'grad_norm': 0.6336131691932678, 'learning_rate': 5.841983964101694e-06, 'epoch': 1.5}

 50%|█████     | 6679/13317 [1:08:50<1:08:01,  1.63it/s]
 50%|█████     | 6680/13317 [1:08:51<1:08:02,  1.63it/s]
                                                        
{'loss': 0.2681, 'grad_norm': 0.5811524391174316, 'learning_rate': 5.840692016757532e-06, 'epoch': 1.5}

 50%|█████     | 6680/13317 [1:08:51<1:08:02,  1.63it/s]
 50%|█████     | 6681/13317 [1:08:51<1:08:04,  1.62it/s]
                                                        
{'loss': 0.1554, 'grad_norm': 0.4961339235305786, 'learning_rate': 5.839400011648929e-06, 'epoch': 1.51}

 50%|█████     | 6681/13317 [1:08:52<1:08:04,  1.62it/s]
 50%|█████     | 6682/13317 [1:08:52<1:07:58,  1.63it/s]
                                                        
{'loss': 0.2166, 'grad_norm': 0.5519833564758301, 'learning_rate': 5.838107948864656e-06, 'epoch': 1.51}

 50%|█████     | 6682/13317 [1:08:52<1:07:58,  1.63it/s]
 50%|█████     | 6683/13317 [1:08:53<1:07:56,  1.63it/s]
                                                        
{'loss': 0.178, 'grad_norm': 0.5037800669670105, 'learning_rate': 5.836815828493494e-06, 'epoch': 1.51}

 50%|█████     | 6683/13317 [1:08:53<1:07:56,  1.63it/s]
 50%|█████     | 6684/13317 [1:08:53<1:07:55,  1.63it/s]
                                                        
{'loss': 0.1848, 'grad_norm': 0.514323353767395, 'learning_rate': 5.835523650624224e-06, 'epoch': 1.51}

 50%|█████     | 6684/13317 [1:08:53<1:07:55,  1.63it/s]
 50%|█████     | 6685/13317 [1:08:54<1:07:53,  1.63it/s]
                                                        
{'loss': 0.1653, 'grad_norm': 0.4387172758579254, 'learning_rate': 5.834231415345633e-06, 'epoch': 1.51}

 50%|█████     | 6685/13317 [1:08:54<1:07:53,  1.63it/s]
 50%|█████     | 6686/13317 [1:08:55<1:07:59,  1.63it/s]
                                                        
{'loss': 0.3239, 'grad_norm': 0.6010230183601379, 'learning_rate': 5.832939122746511e-06, 'epoch': 1.51}

 50%|█████     | 6686/13317 [1:08:55<1:07:59,  1.63it/s]
 50%|█████     | 6687/13317 [1:08:55<1:07:55,  1.63it/s]
                                                        
{'loss': 0.2045, 'grad_norm': 0.5289722084999084, 'learning_rate': 5.831646772915651e-06, 'epoch': 1.51}

 50%|█████     | 6687/13317 [1:08:55<1:07:55,  1.63it/s]
 50%|█████     | 6688/13317 [1:08:56<1:07:52,  1.63it/s]
                                                        
{'loss': 0.158, 'grad_norm': 0.4281359016895294, 'learning_rate': 5.830354365941853e-06, 'epoch': 1.51}

 50%|█████     | 6688/13317 [1:08:56<1:07:52,  1.63it/s]
 50%|█████     | 6689/13317 [1:08:56<1:07:51,  1.63it/s]
                                                        
{'loss': 0.238, 'grad_norm': 0.6099297404289246, 'learning_rate': 5.829061901913917e-06, 'epoch': 1.51}

 50%|█████     | 6689/13317 [1:08:57<1:07:51,  1.63it/s]
 50%|█████     | 6690/13317 [1:08:57<1:07:52,  1.63it/s]
                                                        
{'loss': 0.1938, 'grad_norm': 0.5400424599647522, 'learning_rate': 5.82776938092065e-06, 'epoch': 1.51}

 50%|█████     | 6690/13317 [1:08:57<1:07:52,  1.63it/s]
 50%|█████     | 6691/13317 [1:08:58<1:07:58,  1.62it/s]
                                                        
{'loss': 0.2175, 'grad_norm': 0.5433862805366516, 'learning_rate': 5.826476803050863e-06, 'epoch': 1.51}

 50%|█████     | 6691/13317 [1:08:58<1:07:58,  1.62it/s]
 50%|█████     | 6692/13317 [1:08:58<1:07:55,  1.63it/s]
                                                        
{'loss': 0.2009, 'grad_norm': 0.500703752040863, 'learning_rate': 5.82518416839337e-06, 'epoch': 1.51}

 50%|█████     | 6692/13317 [1:08:58<1:07:55,  1.63it/s]
 50%|█████     | 6693/13317 [1:08:59<1:07:50,  1.63it/s]
                                                        
{'loss': 0.2138, 'grad_norm': 0.5197546482086182, 'learning_rate': 5.823891477036984e-06, 'epoch': 1.51}

 50%|█████     | 6693/13317 [1:08:59<1:07:50,  1.63it/s]
 50%|█████     | 6694/13317 [1:08:59<1:07:49,  1.63it/s]
                                                        
{'loss': 0.2347, 'grad_norm': 0.7996851801872253, 'learning_rate': 5.822598729070532e-06, 'epoch': 1.51}

 50%|█████     | 6694/13317 [1:09:00<1:07:49,  1.63it/s]
 50%|█████     | 6695/13317 [1:09:00<1:07:51,  1.63it/s]
                                                        
{'loss': 0.1772, 'grad_norm': 0.45325547456741333, 'learning_rate': 5.821305924582837e-06, 'epoch': 1.51}

 50%|█████     | 6695/13317 [1:09:00<1:07:51,  1.63it/s]
 50%|█████     | 6696/13317 [1:09:01<1:07:57,  1.62it/s]
                                                        
{'loss': 0.2063, 'grad_norm': 0.5461183786392212, 'learning_rate': 5.820013063662729e-06, 'epoch': 1.51}

 50%|█████     | 6696/13317 [1:09:01<1:07:57,  1.62it/s]
 50%|█████     | 6697/13317 [1:09:01<1:07:52,  1.63it/s]
                                                        
{'loss': 0.2149, 'grad_norm': 0.5922505855560303, 'learning_rate': 5.818720146399043e-06, 'epoch': 1.51}

 50%|█████     | 6697/13317 [1:09:01<1:07:52,  1.63it/s]
 50%|█████     | 6698/13317 [1:09:02<1:07:48,  1.63it/s]
                                                        
{'loss': 0.2009, 'grad_norm': 0.49937912821769714, 'learning_rate': 5.8174271728806105e-06, 'epoch': 1.51}

 50%|█████     | 6698/13317 [1:09:02<1:07:48,  1.63it/s]
 50%|█████     | 6699/13317 [1:09:03<1:07:45,  1.63it/s]
                                                        
{'loss': 0.1946, 'grad_norm': 0.5612397193908691, 'learning_rate': 5.8161341431962795e-06, 'epoch': 1.51}

 50%|█████     | 6699/13317 [1:09:03<1:07:45,  1.63it/s]
 50%|█████     | 6700/13317 [1:09:03<1:07:45,  1.63it/s]
                                                        
{'loss': 0.1888, 'grad_norm': 0.5191967487335205, 'learning_rate': 5.814841057434889e-06, 'epoch': 1.51}

 50%|█████     | 6700/13317 [1:09:03<1:07:45,  1.63it/s]
 50%|█████     | 6701/13317 [1:09:04<1:08:01,  1.62it/s]
                                                        
{'loss': 0.2171, 'grad_norm': 0.6015994548797607, 'learning_rate': 5.813547915685293e-06, 'epoch': 1.51}

 50%|█████     | 6701/13317 [1:09:04<1:08:01,  1.62it/s]
 50%|█████     | 6702/13317 [1:09:04<1:07:53,  1.62it/s]
                                                        
{'loss': 0.1697, 'grad_norm': 0.5140175819396973, 'learning_rate': 5.812254718036337e-06, 'epoch': 1.51}

 50%|█████     | 6702/13317 [1:09:05<1:07:53,  1.62it/s]
 50%|█████     | 6703/13317 [1:09:05<1:07:47,  1.63it/s]
                                                        
{'loss': 0.2347, 'grad_norm': 0.7026673555374146, 'learning_rate': 5.810961464576885e-06, 'epoch': 1.51}

 50%|█████     | 6703/13317 [1:09:05<1:07:47,  1.63it/s]
 50%|█████     | 6704/13317 [1:09:06<1:07:43,  1.63it/s]
                                                        
{'loss': 0.2349, 'grad_norm': 0.5147829055786133, 'learning_rate': 5.809668155395793e-06, 'epoch': 1.51}

 50%|█████     | 6704/13317 [1:09:06<1:07:43,  1.63it/s]
 50%|█████     | 6705/13317 [1:09:06<1:07:49,  1.62it/s]
                                                        
{'loss': 0.175, 'grad_norm': 0.4678829610347748, 'learning_rate': 5.808374790581926e-06, 'epoch': 1.51}

 50%|█████     | 6705/13317 [1:09:06<1:07:49,  1.62it/s]
 50%|█████     | 6706/13317 [1:09:07<1:07:49,  1.62it/s]
                                                        
{'loss': 0.1493, 'grad_norm': 0.4209100902080536, 'learning_rate': 5.807081370224153e-06, 'epoch': 1.51}

 50%|█████     | 6706/13317 [1:09:07<1:07:49,  1.62it/s]
 50%|█████     | 6707/13317 [1:09:07<1:07:47,  1.63it/s]
                                                        
{'loss': 0.2089, 'grad_norm': 0.6128312945365906, 'learning_rate': 5.805787894411341e-06, 'epoch': 1.51}

 50%|█████     | 6707/13317 [1:09:08<1:07:47,  1.63it/s]
 50%|█████     | 6708/13317 [1:09:08<1:07:43,  1.63it/s]
                                                        
{'loss': 0.196, 'grad_norm': 0.49185216426849365, 'learning_rate': 5.804494363232372e-06, 'epoch': 1.51}

 50%|█████     | 6708/13317 [1:09:08<1:07:43,  1.63it/s]
 50%|█████     | 6709/13317 [1:09:09<1:07:39,  1.63it/s]
                                                        
{'loss': 0.2283, 'grad_norm': 0.5498868823051453, 'learning_rate': 5.80320077677612e-06, 'epoch': 1.51}

 50%|█████     | 6709/13317 [1:09:09<1:07:39,  1.63it/s]
 50%|█████     | 6710/13317 [1:09:09<1:07:41,  1.63it/s]
                                                        
{'loss': 0.1881, 'grad_norm': 0.474558562040329, 'learning_rate': 5.80190713513147e-06, 'epoch': 1.51}

 50%|█████     | 6710/13317 [1:09:09<1:07:41,  1.63it/s]
 50%|█████     | 6711/13317 [1:09:10<1:07:58,  1.62it/s]
                                                        
{'loss': 0.2073, 'grad_norm': 0.56226646900177, 'learning_rate': 5.80061343838731e-06, 'epoch': 1.51}

 50%|█████     | 6711/13317 [1:09:10<1:07:58,  1.62it/s]
 50%|█████     | 6712/13317 [1:09:11<1:07:49,  1.62it/s]
                                                        
{'loss': 0.198, 'grad_norm': 0.5519906878471375, 'learning_rate': 5.79931968663253e-06, 'epoch': 1.51}

 50%|█████     | 6712/13317 [1:09:11<1:07:49,  1.62it/s]
 50%|█████     | 6713/13317 [1:09:11<1:07:43,  1.63it/s]
                                                        
{'loss': 0.2303, 'grad_norm': 0.5374262928962708, 'learning_rate': 5.798025879956024e-06, 'epoch': 1.51}

 50%|█████     | 6713/13317 [1:09:11<1:07:43,  1.63it/s]
 50%|█████     | 6714/13317 [1:09:12<1:07:44,  1.62it/s]
                                                        
{'loss': 0.2205, 'grad_norm': 0.5160539150238037, 'learning_rate': 5.796732018446691e-06, 'epoch': 1.51}

 50%|█████     | 6714/13317 [1:09:12<1:07:44,  1.62it/s]
 50%|█████     | 6715/13317 [1:09:12<1:07:46,  1.62it/s]
                                                        
{'loss': 0.2282, 'grad_norm': 0.5676764845848083, 'learning_rate': 5.795438102193432e-06, 'epoch': 1.51}

 50%|█████     | 6715/13317 [1:09:13<1:07:46,  1.62it/s]
 50%|█████     | 6716/13317 [1:09:13<1:07:47,  1.62it/s]
                                                        
{'loss': 0.1916, 'grad_norm': 0.592774510383606, 'learning_rate': 5.794144131285152e-06, 'epoch': 1.51}

 50%|█████     | 6716/13317 [1:09:13<1:07:47,  1.62it/s]
 50%|█████     | 6717/13317 [1:09:14<1:07:43,  1.62it/s]
                                                        
{'loss': 0.2171, 'grad_norm': 0.5394418835639954, 'learning_rate': 5.792850105810764e-06, 'epoch': 1.51}

 50%|█████     | 6717/13317 [1:09:14<1:07:43,  1.62it/s]
 50%|█████     | 6718/13317 [1:09:14<1:07:38,  1.63it/s]
                                                        
{'loss': 0.1762, 'grad_norm': 0.46157246828079224, 'learning_rate': 5.791556025859177e-06, 'epoch': 1.51}

 50%|█████     | 6718/13317 [1:09:14<1:07:38,  1.63it/s]
 50%|█████     | 6719/13317 [1:09:15<1:07:33,  1.63it/s]
                                                        
{'loss': 0.1998, 'grad_norm': 0.5205851197242737, 'learning_rate': 5.790261891519311e-06, 'epoch': 1.51}

 50%|█████     | 6719/13317 [1:09:15<1:07:33,  1.63it/s]
 50%|█████     | 6720/13317 [1:09:15<1:07:32,  1.63it/s]
                                                        
{'loss': 0.269, 'grad_norm': 0.6398618817329407, 'learning_rate': 5.788967702880082e-06, 'epoch': 1.51}

 50%|█████     | 6720/13317 [1:09:16<1:07:32,  1.63it/s]
 50%|█████     | 6721/13317 [1:09:16<1:07:33,  1.63it/s]
                                                        
{'loss': 0.2285, 'grad_norm': 0.5887876749038696, 'learning_rate': 5.787673460030423e-06, 'epoch': 1.51}

 50%|█████     | 6721/13317 [1:09:16<1:07:33,  1.63it/s]
 50%|█████     | 6722/13317 [1:09:17<1:07:30,  1.63it/s]
                                                        
{'loss': 0.1941, 'grad_norm': 0.5140997767448425, 'learning_rate': 5.786379163059256e-06, 'epoch': 1.51}

 50%|█████     | 6722/13317 [1:09:17<1:07:30,  1.63it/s]
 50%|█████     | 6723/13317 [1:09:17<1:07:31,  1.63it/s]
                                                        
{'loss': 0.1778, 'grad_norm': 0.48841941356658936, 'learning_rate': 5.785084812055512e-06, 'epoch': 1.51}

 50%|█████     | 6723/13317 [1:09:17<1:07:31,  1.63it/s]
 50%|█████     | 6724/13317 [1:09:18<1:07:29,  1.63it/s]
                                                        
{'loss': 0.2248, 'grad_norm': 0.5798484086990356, 'learning_rate': 5.7837904071081306e-06, 'epoch': 1.51}

 50%|█████     | 6724/13317 [1:09:18<1:07:29,  1.63it/s]
 50%|█████     | 6725/13317 [1:09:19<1:07:32,  1.63it/s]
                                                        
{'loss': 0.2167, 'grad_norm': 0.6559006571769714, 'learning_rate': 5.782495948306049e-06, 'epoch': 1.51}

 50%|█████     | 6725/13317 [1:09:19<1:07:32,  1.63it/s]
 51%|█████     | 6726/13317 [1:09:19<1:07:33,  1.63it/s]
                                                        
{'loss': 0.1691, 'grad_norm': 0.4858795702457428, 'learning_rate': 5.781201435738212e-06, 'epoch': 1.52}

 51%|█████     | 6726/13317 [1:09:19<1:07:33,  1.63it/s]
 51%|█████     | 6727/13317 [1:09:20<1:07:30,  1.63it/s]
                                                        
{'loss': 0.1892, 'grad_norm': 0.5756695866584778, 'learning_rate': 5.7799068694935625e-06, 'epoch': 1.52}

 51%|█████     | 6727/13317 [1:09:20<1:07:30,  1.63it/s]
 51%|█████     | 6728/13317 [1:09:20<1:07:27,  1.63it/s]
                                                        
{'loss': 0.1856, 'grad_norm': 0.5110322833061218, 'learning_rate': 5.778612249661054e-06, 'epoch': 1.52}

 51%|█████     | 6728/13317 [1:09:21<1:07:27,  1.63it/s]
 51%|█████     | 6729/13317 [1:09:21<1:07:28,  1.63it/s]
                                                        
{'loss': 0.1326, 'grad_norm': 0.43071651458740234, 'learning_rate': 5.77731757632964e-06, 'epoch': 1.52}

 51%|█████     | 6729/13317 [1:09:21<1:07:28,  1.63it/s]
 51%|█████     | 6730/13317 [1:09:22<1:07:32,  1.63it/s]
                                                        
{'loss': 0.1888, 'grad_norm': 0.5573046207427979, 'learning_rate': 5.776022849588278e-06, 'epoch': 1.52}

 51%|█████     | 6730/13317 [1:09:22<1:07:32,  1.63it/s]
 51%|█████     | 6731/13317 [1:09:22<1:07:50,  1.62it/s]
                                                        
{'loss': 0.2422, 'grad_norm': 0.596106767654419, 'learning_rate': 5.77472806952593e-06, 'epoch': 1.52}

 51%|█████     | 6731/13317 [1:09:22<1:07:50,  1.62it/s]
 51%|█████     | 6732/13317 [1:09:23<1:07:41,  1.62it/s]
                                                        
{'loss': 0.2393, 'grad_norm': 0.5506792068481445, 'learning_rate': 5.7734332362315605e-06, 'epoch': 1.52}

 51%|█████     | 6732/13317 [1:09:23<1:07:41,  1.62it/s]
 51%|█████     | 6733/13317 [1:09:23<1:07:32,  1.62it/s]
                                                        
{'loss': 0.1815, 'grad_norm': 0.5258397459983826, 'learning_rate': 5.772138349794139e-06, 'epoch': 1.52}

 51%|█████     | 6733/13317 [1:09:24<1:07:32,  1.62it/s]
 51%|█████     | 6734/13317 [1:09:24<1:07:27,  1.63it/s]
                                                        
{'loss': 0.1943, 'grad_norm': 0.5304751396179199, 'learning_rate': 5.770843410302636e-06, 'epoch': 1.52}

 51%|█████     | 6734/13317 [1:09:24<1:07:27,  1.63it/s]
 51%|█████     | 6735/13317 [1:09:25<1:07:44,  1.62it/s]
                                                        
{'loss': 0.2607, 'grad_norm': 0.6047868728637695, 'learning_rate': 5.76954841784603e-06, 'epoch': 1.52}

 51%|█████     | 6735/13317 [1:09:25<1:07:44,  1.62it/s]
 51%|█████     | 6736/13317 [1:09:25<1:07:42,  1.62it/s]
                                                        
{'loss': 0.2214, 'grad_norm': 0.5527637004852295, 'learning_rate': 5.7682533725133e-06, 'epoch': 1.52}

 51%|█████     | 6736/13317 [1:09:25<1:07:42,  1.62it/s]
 51%|█████     | 6737/13317 [1:09:26<1:07:33,  1.62it/s]
                                                        
{'loss': 0.1826, 'grad_norm': 0.48912784457206726, 'learning_rate': 5.766958274393428e-06, 'epoch': 1.52}

 51%|█████     | 6737/13317 [1:09:26<1:07:33,  1.62it/s]
 51%|█████     | 6738/13317 [1:09:27<1:07:28,  1.63it/s]
                                                        
{'loss': 0.2271, 'grad_norm': 0.5886701345443726, 'learning_rate': 5.765663123575401e-06, 'epoch': 1.52}

 51%|█████     | 6738/13317 [1:09:27<1:07:28,  1.63it/s]
 51%|█████     | 6739/13317 [1:09:27<1:07:26,  1.63it/s]
                                                        
{'loss': 0.1661, 'grad_norm': 0.4174511730670929, 'learning_rate': 5.764367920148212e-06, 'epoch': 1.52}

 51%|█████     | 6739/13317 [1:09:27<1:07:26,  1.63it/s]
 51%|█████     | 6740/13317 [1:09:28<1:07:26,  1.63it/s]
                                                        
{'loss': 0.2577, 'grad_norm': 0.6051071882247925, 'learning_rate': 5.7630726642008526e-06, 'epoch': 1.52}

 51%|█████     | 6740/13317 [1:09:28<1:07:26,  1.63it/s]
 51%|█████     | 6741/13317 [1:09:28<1:07:28,  1.62it/s]
                                                        
{'loss': 0.1761, 'grad_norm': 0.45285236835479736, 'learning_rate': 5.761777355822322e-06, 'epoch': 1.52}

 51%|█████     | 6741/13317 [1:09:29<1:07:28,  1.62it/s]
 51%|█████     | 6742/13317 [1:09:29<1:07:22,  1.63it/s]
                                                        
{'loss': 0.2151, 'grad_norm': 0.6171121001243591, 'learning_rate': 5.760481995101622e-06, 'epoch': 1.52}

 51%|█████     | 6742/13317 [1:09:29<1:07:22,  1.63it/s]
 51%|█████     | 6743/13317 [1:09:30<1:07:19,  1.63it/s]
                                                        
{'loss': 0.2138, 'grad_norm': 0.4778108596801758, 'learning_rate': 5.759186582127756e-06, 'epoch': 1.52}

 51%|█████     | 6743/13317 [1:09:30<1:07:19,  1.63it/s]
 51%|█████     | 6744/13317 [1:09:30<1:07:15,  1.63it/s]
                                                        
{'loss': 0.2183, 'grad_norm': 0.51649010181427, 'learning_rate': 5.757891116989733e-06, 'epoch': 1.52}

 51%|█████     | 6744/13317 [1:09:30<1:07:15,  1.63it/s]
 51%|█████     | 6745/13317 [1:09:31<1:07:18,  1.63it/s]
                                                        
{'loss': 0.246, 'grad_norm': 0.6796680688858032, 'learning_rate': 5.7565955997765665e-06, 'epoch': 1.52}

 51%|█████     | 6745/13317 [1:09:31<1:07:18,  1.63it/s]
 51%|█████     | 6746/13317 [1:09:31<1:07:21,  1.63it/s]
                                                        
{'loss': 0.217, 'grad_norm': 0.5161526799201965, 'learning_rate': 5.755300030577271e-06, 'epoch': 1.52}

 51%|█████     | 6746/13317 [1:09:32<1:07:21,  1.63it/s]
 51%|█████     | 6747/13317 [1:09:32<1:07:19,  1.63it/s]
                                                        
{'loss': 0.1313, 'grad_norm': 0.4209243059158325, 'learning_rate': 5.754004409480865e-06, 'epoch': 1.52}

 51%|█████     | 6747/13317 [1:09:32<1:07:19,  1.63it/s]
 51%|█████     | 6748/13317 [1:09:33<1:07:14,  1.63it/s]
                                                        
{'loss': 0.186, 'grad_norm': 0.49969586730003357, 'learning_rate': 5.752708736576376e-06, 'epoch': 1.52}

 51%|█████     | 6748/13317 [1:09:33<1:07:14,  1.63it/s]
 51%|█████     | 6749/13317 [1:09:33<1:07:14,  1.63it/s]
                                                        
{'loss': 0.1957, 'grad_norm': 0.5377335548400879, 'learning_rate': 5.751413011952824e-06, 'epoch': 1.52}

 51%|█████     | 6749/13317 [1:09:33<1:07:14,  1.63it/s]
 51%|█████     | 6750/13317 [1:09:34<1:07:16,  1.63it/s]
                                                        
{'loss': 0.2137, 'grad_norm': 0.5823622941970825, 'learning_rate': 5.750117235699243e-06, 'epoch': 1.52}

 51%|█████     | 6750/13317 [1:09:34<1:07:16,  1.63it/s]
 51%|█████     | 6751/13317 [1:09:35<1:07:18,  1.63it/s]
                                                        
{'loss': 0.2792, 'grad_norm': 0.7317012548446655, 'learning_rate': 5.748821407904664e-06, 'epoch': 1.52}

 51%|█████     | 6751/13317 [1:09:35<1:07:18,  1.63it/s]
 51%|█████     | 6752/13317 [1:09:35<1:07:15,  1.63it/s]
                                                        
{'loss': 0.167, 'grad_norm': 0.49817490577697754, 'learning_rate': 5.7475255286581275e-06, 'epoch': 1.52}

 51%|█████     | 6752/13317 [1:09:35<1:07:15,  1.63it/s]
 51%|█████     | 6753/13317 [1:09:36<1:07:11,  1.63it/s]
                                                        
{'loss': 0.1978, 'grad_norm': 0.560279905796051, 'learning_rate': 5.746229598048672e-06, 'epoch': 1.52}

 51%|█████     | 6753/13317 [1:09:36<1:07:11,  1.63it/s]
 51%|█████     | 6754/13317 [1:09:36<1:07:08,  1.63it/s]
                                                        
{'loss': 0.2281, 'grad_norm': 0.5609884262084961, 'learning_rate': 5.74493361616534e-06, 'epoch': 1.52}

 51%|█████     | 6754/13317 [1:09:36<1:07:08,  1.63it/s]
 51%|█████     | 6755/13317 [1:09:37<1:07:12,  1.63it/s]
                                                        
{'loss': 0.1693, 'grad_norm': 0.4373992085456848, 'learning_rate': 5.7436375830971835e-06, 'epoch': 1.52}

 51%|█████     | 6755/13317 [1:09:37<1:07:12,  1.63it/s]
 51%|█████     | 6756/13317 [1:09:38<1:07:14,  1.63it/s]
                                                        
{'loss': 0.1952, 'grad_norm': 0.5226536989212036, 'learning_rate': 5.7423414989332495e-06, 'epoch': 1.52}

 51%|█████     | 6756/13317 [1:09:38<1:07:14,  1.63it/s]
 51%|█████     | 6757/13317 [1:09:38<1:07:13,  1.63it/s]
                                                        
{'loss': 0.2267, 'grad_norm': 0.5200282335281372, 'learning_rate': 5.7410453637625966e-06, 'epoch': 1.52}

 51%|█████     | 6757/13317 [1:09:38<1:07:13,  1.63it/s]
 51%|█████     | 6758/13317 [1:09:39<1:07:12,  1.63it/s]
                                                        
{'loss': 0.1922, 'grad_norm': 0.5213318467140198, 'learning_rate': 5.73974917767428e-06, 'epoch': 1.52}

 51%|█████     | 6758/13317 [1:09:39<1:07:12,  1.63it/s]
 51%|█████     | 6759/13317 [1:09:39<1:07:10,  1.63it/s]
                                                        
{'loss': 0.1965, 'grad_norm': 0.5448389649391174, 'learning_rate': 5.738452940757361e-06, 'epoch': 1.52}

 51%|█████     | 6759/13317 [1:09:40<1:07:10,  1.63it/s]
 51%|█████     | 6760/13317 [1:09:40<1:07:14,  1.63it/s]
                                                        
{'loss': 0.1646, 'grad_norm': 0.4422101378440857, 'learning_rate': 5.737156653100908e-06, 'epoch': 1.52}

 51%|█████     | 6760/13317 [1:09:40<1:07:14,  1.63it/s]
 51%|█████     | 6761/13317 [1:09:41<1:07:14,  1.62it/s]
                                                        
{'loss': 0.2311, 'grad_norm': 0.5113460421562195, 'learning_rate': 5.735860314793986e-06, 'epoch': 1.52}

 51%|█████     | 6761/13317 [1:09:41<1:07:14,  1.62it/s]
 51%|█████     | 6762/13317 [1:09:41<1:07:11,  1.63it/s]
                                                        
{'loss': 0.2353, 'grad_norm': 0.5481608510017395, 'learning_rate': 5.73456392592567e-06, 'epoch': 1.52}

 51%|█████     | 6762/13317 [1:09:41<1:07:11,  1.63it/s]
 51%|█████     | 6763/13317 [1:09:42<1:07:07,  1.63it/s]
                                                        
{'loss': 0.297, 'grad_norm': 0.7782899141311646, 'learning_rate': 5.733267486585034e-06, 'epoch': 1.52}

 51%|█████     | 6763/13317 [1:09:42<1:07:07,  1.63it/s]
 51%|█████     | 6764/13317 [1:09:43<1:07:03,  1.63it/s]
                                                        
{'loss': 0.2703, 'grad_norm': 0.5923895835876465, 'learning_rate': 5.731970996861158e-06, 'epoch': 1.52}

 51%|█████     | 6764/13317 [1:09:43<1:07:03,  1.63it/s]
 51%|█████     | 6765/13317 [1:09:43<1:07:05,  1.63it/s]
                                                        
{'loss': 0.1645, 'grad_norm': 0.5084775686264038, 'learning_rate': 5.730674456843125e-06, 'epoch': 1.52}

 51%|█████     | 6765/13317 [1:09:43<1:07:05,  1.63it/s]
 51%|█████     | 6766/13317 [1:09:44<1:07:11,  1.63it/s]
                                                        
{'loss': 0.213, 'grad_norm': 0.606769323348999, 'learning_rate': 5.729377866620019e-06, 'epoch': 1.52}

 51%|█████     | 6766/13317 [1:09:44<1:07:11,  1.63it/s]
 51%|█████     | 6767/13317 [1:09:44<1:07:06,  1.63it/s]
                                                        
{'loss': 0.2128, 'grad_norm': 0.5424793362617493, 'learning_rate': 5.728081226280932e-06, 'epoch': 1.52}

 51%|█████     | 6767/13317 [1:09:44<1:07:06,  1.63it/s]
 51%|█████     | 6768/13317 [1:09:45<1:07:02,  1.63it/s]
                                                        
{'loss': 0.1736, 'grad_norm': 0.48421016335487366, 'learning_rate': 5.726784535914955e-06, 'epoch': 1.52}

 51%|█████     | 6768/13317 [1:09:45<1:07:02,  1.63it/s]
 51%|█████     | 6769/13317 [1:09:46<1:06:59,  1.63it/s]
                                                        
{'loss': 0.177, 'grad_norm': 0.45434075593948364, 'learning_rate': 5.725487795611186e-06, 'epoch': 1.52}

 51%|█████     | 6769/13317 [1:09:46<1:06:59,  1.63it/s]
 51%|█████     | 6770/13317 [1:09:46<1:07:00,  1.63it/s]
                                                        
{'loss': 0.2167, 'grad_norm': 0.5257607102394104, 'learning_rate': 5.724191005458724e-06, 'epoch': 1.53}

 51%|█████     | 6770/13317 [1:09:46<1:07:00,  1.63it/s]
 51%|█████     | 6771/13317 [1:09:47<1:07:03,  1.63it/s]
                                                        
{'loss': 0.2194, 'grad_norm': 0.662413477897644, 'learning_rate': 5.72289416554667e-06, 'epoch': 1.53}

 51%|█████     | 6771/13317 [1:09:47<1:07:03,  1.63it/s]
 51%|█████     | 6772/13317 [1:09:47<1:07:01,  1.63it/s]
                                                        
{'loss': 0.1971, 'grad_norm': 0.5366441607475281, 'learning_rate': 5.7215972759641335e-06, 'epoch': 1.53}

 51%|█████     | 6772/13317 [1:09:48<1:07:01,  1.63it/s]
 51%|█████     | 6773/13317 [1:09:48<1:06:58,  1.63it/s]
                                                        
{'loss': 0.2142, 'grad_norm': 0.5490864515304565, 'learning_rate': 5.720300336800224e-06, 'epoch': 1.53}

 51%|█████     | 6773/13317 [1:09:48<1:06:58,  1.63it/s]
 51%|█████     | 6774/13317 [1:09:49<1:06:55,  1.63it/s]
                                                        
{'loss': 0.1891, 'grad_norm': 0.5254865884780884, 'learning_rate': 5.7190033481440545e-06, 'epoch': 1.53}

 51%|█████     | 6774/13317 [1:09:49<1:06:55,  1.63it/s]
 51%|█████     | 6775/13317 [1:09:49<1:06:56,  1.63it/s]
                                                        
{'loss': 0.1867, 'grad_norm': 0.4952949583530426, 'learning_rate': 5.717706310084744e-06, 'epoch': 1.53}

 51%|█████     | 6775/13317 [1:09:49<1:06:56,  1.63it/s]
 51%|█████     | 6776/13317 [1:09:50<1:07:00,  1.63it/s]
                                                        
{'loss': 0.2062, 'grad_norm': 0.5362591743469238, 'learning_rate': 5.716409222711408e-06, 'epoch': 1.53}

 51%|█████     | 6776/13317 [1:09:50<1:07:00,  1.63it/s]
 51%|█████     | 6777/13317 [1:09:51<1:06:58,  1.63it/s]
                                                        
{'loss': 0.2276, 'grad_norm': 0.6071352362632751, 'learning_rate': 5.715112086113175e-06, 'epoch': 1.53}

 51%|█████     | 6777/13317 [1:09:51<1:06:58,  1.63it/s]
 51%|█████     | 6778/13317 [1:09:51<1:06:55,  1.63it/s]
                                                        
{'loss': 0.2631, 'grad_norm': 0.5844934582710266, 'learning_rate': 5.713814900379169e-06, 'epoch': 1.53}

 51%|█████     | 6778/13317 [1:09:51<1:06:55,  1.63it/s]
 51%|█████     | 6779/13317 [1:09:52<1:06:52,  1.63it/s]
                                                        
{'loss': 0.1704, 'grad_norm': 0.5094215273857117, 'learning_rate': 5.71251766559852e-06, 'epoch': 1.53}

 51%|█████     | 6779/13317 [1:09:52<1:06:52,  1.63it/s]
 51%|█████     | 6780/13317 [1:09:52<1:06:52,  1.63it/s]
                                                        
{'loss': 0.2124, 'grad_norm': 0.5584704279899597, 'learning_rate': 5.711220381860366e-06, 'epoch': 1.53}

 51%|█████     | 6780/13317 [1:09:52<1:06:52,  1.63it/s]
 51%|█████     | 6781/13317 [1:09:53<1:06:58,  1.63it/s]
                                                        
{'loss': 0.2182, 'grad_norm': 0.5637684464454651, 'learning_rate': 5.709923049253839e-06, 'epoch': 1.53}

 51%|█████     | 6781/13317 [1:09:53<1:06:58,  1.63it/s]
 51%|█████     | 6782/13317 [1:09:54<1:06:56,  1.63it/s]
                                                        
{'loss': 0.2034, 'grad_norm': 0.488366037607193, 'learning_rate': 5.708625667868083e-06, 'epoch': 1.53}

 51%|█████     | 6782/13317 [1:09:54<1:06:56,  1.63it/s]
 51%|█████     | 6783/13317 [1:09:54<1:06:53,  1.63it/s]
                                                        
{'loss': 0.1775, 'grad_norm': 0.4734778106212616, 'learning_rate': 5.707328237792239e-06, 'epoch': 1.53}

 51%|█████     | 6783/13317 [1:09:54<1:06:53,  1.63it/s]
 51%|█████     | 6784/13317 [1:09:55<1:06:51,  1.63it/s]
                                                        
{'loss': 0.1897, 'grad_norm': 0.5025120377540588, 'learning_rate': 5.706030759115458e-06, 'epoch': 1.53}

 51%|█████     | 6784/13317 [1:09:55<1:06:51,  1.63it/s]
 51%|█████     | 6785/13317 [1:09:55<1:06:48,  1.63it/s]
                                                        
{'loss': 0.2097, 'grad_norm': 0.5105117559432983, 'learning_rate': 5.704733231926887e-06, 'epoch': 1.53}

 51%|█████     | 6785/13317 [1:09:56<1:06:48,  1.63it/s]
 51%|█████     | 6786/13317 [1:09:56<1:06:55,  1.63it/s]
                                                        
{'loss': 0.2468, 'grad_norm': 0.5609579682350159, 'learning_rate': 5.703435656315681e-06, 'epoch': 1.53}

 51%|█████     | 6786/13317 [1:09:56<1:06:55,  1.63it/s]
 51%|█████     | 6787/13317 [1:09:57<1:06:52,  1.63it/s]
                                                        
{'loss': 0.1775, 'grad_norm': 0.49356335401535034, 'learning_rate': 5.702138032370999e-06, 'epoch': 1.53}

 51%|█████     | 6787/13317 [1:09:57<1:06:52,  1.63it/s]
 51%|█████     | 6788/13317 [1:09:57<1:06:51,  1.63it/s]
                                                        
{'loss': 0.1701, 'grad_norm': 0.5023838877677917, 'learning_rate': 5.7008403601819965e-06, 'epoch': 1.53}

 51%|█████     | 6788/13317 [1:09:57<1:06:51,  1.63it/s]
 51%|█████     | 6789/13317 [1:09:58<1:06:47,  1.63it/s]
                                                        
{'loss': 0.1827, 'grad_norm': 0.4569028317928314, 'learning_rate': 5.699542639837844e-06, 'epoch': 1.53}

 51%|█████     | 6789/13317 [1:09:58<1:06:47,  1.63it/s]
 51%|█████     | 6790/13317 [1:09:58<1:06:46,  1.63it/s]
                                                        
{'loss': 0.2089, 'grad_norm': 0.5477475523948669, 'learning_rate': 5.698244871427701e-06, 'epoch': 1.53}

 51%|█████     | 6790/13317 [1:09:59<1:06:46,  1.63it/s]
 51%|█████     | 6791/13317 [1:09:59<1:06:52,  1.63it/s]
                                                        
{'loss': 0.2232, 'grad_norm': 0.5793611407279968, 'learning_rate': 5.696947055040747e-06, 'epoch': 1.53}

 51%|█████     | 6791/13317 [1:09:59<1:06:52,  1.63it/s]
 51%|█████     | 6792/13317 [1:10:00<1:06:49,  1.63it/s]
                                                        
{'loss': 0.1983, 'grad_norm': 0.5046886205673218, 'learning_rate': 5.695649190766148e-06, 'epoch': 1.53}

 51%|█████     | 6792/13317 [1:10:00<1:06:49,  1.63it/s]
 51%|█████     | 6793/13317 [1:10:00<1:06:46,  1.63it/s]
                                                        
{'loss': 0.2089, 'grad_norm': 0.51475989818573, 'learning_rate': 5.694351278693085e-06, 'epoch': 1.53}

 51%|█████     | 6793/13317 [1:10:00<1:06:46,  1.63it/s]
 51%|█████     | 6794/13317 [1:10:01<1:06:44,  1.63it/s]
                                                        
{'loss': 0.1678, 'grad_norm': 0.4634905159473419, 'learning_rate': 5.6930533189107375e-06, 'epoch': 1.53}

 51%|█████     | 6794/13317 [1:10:01<1:06:44,  1.63it/s]
 51%|█████     | 6795/13317 [1:10:02<1:06:43,  1.63it/s]
                                                        
{'loss': 0.1958, 'grad_norm': 0.5257046222686768, 'learning_rate': 5.691755311508288e-06, 'epoch': 1.53}

 51%|█████     | 6795/13317 [1:10:02<1:06:43,  1.63it/s]
 51%|█████     | 6796/13317 [1:10:02<1:06:47,  1.63it/s]
                                                        
{'loss': 0.2031, 'grad_norm': 0.4766254723072052, 'learning_rate': 5.690457256574926e-06, 'epoch': 1.53}

 51%|█████     | 6796/13317 [1:10:02<1:06:47,  1.63it/s]
 51%|█████     | 6797/13317 [1:10:03<1:06:44,  1.63it/s]
                                                        
{'loss': 0.2334, 'grad_norm': 0.5211377143859863, 'learning_rate': 5.689159154199838e-06, 'epoch': 1.53}

 51%|█████     | 6797/13317 [1:10:03<1:06:44,  1.63it/s]
 51%|█████     | 6798/13317 [1:10:03<1:06:43,  1.63it/s]
                                                        
{'loss': 0.1775, 'grad_norm': 1.131569743156433, 'learning_rate': 5.687861004472219e-06, 'epoch': 1.53}

 51%|█████     | 6798/13317 [1:10:04<1:06:43,  1.63it/s]
 51%|█████     | 6799/13317 [1:10:04<1:06:41,  1.63it/s]
                                                        
{'loss': 0.199, 'grad_norm': 0.5458588004112244, 'learning_rate': 5.6865628074812665e-06, 'epoch': 1.53}

 51%|█████     | 6799/13317 [1:10:04<1:06:41,  1.63it/s]
 51%|█████     | 6800/13317 [1:10:05<1:06:40,  1.63it/s]
                                                        
{'loss': 0.19, 'grad_norm': 0.535820722579956, 'learning_rate': 5.6852645633161795e-06, 'epoch': 1.53}

 51%|█████     | 6800/13317 [1:10:05<1:06:40,  1.63it/s]
 51%|█████     | 6801/13317 [1:10:05<1:06:57,  1.62it/s]
                                                        
{'loss': 0.1983, 'grad_norm': 0.5094990730285645, 'learning_rate': 5.6839662720661615e-06, 'epoch': 1.53}

 51%|█████     | 6801/13317 [1:10:05<1:06:57,  1.62it/s]
 51%|█████     | 6802/13317 [1:10:06<1:06:53,  1.62it/s]
                                                        
{'loss': 0.1714, 'grad_norm': 0.4518454372882843, 'learning_rate': 5.6826679338204195e-06, 'epoch': 1.53}

 51%|█████     | 6802/13317 [1:10:06<1:06:53,  1.62it/s]
 51%|█████     | 6803/13317 [1:10:06<1:06:49,  1.62it/s]
                                                        
{'loss': 0.1652, 'grad_norm': 0.4743281900882721, 'learning_rate': 5.6813695486681595e-06, 'epoch': 1.53}

 51%|█████     | 6803/13317 [1:10:07<1:06:49,  1.62it/s]
 51%|█████     | 6804/13317 [1:10:07<1:06:45,  1.63it/s]
                                                        
{'loss': 0.1678, 'grad_norm': 0.4271320402622223, 'learning_rate': 5.6800711166986e-06, 'epoch': 1.53}

 51%|█████     | 6804/13317 [1:10:07<1:06:45,  1.63it/s]
 51%|█████     | 6805/13317 [1:10:08<1:06:41,  1.63it/s]
                                                        
{'loss': 0.1906, 'grad_norm': 0.5004856586456299, 'learning_rate': 5.678772638000952e-06, 'epoch': 1.53}

 51%|█████     | 6805/13317 [1:10:08<1:06:41,  1.63it/s]
 51%|█████     | 6806/13317 [1:10:08<1:06:43,  1.63it/s]
                                                        
{'loss': 0.135, 'grad_norm': 0.44469425082206726, 'learning_rate': 5.6774741126644384e-06, 'epoch': 1.53}

 51%|█████     | 6806/13317 [1:10:08<1:06:43,  1.63it/s]
 51%|█████     | 6807/13317 [1:10:09<1:06:41,  1.63it/s]
                                                        
{'loss': 0.1322, 'grad_norm': 0.39630424976348877, 'learning_rate': 5.676175540778279e-06, 'epoch': 1.53}

 51%|█████     | 6807/13317 [1:10:09<1:06:41,  1.63it/s]
 51%|█████     | 6808/13317 [1:10:10<1:06:36,  1.63it/s]
                                                        
{'loss': 0.2294, 'grad_norm': 0.5805449485778809, 'learning_rate': 5.6748769224317e-06, 'epoch': 1.53}

 51%|█████     | 6808/13317 [1:10:10<1:06:36,  1.63it/s]
 51%|█████     | 6809/13317 [1:10:10<1:06:35,  1.63it/s]
                                                        
{'loss': 0.1379, 'grad_norm': 0.38922789692878723, 'learning_rate': 5.673578257713933e-06, 'epoch': 1.53}

 51%|█████     | 6809/13317 [1:10:10<1:06:35,  1.63it/s]
 51%|█████     | 6810/13317 [1:10:11<1:06:34,  1.63it/s]
                                                        
{'loss': 0.1923, 'grad_norm': 0.49748554825782776, 'learning_rate': 5.6722795467142055e-06, 'epoch': 1.53}

 51%|█████     | 6810/13317 [1:10:11<1:06:34,  1.63it/s]
 51%|█████     | 6811/13317 [1:10:11<1:06:36,  1.63it/s]
                                                        
{'loss': 0.2388, 'grad_norm': 0.6452785730361938, 'learning_rate': 5.670980789521756e-06, 'epoch': 1.53}

 51%|█████     | 6811/13317 [1:10:12<1:06:36,  1.63it/s]
 51%|█████     | 6812/13317 [1:10:12<1:06:35,  1.63it/s]
                                                        
{'loss': 0.1838, 'grad_norm': 0.5047988295555115, 'learning_rate': 5.669681986225821e-06, 'epoch': 1.53}

 51%|█████     | 6812/13317 [1:10:12<1:06:35,  1.63it/s]
 51%|█████     | 6813/13317 [1:10:13<1:06:33,  1.63it/s]
                                                        
{'loss': 0.2284, 'grad_norm': 0.5765100121498108, 'learning_rate': 5.668383136915644e-06, 'epoch': 1.53}

 51%|█████     | 6813/13317 [1:10:13<1:06:33,  1.63it/s]
 51%|█████     | 6814/13317 [1:10:13<1:06:33,  1.63it/s]
                                                        
{'loss': 0.1967, 'grad_norm': 0.5365408658981323, 'learning_rate': 5.6670842416804676e-06, 'epoch': 1.54}

 51%|█████     | 6814/13317 [1:10:13<1:06:33,  1.63it/s]
 51%|█████     | 6815/13317 [1:10:14<1:06:31,  1.63it/s]
                                                        
{'loss': 0.181, 'grad_norm': 0.4827931821346283, 'learning_rate': 5.665785300609541e-06, 'epoch': 1.54}

 51%|█████     | 6815/13317 [1:10:14<1:06:31,  1.63it/s]
 51%|█████     | 6816/13317 [1:10:14<1:06:35,  1.63it/s]
                                                        
{'loss': 0.1693, 'grad_norm': 0.48592302203178406, 'learning_rate': 5.664486313792115e-06, 'epoch': 1.54}

 51%|█████     | 6816/13317 [1:10:15<1:06:35,  1.63it/s]
 51%|█████     | 6817/13317 [1:10:15<1:06:33,  1.63it/s]
                                                        
{'loss': 0.2219, 'grad_norm': 0.49979478120803833, 'learning_rate': 5.663187281317442e-06, 'epoch': 1.54}

 51%|█████     | 6817/13317 [1:10:15<1:06:33,  1.63it/s]
 51%|█████     | 6818/13317 [1:10:16<1:06:29,  1.63it/s]
                                                        
{'loss': 0.1799, 'grad_norm': 0.5272715091705322, 'learning_rate': 5.661888203274784e-06, 'epoch': 1.54}

 51%|█████     | 6818/13317 [1:10:16<1:06:29,  1.63it/s]
 51%|█████     | 6819/13317 [1:10:16<1:06:26,  1.63it/s]
                                                        
{'loss': 0.2264, 'grad_norm': 0.5689626336097717, 'learning_rate': 5.6605890797533955e-06, 'epoch': 1.54}

 51%|█████     | 6819/13317 [1:10:16<1:06:26,  1.63it/s]
 51%|█████     | 6820/13317 [1:10:17<1:06:25,  1.63it/s]
                                                        
{'loss': 0.2121, 'grad_norm': 0.5376394987106323, 'learning_rate': 5.659289910842544e-06, 'epoch': 1.54}

 51%|█████     | 6820/13317 [1:10:17<1:06:25,  1.63it/s]
 51%|█████     | 6821/13317 [1:10:18<1:06:28,  1.63it/s]
                                                        
{'loss': 0.1941, 'grad_norm': 0.5027828216552734, 'learning_rate': 5.6579906966314936e-06, 'epoch': 1.54}

 51%|█████     | 6821/13317 [1:10:18<1:06:28,  1.63it/s]
 51%|█████     | 6822/13317 [1:10:18<1:06:26,  1.63it/s]
                                                        
{'loss': 0.1803, 'grad_norm': 0.4669327139854431, 'learning_rate': 5.656691437209516e-06, 'epoch': 1.54}

 51%|█████     | 6822/13317 [1:10:18<1:06:26,  1.63it/s]
 51%|█████     | 6823/13317 [1:10:19<1:06:24,  1.63it/s]
                                                        
{'loss': 0.2003, 'grad_norm': 0.5157067775726318, 'learning_rate': 5.655392132665885e-06, 'epoch': 1.54}

 51%|█████     | 6823/13317 [1:10:19<1:06:24,  1.63it/s]
 51%|█████     | 6824/13317 [1:10:19<1:06:23,  1.63it/s]
                                                        
{'loss': 0.197, 'grad_norm': 0.4923931658267975, 'learning_rate': 5.654092783089874e-06, 'epoch': 1.54}

 51%|█████     | 6824/13317 [1:10:20<1:06:23,  1.63it/s]
 51%|█████▏    | 6825/13317 [1:10:20<1:06:22,  1.63it/s]
                                                        
{'loss': 0.2473, 'grad_norm': 0.5621438026428223, 'learning_rate': 5.6527933885707635e-06, 'epoch': 1.54}

 51%|█████▏    | 6825/13317 [1:10:20<1:06:22,  1.63it/s]
 51%|█████▏    | 6826/13317 [1:10:21<1:06:25,  1.63it/s]
                                                        
{'loss': 0.1826, 'grad_norm': 0.49002817273139954, 'learning_rate': 5.651493949197834e-06, 'epoch': 1.54}

 51%|█████▏    | 6826/13317 [1:10:21<1:06:25,  1.63it/s]
 51%|█████▏    | 6827/13317 [1:10:21<1:06:23,  1.63it/s]
                                                        
{'loss': 0.1828, 'grad_norm': 0.4822761118412018, 'learning_rate': 5.6501944650603725e-06, 'epoch': 1.54}

 51%|█████▏    | 6827/13317 [1:10:21<1:06:23,  1.63it/s]
 51%|█████▏    | 6828/13317 [1:10:22<1:06:23,  1.63it/s]
                                                        
{'loss': 0.2034, 'grad_norm': 0.5216538310050964, 'learning_rate': 5.648894936247668e-06, 'epoch': 1.54}

 51%|█████▏    | 6828/13317 [1:10:22<1:06:23,  1.63it/s]
 51%|█████▏    | 6829/13317 [1:10:22<1:06:23,  1.63it/s]
                                                        
{'loss': 0.2018, 'grad_norm': 0.5114476680755615, 'learning_rate': 5.64759536284901e-06, 'epoch': 1.54}

 51%|█████▏    | 6829/13317 [1:10:23<1:06:23,  1.63it/s]
 51%|█████▏    | 6830/13317 [1:10:23<1:06:21,  1.63it/s]
                                                        
{'loss': 0.1505, 'grad_norm': 0.503437340259552, 'learning_rate': 5.646295744953693e-06, 'epoch': 1.54}

 51%|█████▏    | 6830/13317 [1:10:23<1:06:21,  1.63it/s]
 51%|█████▏    | 6831/13317 [1:10:24<1:06:25,  1.63it/s]
                                                        
{'loss': 0.2395, 'grad_norm': 0.4735241234302521, 'learning_rate': 5.644996082651018e-06, 'epoch': 1.54}

 51%|█████▏    | 6831/13317 [1:10:24<1:06:25,  1.63it/s]
 51%|█████▏    | 6832/13317 [1:10:24<1:06:23,  1.63it/s]
                                                        
{'loss': 0.197, 'grad_norm': 0.497791588306427, 'learning_rate': 5.64369637603028e-06, 'epoch': 1.54}

 51%|█████▏    | 6832/13317 [1:10:24<1:06:23,  1.63it/s]
 51%|█████▏    | 6833/13317 [1:10:25<1:06:21,  1.63it/s]
                                                        
{'loss': 0.1676, 'grad_norm': 0.4403837323188782, 'learning_rate': 5.642396625180786e-06, 'epoch': 1.54}

 51%|█████▏    | 6833/13317 [1:10:25<1:06:21,  1.63it/s]
 51%|█████▏    | 6834/13317 [1:10:26<1:06:19,  1.63it/s]
                                                        
{'loss': 0.1916, 'grad_norm': 0.5238213539123535, 'learning_rate': 5.641096830191843e-06, 'epoch': 1.54}

 51%|█████▏    | 6834/13317 [1:10:26<1:06:19,  1.63it/s]
 51%|█████▏    | 6835/13317 [1:10:26<1:06:18,  1.63it/s]
                                                        
{'loss': 0.1928, 'grad_norm': 0.5014521479606628, 'learning_rate': 5.6397969911527586e-06, 'epoch': 1.54}

 51%|█████▏    | 6835/13317 [1:10:26<1:06:18,  1.63it/s]
 51%|█████▏    | 6836/13317 [1:10:27<1:06:21,  1.63it/s]
                                                        
{'loss': 0.196, 'grad_norm': 0.5812679529190063, 'learning_rate': 5.638497108152848e-06, 'epoch': 1.54}

 51%|█████▏    | 6836/13317 [1:10:27<1:06:21,  1.63it/s]
 51%|█████▏    | 6837/13317 [1:10:27<1:06:22,  1.63it/s]
                                                        
{'loss': 0.2018, 'grad_norm': 0.5215535163879395, 'learning_rate': 5.637197181281423e-06, 'epoch': 1.54}

 51%|█████▏    | 6837/13317 [1:10:27<1:06:22,  1.63it/s]
 51%|█████▏    | 6838/13317 [1:10:28<1:06:21,  1.63it/s]
                                                        
{'loss': 0.1771, 'grad_norm': 0.5529235005378723, 'learning_rate': 5.635897210627808e-06, 'epoch': 1.54}

 51%|█████▏    | 6838/13317 [1:10:28<1:06:21,  1.63it/s]
 51%|█████▏    | 6839/13317 [1:10:29<1:06:17,  1.63it/s]
                                                        
{'loss': 0.2263, 'grad_norm': 0.5270705223083496, 'learning_rate': 5.63459719628132e-06, 'epoch': 1.54}

 51%|█████▏    | 6839/13317 [1:10:29<1:06:17,  1.63it/s]
 51%|█████▏    | 6840/13317 [1:10:29<1:06:16,  1.63it/s]
                                                        
{'loss': 0.2822, 'grad_norm': 0.5296881794929504, 'learning_rate': 5.633297138331285e-06, 'epoch': 1.54}

 51%|█████▏    | 6840/13317 [1:10:29<1:06:16,  1.63it/s]
 51%|█████▏    | 6841/13317 [1:10:30<1:06:32,  1.62it/s]
                                                        
{'loss': 0.2278, 'grad_norm': 0.5209159851074219, 'learning_rate': 5.631997036867032e-06, 'epoch': 1.54}

 51%|█████▏    | 6841/13317 [1:10:30<1:06:32,  1.62it/s]
 51%|█████▏    | 6842/13317 [1:10:30<1:06:25,  1.62it/s]
                                                        
{'loss': 0.1448, 'grad_norm': 0.4253266155719757, 'learning_rate': 5.63069689197789e-06, 'epoch': 1.54}

 51%|█████▏    | 6842/13317 [1:10:31<1:06:25,  1.62it/s]
 51%|█████▏    | 6843/13317 [1:10:31<1:06:21,  1.63it/s]
                                                        
{'loss': 0.1837, 'grad_norm': 0.48440366983413696, 'learning_rate': 5.629396703753196e-06, 'epoch': 1.54}

 51%|█████▏    | 6843/13317 [1:10:31<1:06:21,  1.63it/s]
 51%|█████▏    | 6844/13317 [1:10:32<1:06:21,  1.63it/s]
                                                        
{'loss': 0.1973, 'grad_norm': 0.5112410187721252, 'learning_rate': 5.628096472282281e-06, 'epoch': 1.54}

 51%|█████▏    | 6844/13317 [1:10:32<1:06:21,  1.63it/s]
 51%|█████▏    | 6845/13317 [1:10:32<1:06:22,  1.63it/s]
                                                        
{'loss': 0.2165, 'grad_norm': 0.5731823444366455, 'learning_rate': 5.62679619765449e-06, 'epoch': 1.54}

 51%|█████▏    | 6845/13317 [1:10:33<1:06:22,  1.63it/s]
 51%|█████▏    | 6846/13317 [1:10:33<1:09:50,  1.54it/s]
                                                        
{'loss': 0.3844, 'grad_norm': 0.6784449815750122, 'learning_rate': 5.625495879959162e-06, 'epoch': 1.54}

 51%|█████▏    | 6846/13317 [1:10:33<1:09:50,  1.54it/s]
 51%|█████▏    | 6847/13317 [1:10:34<1:08:49,  1.57it/s]
                                                        
{'loss': 0.2162, 'grad_norm': 0.6471434831619263, 'learning_rate': 5.624195519285646e-06, 'epoch': 1.54}

 51%|█████▏    | 6847/13317 [1:10:34<1:08:49,  1.57it/s]
 51%|█████▏    | 6848/13317 [1:10:34<1:08:01,  1.59it/s]
                                                        
{'loss': 0.1428, 'grad_norm': 0.4250980019569397, 'learning_rate': 5.622895115723288e-06, 'epoch': 1.54}

 51%|█████▏    | 6848/13317 [1:10:34<1:08:01,  1.59it/s]
 51%|█████▏    | 6849/13317 [1:10:35<1:07:26,  1.60it/s]
                                                        
{'loss': 0.1485, 'grad_norm': 0.47087159752845764, 'learning_rate': 5.62159466936144e-06, 'epoch': 1.54}

 51%|█████▏    | 6849/13317 [1:10:35<1:07:26,  1.60it/s]
 51%|█████▏    | 6850/13317 [1:10:35<1:07:04,  1.61it/s]
                                                        
{'loss': 0.2201, 'grad_norm': 0.5873328447341919, 'learning_rate': 5.620294180289457e-06, 'epoch': 1.54}

 51%|█████▏    | 6850/13317 [1:10:36<1:07:04,  1.61it/s]
 51%|█████▏    | 6851/13317 [1:10:36<1:06:55,  1.61it/s]
                                                        
{'loss': 0.2037, 'grad_norm': 0.5765529870986938, 'learning_rate': 5.6189936485966955e-06, 'epoch': 1.54}

 51%|█████▏    | 6851/13317 [1:10:36<1:06:55,  1.61it/s]
 51%|█████▏    | 6852/13317 [1:10:37<1:10:08,  1.54it/s]
                                                        
{'loss': 0.1478, 'grad_norm': 0.4412636458873749, 'learning_rate': 5.617693074372518e-06, 'epoch': 1.54}

 51%|█████▏    | 6852/13317 [1:10:37<1:10:08,  1.54it/s]
 51%|█████▏    | 6853/13317 [1:10:37<1:08:56,  1.56it/s]
                                                        
{'loss': 0.1808, 'grad_norm': 0.553290069103241, 'learning_rate': 5.616392457706284e-06, 'epoch': 1.54}

 51%|█████▏    | 6853/13317 [1:10:38<1:08:56,  1.56it/s]
 51%|█████▏    | 6854/13317 [1:10:38<1:08:06,  1.58it/s]
                                                        
{'loss': 0.1985, 'grad_norm': 0.5340937376022339, 'learning_rate': 5.615091798687362e-06, 'epoch': 1.54}

 51%|█████▏    | 6854/13317 [1:10:38<1:08:06,  1.58it/s]
 51%|█████▏    | 6855/13317 [1:10:39<1:07:30,  1.60it/s]
                                                        
{'loss': 0.1955, 'grad_norm': 0.5323349833488464, 'learning_rate': 5.613791097405121e-06, 'epoch': 1.54}

 51%|█████▏    | 6855/13317 [1:10:39<1:07:30,  1.60it/s]
 51%|█████▏    | 6856/13317 [1:10:39<1:07:04,  1.61it/s]
                                                        
{'loss': 0.186, 'grad_norm': 0.4879344403743744, 'learning_rate': 5.612490353948933e-06, 'epoch': 1.54}

 51%|█████▏    | 6856/13317 [1:10:39<1:07:04,  1.61it/s]
 51%|█████▏    | 6857/13317 [1:10:40<1:06:47,  1.61it/s]
                                                        
{'loss': 0.2467, 'grad_norm': 0.6438469886779785, 'learning_rate': 5.611189568408173e-06, 'epoch': 1.54}

 51%|█████▏    | 6857/13317 [1:10:40<1:06:47,  1.61it/s]
 51%|█████▏    | 6858/13317 [1:10:41<1:10:04,  1.54it/s]
                                                        
{'loss': 0.2394, 'grad_norm': 0.5873123407363892, 'learning_rate': 5.609888740872218e-06, 'epoch': 1.54}

 51%|█████▏    | 6858/13317 [1:10:41<1:10:04,  1.54it/s]
 52%|█████▏    | 6859/13317 [1:10:41<1:08:53,  1.56it/s]
                                                        
{'loss': 0.2003, 'grad_norm': 0.5571212768554688, 'learning_rate': 5.608587871430448e-06, 'epoch': 1.55}

 52%|█████▏    | 6859/13317 [1:10:41<1:08:53,  1.56it/s]
 52%|█████▏    | 6860/13317 [1:10:42<1:08:03,  1.58it/s]
                                                        
{'loss': 0.1784, 'grad_norm': 0.5455048680305481, 'learning_rate': 5.607286960172249e-06, 'epoch': 1.55}

 52%|█████▏    | 6860/13317 [1:10:42<1:08:03,  1.58it/s]
 52%|█████▏    | 6861/13317 [1:10:42<1:07:24,  1.60it/s]
                                                        
{'loss': 0.1627, 'grad_norm': 0.406173974275589, 'learning_rate': 5.6059860071870045e-06, 'epoch': 1.55}

 52%|█████▏    | 6861/13317 [1:10:43<1:07:24,  1.60it/s]
 52%|█████▏    | 6862/13317 [1:10:43<1:06:59,  1.61it/s]
                                                        
{'loss': 0.1438, 'grad_norm': 0.38732701539993286, 'learning_rate': 5.604685012564105e-06, 'epoch': 1.55}

 52%|█████▏    | 6862/13317 [1:10:43<1:06:59,  1.61it/s]
 52%|█████▏    | 6863/13317 [1:10:44<1:06:42,  1.61it/s]
                                                        
{'loss': 0.2517, 'grad_norm': 0.8559027910232544, 'learning_rate': 5.6033839763929436e-06, 'epoch': 1.55}

 52%|█████▏    | 6863/13317 [1:10:44<1:06:42,  1.61it/s]
 52%|█████▏    | 6864/13317 [1:10:44<1:06:34,  1.62it/s]
                                                        
{'loss': 0.2443, 'grad_norm': 0.5462571978569031, 'learning_rate': 5.602082898762913e-06, 'epoch': 1.55}

 52%|█████▏    | 6864/13317 [1:10:44<1:06:34,  1.62it/s]
 52%|█████▏    | 6865/13317 [1:10:45<1:06:21,  1.62it/s]
                                                        
{'loss': 0.2263, 'grad_norm': 0.5614378452301025, 'learning_rate': 5.600781779763414e-06, 'epoch': 1.55}

 52%|█████▏    | 6865/13317 [1:10:45<1:06:21,  1.62it/s]
 52%|█████▏    | 6866/13317 [1:10:46<1:06:12,  1.62it/s]
                                                        
{'loss': 0.2032, 'grad_norm': 0.5728249549865723, 'learning_rate': 5.5994806194838435e-06, 'epoch': 1.55}

 52%|█████▏    | 6866/13317 [1:10:46<1:06:12,  1.62it/s]
 52%|█████▏    | 6867/13317 [1:10:46<1:06:07,  1.63it/s]
                                                        
{'loss': 0.1643, 'grad_norm': 0.4580096900463104, 'learning_rate': 5.598179418013607e-06, 'epoch': 1.55}

 52%|█████▏    | 6867/13317 [1:10:46<1:06:07,  1.63it/s]
 52%|█████▏    | 6868/13317 [1:10:47<1:06:05,  1.63it/s]
                                                        
{'loss': 0.2091, 'grad_norm': 0.6097751259803772, 'learning_rate': 5.596878175442113e-06, 'epoch': 1.55}

 52%|█████▏    | 6868/13317 [1:10:47<1:06:05,  1.63it/s]
 52%|█████▏    | 6869/13317 [1:10:47<1:06:04,  1.63it/s]
                                                        
{'loss': 0.1806, 'grad_norm': 0.492161363363266, 'learning_rate': 5.595576891858767e-06, 'epoch': 1.55}

 52%|█████▏    | 6869/13317 [1:10:47<1:06:04,  1.63it/s]
 52%|█████▏    | 6870/13317 [1:10:48<1:06:01,  1.63it/s]
                                                        
{'loss': 0.2018, 'grad_norm': 0.49287253618240356, 'learning_rate': 5.594275567352984e-06, 'epoch': 1.55}

 52%|█████▏    | 6870/13317 [1:10:48<1:06:01,  1.63it/s]
 52%|█████▏    | 6871/13317 [1:10:49<1:05:58,  1.63it/s]
                                                        
{'loss': 0.2328, 'grad_norm': 0.5852052569389343, 'learning_rate': 5.592974202014174e-06, 'epoch': 1.55}

 52%|█████▏    | 6871/13317 [1:10:49<1:05:58,  1.63it/s]
 52%|█████▏    | 6872/13317 [1:10:49<1:05:58,  1.63it/s]
                                                        
{'loss': 0.1709, 'grad_norm': 0.4645231366157532, 'learning_rate': 5.591672795931761e-06, 'epoch': 1.55}

 52%|█████▏    | 6872/13317 [1:10:49<1:05:58,  1.63it/s]
 52%|█████▏    | 6873/13317 [1:10:50<1:05:57,  1.63it/s]
                                                        
{'loss': 0.2406, 'grad_norm': 0.6288430690765381, 'learning_rate': 5.59037134919516e-06, 'epoch': 1.55}

 52%|█████▏    | 6873/13317 [1:10:50<1:05:57,  1.63it/s]
 52%|█████▏    | 6874/13317 [1:10:50<1:06:01,  1.63it/s]
                                                        
{'loss': 0.2354, 'grad_norm': 0.619368851184845, 'learning_rate': 5.589069861893799e-06, 'epoch': 1.55}

 52%|█████▏    | 6874/13317 [1:10:51<1:06:01,  1.63it/s]
 52%|█████▏    | 6875/13317 [1:10:51<1:05:57,  1.63it/s]
                                                        
{'loss': 0.1464, 'grad_norm': 0.5076063275337219, 'learning_rate': 5.587768334117099e-06, 'epoch': 1.55}

 52%|█████▏    | 6875/13317 [1:10:51<1:05:57,  1.63it/s]
 52%|█████▏    | 6876/13317 [1:10:52<1:05:55,  1.63it/s]
                                                        
{'loss': 0.1916, 'grad_norm': 0.4913501441478729, 'learning_rate': 5.586466765954491e-06, 'epoch': 1.55}

 52%|█████▏    | 6876/13317 [1:10:52<1:05:55,  1.63it/s]
 52%|█████▏    | 6877/13317 [1:10:52<1:05:52,  1.63it/s]
                                                        
{'loss': 0.168, 'grad_norm': 0.4808025360107422, 'learning_rate': 5.585165157495409e-06, 'epoch': 1.55}

 52%|█████▏    | 6877/13317 [1:10:52<1:05:52,  1.63it/s]
 52%|█████▏    | 6878/13317 [1:10:53<1:05:52,  1.63it/s]
                                                        
{'loss': 0.1592, 'grad_norm': 0.4619109034538269, 'learning_rate': 5.583863508829281e-06, 'epoch': 1.55}

 52%|█████▏    | 6878/13317 [1:10:53<1:05:52,  1.63it/s]
 52%|█████▏    | 6879/13317 [1:10:53<1:05:57,  1.63it/s]
                                                        
{'loss': 0.1615, 'grad_norm': 0.48559340834617615, 'learning_rate': 5.582561820045552e-06, 'epoch': 1.55}

 52%|█████▏    | 6879/13317 [1:10:54<1:05:57,  1.63it/s]
 52%|█████▏    | 6880/13317 [1:10:54<1:05:53,  1.63it/s]
                                                        
{'loss': 0.2355, 'grad_norm': 0.6570034027099609, 'learning_rate': 5.581260091233655e-06, 'epoch': 1.55}

 52%|█████▏    | 6880/13317 [1:10:54<1:05:53,  1.63it/s]
 52%|█████▏    | 6881/13317 [1:10:55<1:05:51,  1.63it/s]
                                                        
{'loss': 0.2246, 'grad_norm': 0.5265800952911377, 'learning_rate': 5.579958322483037e-06, 'epoch': 1.55}

 52%|█████▏    | 6881/13317 [1:10:55<1:05:51,  1.63it/s]
 52%|█████▏    | 6882/13317 [1:10:55<1:05:49,  1.63it/s]
                                                        
{'loss': 0.1632, 'grad_norm': 0.4618828594684601, 'learning_rate': 5.578656513883139e-06, 'epoch': 1.55}

 52%|█████▏    | 6882/13317 [1:10:55<1:05:49,  1.63it/s]
 52%|█████▏    | 6883/13317 [1:10:56<1:05:48,  1.63it/s]
                                                        
{'loss': 0.2363, 'grad_norm': 0.5161967277526855, 'learning_rate': 5.577354665523413e-06, 'epoch': 1.55}

 52%|█████▏    | 6883/13317 [1:10:56<1:05:48,  1.63it/s]
 52%|█████▏    | 6884/13317 [1:10:57<1:05:48,  1.63it/s]
                                                        
{'loss': 0.1851, 'grad_norm': 0.49604710936546326, 'learning_rate': 5.576052777493309e-06, 'epoch': 1.55}

 52%|█████▏    | 6884/13317 [1:10:57<1:05:48,  1.63it/s]
 52%|█████▏    | 6885/13317 [1:10:57<1:05:55,  1.63it/s]
                                                        
{'loss': 0.2223, 'grad_norm': 0.5873522162437439, 'learning_rate': 5.57475084988228e-06, 'epoch': 1.55}

 52%|█████▏    | 6885/13317 [1:10:57<1:05:55,  1.63it/s]
 52%|█████▏    | 6886/13317 [1:10:58<1:05:54,  1.63it/s]
                                                        
{'loss': 0.213, 'grad_norm': 0.5567418336868286, 'learning_rate': 5.573448882779782e-06, 'epoch': 1.55}

 52%|█████▏    | 6886/13317 [1:10:58<1:05:54,  1.63it/s]
 52%|█████▏    | 6887/13317 [1:10:58<1:05:54,  1.63it/s]
                                                        
{'loss': 0.1788, 'grad_norm': 0.5051569938659668, 'learning_rate': 5.572146876275273e-06, 'epoch': 1.55}

 52%|█████▏    | 6887/13317 [1:10:59<1:05:54,  1.63it/s]
 52%|█████▏    | 6888/13317 [1:10:59<1:05:52,  1.63it/s]
                                                        
{'loss': 0.1769, 'grad_norm': 0.4503238797187805, 'learning_rate': 5.570844830458216e-06, 'epoch': 1.55}

 52%|█████▏    | 6888/13317 [1:10:59<1:05:52,  1.63it/s]
 52%|█████▏    | 6889/13317 [1:11:00<1:05:54,  1.63it/s]
                                                        
{'loss': 0.2833, 'grad_norm': 0.6610566973686218, 'learning_rate': 5.569542745418075e-06, 'epoch': 1.55}

 52%|█████▏    | 6889/13317 [1:11:00<1:05:54,  1.63it/s]
 52%|█████▏    | 6890/13317 [1:11:00<1:05:52,  1.63it/s]
                                                        
{'loss': 0.2206, 'grad_norm': 0.6416619420051575, 'learning_rate': 5.568240621244317e-06, 'epoch': 1.55}

 52%|█████▏    | 6890/13317 [1:11:00<1:05:52,  1.63it/s]
 52%|█████▏    | 6891/13317 [1:11:01<1:05:55,  1.62it/s]
                                                        
{'loss': 0.1658, 'grad_norm': 0.5165424346923828, 'learning_rate': 5.566938458026411e-06, 'epoch': 1.55}

 52%|█████▏    | 6891/13317 [1:11:01<1:05:55,  1.62it/s]
 52%|█████▏    | 6892/13317 [1:11:01<1:05:52,  1.63it/s]
                                                        
{'loss': 0.1851, 'grad_norm': 0.5155500173568726, 'learning_rate': 5.565636255853832e-06, 'epoch': 1.55}

 52%|█████▏    | 6892/13317 [1:11:02<1:05:52,  1.63it/s]
 52%|█████▏    | 6893/13317 [1:11:02<1:05:47,  1.63it/s]
                                                        
{'loss': 0.1667, 'grad_norm': 0.46313342452049255, 'learning_rate': 5.56433401481605e-06, 'epoch': 1.55}

 52%|█████▏    | 6893/13317 [1:11:02<1:05:47,  1.63it/s]
 52%|█████▏    | 6894/13317 [1:11:03<1:05:52,  1.63it/s]
                                                        
{'loss': 0.2103, 'grad_norm': 0.5713981986045837, 'learning_rate': 5.563031735002548e-06, 'epoch': 1.55}

 52%|█████▏    | 6894/13317 [1:11:03<1:05:52,  1.63it/s]
 52%|█████▏    | 6895/13317 [1:11:03<1:05:46,  1.63it/s]
                                                        
{'loss': 0.1993, 'grad_norm': 0.6222763657569885, 'learning_rate': 5.561729416502804e-06, 'epoch': 1.55}

 52%|█████▏    | 6895/13317 [1:11:03<1:05:46,  1.63it/s]
 52%|█████▏    | 6896/13317 [1:11:04<1:05:45,  1.63it/s]
                                                        
{'loss': 0.2163, 'grad_norm': 0.5716380476951599, 'learning_rate': 5.560427059406302e-06, 'epoch': 1.55}

 52%|█████▏    | 6896/13317 [1:11:04<1:05:45,  1.63it/s]
 52%|█████▏    | 6897/13317 [1:11:05<1:05:43,  1.63it/s]
                                                        
{'loss': 0.1897, 'grad_norm': 0.5012927055358887, 'learning_rate': 5.559124663802527e-06, 'epoch': 1.55}

 52%|█████▏    | 6897/13317 [1:11:05<1:05:43,  1.63it/s]
 52%|█████▏    | 6898/13317 [1:11:05<1:05:43,  1.63it/s]
                                                        
{'loss': 0.2092, 'grad_norm': 0.6024609208106995, 'learning_rate': 5.557822229780967e-06, 'epoch': 1.55}

 52%|█████▏    | 6898/13317 [1:11:05<1:05:43,  1.63it/s]
 52%|█████▏    | 6899/13317 [1:11:06<1:05:47,  1.63it/s]
                                                        
{'loss': 0.2013, 'grad_norm': 0.5321360230445862, 'learning_rate': 5.556519757431115e-06, 'epoch': 1.55}

 52%|█████▏    | 6899/13317 [1:11:06<1:05:47,  1.63it/s]
 52%|█████▏    | 6900/13317 [1:11:06<1:05:44,  1.63it/s]
                                                        
{'loss': 0.1669, 'grad_norm': 0.4925098717212677, 'learning_rate': 5.5552172468424605e-06, 'epoch': 1.55}

 52%|█████▏    | 6900/13317 [1:11:07<1:05:44,  1.63it/s]
 52%|█████▏    | 6901/13317 [1:11:07<1:05:39,  1.63it/s]
                                                        
{'loss': 0.2339, 'grad_norm': 0.5471585392951965, 'learning_rate': 5.5539146981045044e-06, 'epoch': 1.55}

 52%|█████▏    | 6901/13317 [1:11:07<1:05:39,  1.63it/s]
 52%|█████▏    | 6902/13317 [1:11:08<1:05:43,  1.63it/s]
                                                        
{'loss': 0.1757, 'grad_norm': 0.5026326179504395, 'learning_rate': 5.552612111306742e-06, 'epoch': 1.55}

 52%|█████▏    | 6902/13317 [1:11:08<1:05:43,  1.63it/s]
 52%|█████▏    | 6903/13317 [1:11:08<1:05:41,  1.63it/s]
                                                        
{'loss': 0.1699, 'grad_norm': 0.49965500831604004, 'learning_rate': 5.551309486538678e-06, 'epoch': 1.56}

 52%|█████▏    | 6903/13317 [1:11:08<1:05:41,  1.63it/s]
 52%|█████▏    | 6904/13317 [1:11:09<1:05:42,  1.63it/s]
                                                        
{'loss': 0.2459, 'grad_norm': 0.5836755633354187, 'learning_rate': 5.550006823889814e-06, 'epoch': 1.56}

 52%|█████▏    | 6904/13317 [1:11:09<1:05:42,  1.63it/s]
 52%|█████▏    | 6905/13317 [1:11:09<1:05:39,  1.63it/s]
                                                        
{'loss': 0.209, 'grad_norm': 0.4810606837272644, 'learning_rate': 5.548704123449657e-06, 'epoch': 1.56}

 52%|█████▏    | 6905/13317 [1:11:10<1:05:39,  1.63it/s]
 52%|█████▏    | 6906/13317 [1:11:10<1:05:43,  1.63it/s]
                                                        
{'loss': 0.1383, 'grad_norm': 0.4009459614753723, 'learning_rate': 5.547401385307719e-06, 'epoch': 1.56}

 52%|█████▏    | 6906/13317 [1:11:10<1:05:43,  1.63it/s]
 52%|█████▏    | 6907/13317 [1:11:11<1:05:43,  1.63it/s]
                                                        
{'loss': 0.1706, 'grad_norm': 0.4926522672176361, 'learning_rate': 5.546098609553509e-06, 'epoch': 1.56}

 52%|█████▏    | 6907/13317 [1:11:11<1:05:43,  1.63it/s]
 52%|█████▏    | 6908/13317 [1:11:11<1:05:38,  1.63it/s]
                                                        
{'loss': 0.185, 'grad_norm': 0.5369823575019836, 'learning_rate': 5.544795796276541e-06, 'epoch': 1.56}

 52%|█████▏    | 6908/13317 [1:11:11<1:05:38,  1.63it/s]
 52%|█████▏    | 6909/13317 [1:11:12<1:05:41,  1.63it/s]
                                                        
{'loss': 0.2385, 'grad_norm': 0.5954380035400391, 'learning_rate': 5.543492945566334e-06, 'epoch': 1.56}

 52%|█████▏    | 6909/13317 [1:11:12<1:05:41,  1.63it/s]
 52%|█████▏    | 6910/13317 [1:11:13<1:05:38,  1.63it/s]
                                                        
{'loss': 0.1894, 'grad_norm': 0.5182493329048157, 'learning_rate': 5.542190057512407e-06, 'epoch': 1.56}

 52%|█████▏    | 6910/13317 [1:11:13<1:05:38,  1.63it/s]
 52%|█████▏    | 6911/13317 [1:11:13<1:05:39,  1.63it/s]
                                                        
{'loss': 0.1904, 'grad_norm': 0.5490275621414185, 'learning_rate': 5.540887132204283e-06, 'epoch': 1.56}

 52%|█████▏    | 6911/13317 [1:11:13<1:05:39,  1.63it/s]
 52%|█████▏    | 6912/13317 [1:11:14<1:05:36,  1.63it/s]
                                                        
{'loss': 0.141, 'grad_norm': 0.4267297089099884, 'learning_rate': 5.539584169731486e-06, 'epoch': 1.56}

 52%|█████▏    | 6912/13317 [1:11:14<1:05:36,  1.63it/s]
 52%|█████▏    | 6913/13317 [1:11:14<1:05:32,  1.63it/s]
                                                        
{'loss': 0.2051, 'grad_norm': 0.5373473763465881, 'learning_rate': 5.538281170183544e-06, 'epoch': 1.56}

 52%|█████▏    | 6913/13317 [1:11:15<1:05:32,  1.63it/s]
 52%|█████▏    | 6914/13317 [1:11:15<1:05:36,  1.63it/s]
                                                        
{'loss': 0.2078, 'grad_norm': 0.5544183254241943, 'learning_rate': 5.536978133649985e-06, 'epoch': 1.56}

 52%|█████▏    | 6914/13317 [1:11:15<1:05:36,  1.63it/s]
 52%|█████▏    | 6915/13317 [1:11:16<1:05:31,  1.63it/s]
                                                        
{'loss': 0.1959, 'grad_norm': 0.5074737071990967, 'learning_rate': 5.535675060220342e-06, 'epoch': 1.56}

 52%|█████▏    | 6915/13317 [1:11:16<1:05:31,  1.63it/s]
 52%|█████▏    | 6916/13317 [1:11:16<1:05:33,  1.63it/s]
                                                        
{'loss': 0.205, 'grad_norm': 0.49678707122802734, 'learning_rate': 5.534371949984152e-06, 'epoch': 1.56}

 52%|█████▏    | 6916/13317 [1:11:16<1:05:33,  1.63it/s]
 52%|█████▏    | 6917/13317 [1:11:17<1:05:30,  1.63it/s]
                                                        
{'loss': 0.2287, 'grad_norm': 0.5587177276611328, 'learning_rate': 5.53306880303095e-06, 'epoch': 1.56}

 52%|█████▏    | 6917/13317 [1:11:17<1:05:30,  1.63it/s]
 52%|█████▏    | 6918/13317 [1:11:17<1:05:27,  1.63it/s]
                                                        
{'loss': 0.1697, 'grad_norm': 0.5004328489303589, 'learning_rate': 5.531765619450278e-06, 'epoch': 1.56}

 52%|█████▏    | 6918/13317 [1:11:18<1:05:27,  1.63it/s]
 52%|█████▏    | 6919/13317 [1:11:18<1:05:30,  1.63it/s]
                                                        
{'loss': 0.1591, 'grad_norm': 0.4691566526889801, 'learning_rate': 5.530462399331677e-06, 'epoch': 1.56}

 52%|█████▏    | 6919/13317 [1:11:18<1:05:30,  1.63it/s]
 52%|█████▏    | 6920/13317 [1:11:19<1:05:28,  1.63it/s]
                                                        
{'loss': 0.1568, 'grad_norm': 0.5196841955184937, 'learning_rate': 5.529159142764692e-06, 'epoch': 1.56}

 52%|█████▏    | 6920/13317 [1:11:19<1:05:28,  1.63it/s]
 52%|█████▏    | 6921/13317 [1:11:19<1:05:30,  1.63it/s]
                                                        
{'loss': 0.2368, 'grad_norm': 0.6081128120422363, 'learning_rate': 5.527855849838872e-06, 'epoch': 1.56}

 52%|█████▏    | 6921/13317 [1:11:19<1:05:30,  1.63it/s]
 52%|█████▏    | 6922/13317 [1:11:20<1:05:29,  1.63it/s]
                                                        
{'loss': 0.2129, 'grad_norm': 0.5785732269287109, 'learning_rate': 5.526552520643766e-06, 'epoch': 1.56}

 52%|█████▏    | 6922/13317 [1:11:20<1:05:29,  1.63it/s]
 52%|█████▏    | 6923/13317 [1:11:21<1:05:30,  1.63it/s]
                                                        
{'loss': 0.1921, 'grad_norm': 0.5129921436309814, 'learning_rate': 5.525249155268926e-06, 'epoch': 1.56}

 52%|█████▏    | 6923/13317 [1:11:21<1:05:30,  1.63it/s]
 52%|█████▏    | 6924/13317 [1:11:21<1:05:27,  1.63it/s]
                                                        
{'loss': 0.2229, 'grad_norm': 0.5517379641532898, 'learning_rate': 5.523945753803908e-06, 'epoch': 1.56}

 52%|█████▏    | 6924/13317 [1:11:21<1:05:27,  1.63it/s]
 52%|█████▏    | 6925/13317 [1:11:22<1:05:24,  1.63it/s]
                                                        
{'loss': 0.1732, 'grad_norm': 0.440237820148468, 'learning_rate': 5.522642316338268e-06, 'epoch': 1.56}

 52%|█████▏    | 6925/13317 [1:11:22<1:05:24,  1.63it/s]
 52%|█████▏    | 6926/13317 [1:11:22<1:05:25,  1.63it/s]
                                                        
{'loss': 0.1874, 'grad_norm': 0.5968524813652039, 'learning_rate': 5.521338842961569e-06, 'epoch': 1.56}

 52%|█████▏    | 6926/13317 [1:11:23<1:05:25,  1.63it/s]
 52%|█████▏    | 6927/13317 [1:11:23<1:05:22,  1.63it/s]
                                                        
{'loss': 0.1999, 'grad_norm': 0.43646642565727234, 'learning_rate': 5.520035333763368e-06, 'epoch': 1.56}

 52%|█████▏    | 6927/13317 [1:11:23<1:05:22,  1.63it/s]
 52%|█████▏    | 6928/13317 [1:11:24<1:05:20,  1.63it/s]
                                                        
{'loss': 0.1617, 'grad_norm': 0.4307858943939209, 'learning_rate': 5.518731788833237e-06, 'epoch': 1.56}

 52%|█████▏    | 6928/13317 [1:11:24<1:05:20,  1.63it/s]
 52%|█████▏    | 6929/13317 [1:11:24<1:05:26,  1.63it/s]
                                                        
{'loss': 0.1541, 'grad_norm': 0.4223570227622986, 'learning_rate': 5.5174282082607385e-06, 'epoch': 1.56}

 52%|█████▏    | 6929/13317 [1:11:24<1:05:26,  1.63it/s]
 52%|█████▏    | 6930/13317 [1:11:25<1:05:23,  1.63it/s]
                                                        
{'loss': 0.2804, 'grad_norm': 0.6235194206237793, 'learning_rate': 5.516124592135443e-06, 'epoch': 1.56}

 52%|█████▏    | 6930/13317 [1:11:25<1:05:23,  1.63it/s]
 52%|█████▏    | 6931/13317 [1:11:25<1:05:27,  1.63it/s]
                                                        
{'loss': 0.2338, 'grad_norm': 0.6387377381324768, 'learning_rate': 5.514820940546924e-06, 'epoch': 1.56}

 52%|█████▏    | 6931/13317 [1:11:26<1:05:27,  1.63it/s]
 52%|█████▏    | 6932/13317 [1:11:26<1:05:27,  1.63it/s]
                                                        
{'loss': 0.2449, 'grad_norm': 0.6545002460479736, 'learning_rate': 5.513517253584755e-06, 'epoch': 1.56}

 52%|█████▏    | 6932/13317 [1:11:26<1:05:27,  1.63it/s]
 52%|█████▏    | 6933/13317 [1:11:27<1:05:23,  1.63it/s]
                                                        
{'loss': 0.1572, 'grad_norm': 0.42002758383750916, 'learning_rate': 5.512213531338514e-06, 'epoch': 1.56}

 52%|█████▏    | 6933/13317 [1:11:27<1:05:23,  1.63it/s]
 52%|█████▏    | 6934/13317 [1:11:27<1:05:24,  1.63it/s]
                                                        
{'loss': 0.2293, 'grad_norm': 0.6958950161933899, 'learning_rate': 5.510909773897778e-06, 'epoch': 1.56}

 52%|█████▏    | 6934/13317 [1:11:27<1:05:24,  1.63it/s]
 52%|█████▏    | 6935/13317 [1:11:28<1:05:23,  1.63it/s]
                                                        
{'loss': 0.1981, 'grad_norm': 0.5014004707336426, 'learning_rate': 5.509605981352134e-06, 'epoch': 1.56}

 52%|█████▏    | 6935/13317 [1:11:28<1:05:23,  1.63it/s]
 52%|█████▏    | 6936/13317 [1:11:29<1:05:23,  1.63it/s]
                                                        
{'loss': 0.224, 'grad_norm': 0.5436384081840515, 'learning_rate': 5.508302153791161e-06, 'epoch': 1.56}

 52%|█████▏    | 6936/13317 [1:11:29<1:05:23,  1.63it/s]
 52%|█████▏    | 6937/13317 [1:11:29<1:05:19,  1.63it/s]
                                                        
{'loss': 0.1725, 'grad_norm': 0.5131723284721375, 'learning_rate': 5.506998291304448e-06, 'epoch': 1.56}

 52%|█████▏    | 6937/13317 [1:11:29<1:05:19,  1.63it/s]
 52%|█████▏    | 6938/13317 [1:11:30<1:05:19,  1.63it/s]
                                                        
{'loss': 0.2041, 'grad_norm': 0.5414015054702759, 'learning_rate': 5.505694393981586e-06, 'epoch': 1.56}

 52%|█████▏    | 6938/13317 [1:11:30<1:05:19,  1.63it/s]
 52%|█████▏    | 6939/13317 [1:11:30<1:05:20,  1.63it/s]
                                                        
{'loss': 0.1767, 'grad_norm': 0.4879447817802429, 'learning_rate': 5.504390461912163e-06, 'epoch': 1.56}

 52%|█████▏    | 6939/13317 [1:11:30<1:05:20,  1.63it/s]
 52%|█████▏    | 6940/13317 [1:11:31<1:05:19,  1.63it/s]
                                                        
{'loss': 0.2094, 'grad_norm': 0.4993137717247009, 'learning_rate': 5.503086495185777e-06, 'epoch': 1.56}

 52%|█████▏    | 6940/13317 [1:11:31<1:05:19,  1.63it/s]
 52%|█████▏    | 6941/13317 [1:11:32<1:05:19,  1.63it/s]
                                                        
{'loss': 0.1925, 'grad_norm': 0.5352160334587097, 'learning_rate': 5.501782493892019e-06, 'epoch': 1.56}

 52%|█████▏    | 6941/13317 [1:11:32<1:05:19,  1.63it/s]
 52%|█████▏    | 6942/13317 [1:11:32<1:05:16,  1.63it/s]
                                                        
{'loss': 0.2464, 'grad_norm': 0.5038251876831055, 'learning_rate': 5.500478458120493e-06, 'epoch': 1.56}

 52%|█████▏    | 6942/13317 [1:11:32<1:05:16,  1.63it/s]
 52%|█████▏    | 6943/13317 [1:11:33<1:05:14,  1.63it/s]
                                                        
{'loss': 0.158, 'grad_norm': 0.4556039571762085, 'learning_rate': 5.499174387960796e-06, 'epoch': 1.56}

 52%|█████▏    | 6943/13317 [1:11:33<1:05:14,  1.63it/s]
 52%|█████▏    | 6944/13317 [1:11:33<1:05:17,  1.63it/s]
                                                        
{'loss': 0.2055, 'grad_norm': 0.5610651969909668, 'learning_rate': 5.497870283502535e-06, 'epoch': 1.56}

 52%|█████▏    | 6944/13317 [1:11:34<1:05:17,  1.63it/s]
 52%|█████▏    | 6945/13317 [1:11:34<1:05:16,  1.63it/s]
                                                        
{'loss': 0.206, 'grad_norm': 0.5146677494049072, 'learning_rate': 5.496566144835313e-06, 'epoch': 1.56}

 52%|█████▏    | 6945/13317 [1:11:34<1:05:16,  1.63it/s]
 52%|█████▏    | 6946/13317 [1:11:35<1:05:13,  1.63it/s]
                                                        
{'loss': 0.2075, 'grad_norm': 0.537939190864563, 'learning_rate': 5.49526197204874e-06, 'epoch': 1.56}

 52%|█████▏    | 6946/13317 [1:11:35<1:05:13,  1.63it/s]
 52%|█████▏    | 6947/13317 [1:11:35<1:05:13,  1.63it/s]
                                                        
{'loss': 0.1732, 'grad_norm': 0.46036484837532043, 'learning_rate': 5.493957765232423e-06, 'epoch': 1.56}

 52%|█████▏    | 6947/13317 [1:11:35<1:05:13,  1.63it/s]
 52%|█████▏    | 6948/13317 [1:11:36<1:05:11,  1.63it/s]
                                                        
{'loss': 0.223, 'grad_norm': 0.6221325397491455, 'learning_rate': 5.49265352447598e-06, 'epoch': 1.57}

 52%|█████▏    | 6948/13317 [1:11:36<1:05:11,  1.63it/s]
 52%|█████▏    | 6949/13317 [1:11:37<1:05:15,  1.63it/s]
                                                        
{'loss': 0.235, 'grad_norm': 0.5207458734512329, 'learning_rate': 5.4913492498690224e-06, 'epoch': 1.57}

 52%|█████▏    | 6949/13317 [1:11:37<1:05:15,  1.63it/s]
 52%|█████▏    | 6950/13317 [1:11:37<1:05:12,  1.63it/s]
                                                        
{'loss': 0.2101, 'grad_norm': 0.5496876239776611, 'learning_rate': 5.490044941501169e-06, 'epoch': 1.57}

 52%|█████▏    | 6950/13317 [1:11:37<1:05:12,  1.63it/s]
 52%|█████▏    | 6951/13317 [1:11:38<1:05:17,  1.63it/s]
                                                        
{'loss': 0.2024, 'grad_norm': 0.49304476380348206, 'learning_rate': 5.4887405994620404e-06, 'epoch': 1.57}

 52%|█████▏    | 6951/13317 [1:11:38<1:05:17,  1.63it/s]
 52%|█████▏    | 6952/13317 [1:11:38<1:05:11,  1.63it/s]
                                                        
{'loss': 0.2494, 'grad_norm': 0.5997208952903748, 'learning_rate': 5.487436223841256e-06, 'epoch': 1.57}

 52%|█████▏    | 6952/13317 [1:11:38<1:05:11,  1.63it/s]
 52%|█████▏    | 6953/13317 [1:11:39<1:05:23,  1.62it/s]
                                                        
{'loss': 0.2481, 'grad_norm': 0.6559901237487793, 'learning_rate': 5.486131814728444e-06, 'epoch': 1.57}

 52%|█████▏    | 6953/13317 [1:11:39<1:05:23,  1.62it/s]
 52%|█████▏    | 6954/13317 [1:11:40<1:05:19,  1.62it/s]
                                                        
{'loss': 0.2038, 'grad_norm': 0.5964793562889099, 'learning_rate': 5.484827372213227e-06, 'epoch': 1.57}

 52%|█████▏    | 6954/13317 [1:11:40<1:05:19,  1.62it/s]
 52%|█████▏    | 6955/13317 [1:11:40<1:05:15,  1.62it/s]
                                                        
{'loss': 0.224, 'grad_norm': 0.5893142819404602, 'learning_rate': 5.483522896385238e-06, 'epoch': 1.57}

 52%|█████▏    | 6955/13317 [1:11:40<1:05:15,  1.62it/s]
 52%|█████▏    | 6956/13317 [1:11:41<1:05:17,  1.62it/s]
                                                        
{'loss': 0.22, 'grad_norm': 0.5249609351158142, 'learning_rate': 5.482218387334106e-06, 'epoch': 1.57}

 52%|█████▏    | 6956/13317 [1:11:41<1:05:17,  1.62it/s]
 52%|█████▏    | 6957/13317 [1:11:41<1:05:15,  1.62it/s]
                                                        
{'loss': 0.2087, 'grad_norm': 0.60160893201828, 'learning_rate': 5.480913845149464e-06, 'epoch': 1.57}

 52%|█████▏    | 6957/13317 [1:11:42<1:05:15,  1.62it/s]
 52%|█████▏    | 6958/13317 [1:11:42<1:05:12,  1.63it/s]
                                                        
{'loss': 0.1653, 'grad_norm': 0.4641706347465515, 'learning_rate': 5.479609269920949e-06, 'epoch': 1.57}

 52%|█████▏    | 6958/13317 [1:11:42<1:05:12,  1.63it/s]
 52%|█████▏    | 6959/13317 [1:11:43<1:05:14,  1.62it/s]
                                                        
{'loss': 0.2179, 'grad_norm': 0.5804892778396606, 'learning_rate': 5.4783046617382e-06, 'epoch': 1.57}

 52%|█████▏    | 6959/13317 [1:11:43<1:05:14,  1.62it/s]
 52%|█████▏    | 6960/13317 [1:11:43<1:05:09,  1.63it/s]
                                                        
{'loss': 0.1652, 'grad_norm': 0.4255198538303375, 'learning_rate': 5.477000020690856e-06, 'epoch': 1.57}

 52%|█████▏    | 6960/13317 [1:11:43<1:05:09,  1.63it/s]
 52%|█████▏    | 6961/13317 [1:11:44<1:05:11,  1.62it/s]
                                                        
{'loss': 0.2061, 'grad_norm': 0.5213895440101624, 'learning_rate': 5.475695346868558e-06, 'epoch': 1.57}

 52%|█████▏    | 6961/13317 [1:11:44<1:05:11,  1.62it/s]
 52%|█████▏    | 6962/13317 [1:11:45<1:05:06,  1.63it/s]
                                                        
{'loss': 0.1694, 'grad_norm': 0.5234227180480957, 'learning_rate': 5.474390640360956e-06, 'epoch': 1.57}

 52%|█████▏    | 6962/13317 [1:11:45<1:05:06,  1.63it/s]
 52%|█████▏    | 6963/13317 [1:11:45<1:05:02,  1.63it/s]
                                                        
{'loss': 0.2328, 'grad_norm': 0.5446740984916687, 'learning_rate': 5.473085901257693e-06, 'epoch': 1.57}

 52%|█████▏    | 6963/13317 [1:11:45<1:05:02,  1.63it/s]
 52%|█████▏    | 6964/13317 [1:11:46<1:05:07,  1.63it/s]
                                                        
{'loss': 0.2037, 'grad_norm': 0.5494585037231445, 'learning_rate': 5.471781129648419e-06, 'epoch': 1.57}

 52%|█████▏    | 6964/13317 [1:11:46<1:05:07,  1.63it/s]
 52%|█████▏    | 6965/13317 [1:11:46<1:05:03,  1.63it/s]
                                                        
{'loss': 0.1503, 'grad_norm': 0.45603489875793457, 'learning_rate': 5.470476325622785e-06, 'epoch': 1.57}

 52%|█████▏    | 6965/13317 [1:11:46<1:05:03,  1.63it/s]
 52%|█████▏    | 6966/13317 [1:11:47<1:05:25,  1.62it/s]
                                                        
{'loss': 0.1632, 'grad_norm': 0.4945991337299347, 'learning_rate': 5.469171489270449e-06, 'epoch': 1.57}

 52%|█████▏    | 6966/13317 [1:11:47<1:05:25,  1.62it/s]
 52%|█████▏    | 6967/13317 [1:11:48<1:05:21,  1.62it/s]
                                                        
{'loss': 0.1935, 'grad_norm': 0.524943470954895, 'learning_rate': 5.467866620681063e-06, 'epoch': 1.57}

 52%|█████▏    | 6967/13317 [1:11:48<1:05:21,  1.62it/s]
 52%|█████▏    | 6968/13317 [1:11:48<1:05:11,  1.62it/s]
                                                        
{'loss': 0.1633, 'grad_norm': 0.4710599482059479, 'learning_rate': 5.466561719944284e-06, 'epoch': 1.57}

 52%|█████▏    | 6968/13317 [1:11:48<1:05:11,  1.62it/s]
 52%|█████▏    | 6969/13317 [1:11:49<1:05:09,  1.62it/s]
                                                        
{'loss': 0.1692, 'grad_norm': 0.4915845990180969, 'learning_rate': 5.4652567871497775e-06, 'epoch': 1.57}

 52%|█████▏    | 6969/13317 [1:11:49<1:05:09,  1.62it/s]
 52%|█████▏    | 6970/13317 [1:11:49<1:05:06,  1.62it/s]
                                                        
{'loss': 0.1456, 'grad_norm': 0.4680841565132141, 'learning_rate': 5.463951822387201e-06, 'epoch': 1.57}

 52%|█████▏    | 6970/13317 [1:11:50<1:05:06,  1.62it/s]
 52%|█████▏    | 6971/13317 [1:11:50<1:05:07,  1.62it/s]
                                                        
{'loss': 0.1535, 'grad_norm': 0.42744630575180054, 'learning_rate': 5.462646825746224e-06, 'epoch': 1.57}

 52%|█████▏    | 6971/13317 [1:11:50<1:05:07,  1.62it/s]
 52%|█████▏    | 6972/13317 [1:11:51<1:05:07,  1.62it/s]
                                                        
{'loss': 0.1789, 'grad_norm': 0.46420496702194214, 'learning_rate': 5.46134179731651e-06, 'epoch': 1.57}

 52%|█████▏    | 6972/13317 [1:11:51<1:05:07,  1.62it/s]
 52%|█████▏    | 6973/13317 [1:11:51<1:05:03,  1.63it/s]
                                                        
{'loss': 0.2005, 'grad_norm': 0.4938583970069885, 'learning_rate': 5.4600367371877304e-06, 'epoch': 1.57}

 52%|█████▏    | 6973/13317 [1:11:51<1:05:03,  1.63it/s]
 52%|█████▏    | 6974/13317 [1:11:52<1:05:07,  1.62it/s]
                                                        
{'loss': 0.215, 'grad_norm': 0.49049219489097595, 'learning_rate': 5.458731645449556e-06, 'epoch': 1.57}

 52%|█████▏    | 6974/13317 [1:11:52<1:05:07,  1.62it/s]
 52%|█████▏    | 6975/13317 [1:11:53<1:05:03,  1.62it/s]
                                                        
{'loss': 0.2587, 'grad_norm': 0.6041080355644226, 'learning_rate': 5.457426522191662e-06, 'epoch': 1.57}

 52%|█████▏    | 6975/13317 [1:11:53<1:05:03,  1.62it/s]
 52%|█████▏    | 6976/13317 [1:11:53<1:05:01,  1.63it/s]
                                                        
{'loss': 0.2289, 'grad_norm': 0.5902938842773438, 'learning_rate': 5.456121367503721e-06, 'epoch': 1.57}

 52%|█████▏    | 6976/13317 [1:11:53<1:05:01,  1.63it/s]
 52%|█████▏    | 6977/13317 [1:11:54<1:04:58,  1.63it/s]
                                                        
{'loss': 0.2057, 'grad_norm': 0.5940871834754944, 'learning_rate': 5.454816181475413e-06, 'epoch': 1.57}

 52%|█████▏    | 6977/13317 [1:11:54<1:04:58,  1.63it/s]
 52%|█████▏    | 6978/13317 [1:11:54<1:04:55,  1.63it/s]
                                                        
{'loss': 0.2053, 'grad_norm': 0.5183795690536499, 'learning_rate': 5.453510964196417e-06, 'epoch': 1.57}

 52%|█████▏    | 6978/13317 [1:11:54<1:04:55,  1.63it/s]
 52%|█████▏    | 6979/13317 [1:11:55<1:05:01,  1.62it/s]
                                                        
{'loss': 0.2222, 'grad_norm': 0.5537951588630676, 'learning_rate': 5.452205715756415e-06, 'epoch': 1.57}

 52%|█████▏    | 6979/13317 [1:11:55<1:05:01,  1.62it/s]
 52%|█████▏    | 6980/13317 [1:11:56<1:04:56,  1.63it/s]
                                                        
{'loss': 0.1628, 'grad_norm': 0.45181766152381897, 'learning_rate': 5.450900436245096e-06, 'epoch': 1.57}

 52%|█████▏    | 6980/13317 [1:11:56<1:04:56,  1.63it/s]
 52%|█████▏    | 6981/13317 [1:11:56<1:04:57,  1.63it/s]
                                                        
{'loss': 0.1844, 'grad_norm': 0.4725763499736786, 'learning_rate': 5.4495951257521385e-06, 'epoch': 1.57}

 52%|█████▏    | 6981/13317 [1:11:56<1:04:57,  1.63it/s]
 52%|█████▏    | 6982/13317 [1:11:57<1:05:14,  1.62it/s]
                                                        
{'loss': 0.2073, 'grad_norm': 0.5643550753593445, 'learning_rate': 5.448289784367239e-06, 'epoch': 1.57}

 52%|█████▏    | 6982/13317 [1:11:57<1:05:14,  1.62it/s]
 52%|█████▏    | 6983/13317 [1:11:57<1:05:04,  1.62it/s]
                                                        
{'loss': 0.17, 'grad_norm': 0.5047155022621155, 'learning_rate': 5.446984412180082e-06, 'epoch': 1.57}

 52%|█████▏    | 6983/13317 [1:11:58<1:05:04,  1.62it/s]
 52%|█████▏    | 6984/13317 [1:11:58<1:05:04,  1.62it/s]
                                                        
{'loss': 0.2128, 'grad_norm': 0.5803389549255371, 'learning_rate': 5.445679009280364e-06, 'epoch': 1.57}

 52%|█████▏    | 6984/13317 [1:11:58<1:05:04,  1.62it/s]
 52%|█████▏    | 6985/13317 [1:11:59<1:04:58,  1.62it/s]
                                                        
{'loss': 0.1607, 'grad_norm': 0.42923039197921753, 'learning_rate': 5.44437357575778e-06, 'epoch': 1.57}

 52%|█████▏    | 6985/13317 [1:11:59<1:04:58,  1.62it/s]
 52%|█████▏    | 6986/13317 [1:11:59<1:04:56,  1.62it/s]
                                                        
{'loss': 0.2301, 'grad_norm': 0.5541378855705261, 'learning_rate': 5.443068111702026e-06, 'epoch': 1.57}

 52%|█████▏    | 6986/13317 [1:11:59<1:04:56,  1.62it/s]
 52%|█████▏    | 6987/13317 [1:12:00<1:04:57,  1.62it/s]
                                                        
{'loss': 0.3429, 'grad_norm': 0.4764331877231598, 'learning_rate': 5.441762617202801e-06, 'epoch': 1.57}

 52%|█████▏    | 6987/13317 [1:12:00<1:04:57,  1.62it/s]
 52%|█████▏    | 6988/13317 [1:12:01<1:04:53,  1.63it/s]
                                                        
{'loss': 0.1841, 'grad_norm': 0.49303990602493286, 'learning_rate': 5.4404570923498064e-06, 'epoch': 1.57}

 52%|█████▏    | 6988/13317 [1:12:01<1:04:53,  1.63it/s]
 52%|█████▏    | 6989/13317 [1:12:01<1:04:56,  1.62it/s]
                                                        
{'loss': 0.237, 'grad_norm': 0.5377315282821655, 'learning_rate': 5.439151537232747e-06, 'epoch': 1.57}

 52%|█████▏    | 6989/13317 [1:12:01<1:04:56,  1.62it/s]
 52%|█████▏    | 6990/13317 [1:12:02<1:04:51,  1.63it/s]
                                                        
{'loss': 0.2099, 'grad_norm': 0.5288496017456055, 'learning_rate': 5.4378459519413265e-06, 'epoch': 1.57}

 52%|█████▏    | 6990/13317 [1:12:02<1:04:51,  1.63it/s]
 52%|█████▏    | 6991/13317 [1:12:02<1:04:50,  1.63it/s]
                                                        
{'loss': 0.1708, 'grad_norm': 0.4803270697593689, 'learning_rate': 5.436540336565253e-06, 'epoch': 1.57}

 52%|█████▏    | 6991/13317 [1:12:02<1:04:50,  1.63it/s]
 53%|█████▎    | 6992/13317 [1:12:03<1:04:45,  1.63it/s]
                                                        
{'loss': 0.1743, 'grad_norm': 0.5845608115196228, 'learning_rate': 5.435234691194236e-06, 'epoch': 1.58}

 53%|█████▎    | 6992/13317 [1:12:03<1:04:45,  1.63it/s]
 53%|█████▎    | 6993/13317 [1:12:04<1:04:44,  1.63it/s]
                                                        
{'loss': 0.209, 'grad_norm': 0.5065545439720154, 'learning_rate': 5.433929015917988e-06, 'epoch': 1.58}

 53%|█████▎    | 6993/13317 [1:12:04<1:04:44,  1.63it/s]
 53%|█████▎    | 6994/13317 [1:12:04<1:04:49,  1.63it/s]
                                                        
{'loss': 0.2088, 'grad_norm': 0.5338917374610901, 'learning_rate': 5.432623310826222e-06, 'epoch': 1.58}

 53%|█████▎    | 6994/13317 [1:12:04<1:04:49,  1.63it/s]
 53%|█████▎    | 6995/13317 [1:12:05<1:04:44,  1.63it/s]
                                                        
{'loss': 0.2038, 'grad_norm': 0.5489493608474731, 'learning_rate': 5.4313175760086535e-06, 'epoch': 1.58}

 53%|█████▎    | 6995/13317 [1:12:05<1:04:44,  1.63it/s]
 53%|█████▎    | 6996/13317 [1:12:05<1:04:44,  1.63it/s]
                                                        
{'loss': 0.2161, 'grad_norm': 0.4225448668003082, 'learning_rate': 5.430011811555002e-06, 'epoch': 1.58}

 53%|█████▎    | 6996/13317 [1:12:06<1:04:44,  1.63it/s]
 53%|█████▎    | 6997/13317 [1:12:06<1:04:40,  1.63it/s]
                                                        
{'loss': 0.3159, 'grad_norm': 0.5894599556922913, 'learning_rate': 5.428706017554985e-06, 'epoch': 1.58}

 53%|█████▎    | 6997/13317 [1:12:06<1:04:40,  1.63it/s]
 53%|█████▎    | 6998/13317 [1:12:07<1:04:38,  1.63it/s]
                                                        
{'loss': 0.2377, 'grad_norm': 0.5460792779922485, 'learning_rate': 5.427400194098325e-06, 'epoch': 1.58}

 53%|█████▎    | 6998/13317 [1:12:07<1:04:38,  1.63it/s]
 53%|█████▎    | 6999/13317 [1:12:07<1:04:40,  1.63it/s]
                                                        
{'loss': 0.1848, 'grad_norm': 0.5396097898483276, 'learning_rate': 5.4260943412747454e-06, 'epoch': 1.58}

 53%|█████▎    | 6999/13317 [1:12:07<1:04:40,  1.63it/s]
 53%|█████▎    | 7000/13317 [1:12:08<1:04:37,  1.63it/s]
                                                        
{'loss': 0.2526, 'grad_norm': 0.5697560906410217, 'learning_rate': 5.4247884591739745e-06, 'epoch': 1.58}

 53%|█████▎    | 7000/13317 [1:12:08<1:04:37,  1.63it/s]

  0%|          | 0/13 [00:00<?, ?it/s]

 15%|█▌        | 2/13 [00:00<00:00, 13.14it/s]

 31%|███       | 4/13 [00:00<00:01,  8.35it/s]

 38%|███▊      | 5/13 [00:00<00:01,  7.76it/s]

 46%|████▌     | 6/13 [00:00<00:00,  7.40it/s]

 54%|█████▍    | 7/13 [00:00<00:00,  7.15it/s]

 62%|██████▏   | 8/13 [00:01<00:00,  6.99it/s]

 69%|██████▉   | 9/13 [00:01<00:00,  6.88it/s]

 77%|███████▋  | 10/13 [00:01<00:00,  6.81it/s]

 85%|████████▍ | 11/13 [00:01<00:00,  6.76it/s]

 92%|█████████▏| 12/13 [00:01<00:00,  6.73it/s]

100%|██████████| 13/13 [00:01<00:00,  6.75it/s]
                                                        

                                               
{'eval_loss': 0.5564929842948914, 'eval_runtime': 1.9589, 'eval_samples_per_second': 102.096, 'eval_steps_per_second': 6.636, 'epoch': 1.58}

 53%|█████▎    | 7000/13317 [1:12:10<1:04:37,  1.63it/s]

100%|██████████| 13/13 [00:01<00:00,  6.75it/s]

                                               
 53%|█████▎    | 7001/13317 [1:12:10<2:06:38,  1.20s/it]
                                                        
{'loss': 0.2314, 'grad_norm': 0.55849289894104, 'learning_rate': 5.423482547885737e-06, 'epoch': 1.58}

 53%|█████▎    | 7001/13317 [1:12:11<2:06:38,  1.20s/it]
 53%|█████▎    | 7002/13317 [1:12:11<1:48:03,  1.03s/it]
                                                        
{'loss': 0.2961, 'grad_norm': 0.7062312960624695, 'learning_rate': 5.422176607499768e-06, 'epoch': 1.58}

 53%|█████▎    | 7002/13317 [1:12:11<1:48:03,  1.03s/it]
 53%|█████▎    | 7003/13317 [1:12:12<1:34:59,  1.11it/s]
                                                        
{'loss': 0.1898, 'grad_norm': 0.4979807436466217, 'learning_rate': 5.420870638105794e-06, 'epoch': 1.58}

 53%|█████▎    | 7003/13317 [1:12:12<1:34:59,  1.11it/s]
 53%|█████▎    | 7004/13317 [1:12:12<1:25:54,  1.22it/s]
                                                        
{'loss': 0.1544, 'grad_norm': 0.429878830909729, 'learning_rate': 5.419564639793551e-06, 'epoch': 1.58}

 53%|█████▎    | 7004/13317 [1:12:12<1:25:54,  1.22it/s]
 53%|█████▎    | 7005/13317 [1:12:13<1:19:28,  1.32it/s]
                                                        
{'loss': 0.2161, 'grad_norm': 0.6393535733222961, 'learning_rate': 5.418258612652774e-06, 'epoch': 1.58}

 53%|█████▎    | 7005/13317 [1:12:13<1:19:28,  1.32it/s]
 53%|█████▎    | 7006/13317 [1:12:14<1:14:58,  1.40it/s]
                                                        
{'loss': 0.2601, 'grad_norm': 0.587588906288147, 'learning_rate': 5.416952556773203e-06, 'epoch': 1.58}

 53%|█████▎    | 7006/13317 [1:12:14<1:14:58,  1.40it/s]
 53%|█████▎    | 7007/13317 [1:12:14<1:11:48,  1.46it/s]
                                                        
{'loss': 0.157, 'grad_norm': 0.4822555482387543, 'learning_rate': 5.415646472244577e-06, 'epoch': 1.58}

 53%|█████▎    | 7007/13317 [1:12:14<1:11:48,  1.46it/s]
 53%|█████▎    | 7008/13317 [1:12:15<1:09:37,  1.51it/s]
                                                        
{'loss': 0.1816, 'grad_norm': 0.5129657983779907, 'learning_rate': 5.4143403591566346e-06, 'epoch': 1.58}

 53%|█████▎    | 7008/13317 [1:12:15<1:09:37,  1.51it/s]
 53%|█████▎    | 7009/13317 [1:12:15<1:08:08,  1.54it/s]
                                                        
{'loss': 0.197, 'grad_norm': 0.5371457934379578, 'learning_rate': 5.413034217599125e-06, 'epoch': 1.58}

 53%|█████▎    | 7009/13317 [1:12:16<1:08:08,  1.54it/s]
 53%|█████▎    | 7010/13317 [1:12:16<1:07:02,  1.57it/s]
                                                        
{'loss': 0.2365, 'grad_norm': 0.5949476361274719, 'learning_rate': 5.411728047661789e-06, 'epoch': 1.58}

 53%|█████▎    | 7010/13317 [1:12:16<1:07:02,  1.57it/s]
 53%|█████▎    | 7011/13317 [1:12:17<1:06:15,  1.59it/s]
                                                        
{'loss': 0.1986, 'grad_norm': 0.47350218892097473, 'learning_rate': 5.4104218494343795e-06, 'epoch': 1.58}

 53%|█████▎    | 7011/13317 [1:12:17<1:06:15,  1.59it/s]
 53%|█████▎    | 7012/13317 [1:12:17<1:05:46,  1.60it/s]
                                                        
{'loss': 0.2676, 'grad_norm': 0.6843650937080383, 'learning_rate': 5.4091156230066425e-06, 'epoch': 1.58}

 53%|█████▎    | 7012/13317 [1:12:17<1:05:46,  1.60it/s]
 53%|█████▎    | 7013/13317 [1:12:18<1:05:25,  1.61it/s]
                                                        
{'loss': 0.2461, 'grad_norm': 0.6256868839263916, 'learning_rate': 5.407809368468329e-06, 'epoch': 1.58}

 53%|█████▎    | 7013/13317 [1:12:18<1:05:25,  1.61it/s]
 53%|█████▎    | 7014/13317 [1:12:18<1:05:12,  1.61it/s]
                                                        
{'loss': 0.1734, 'grad_norm': 0.5077915787696838, 'learning_rate': 5.406503085909196e-06, 'epoch': 1.58}

 53%|█████▎    | 7014/13317 [1:12:19<1:05:12,  1.61it/s]
 53%|█████▎    | 7015/13317 [1:12:19<1:04:59,  1.62it/s]
                                                        
{'loss': 0.1973, 'grad_norm': 0.5333491563796997, 'learning_rate': 5.405196775418995e-06, 'epoch': 1.58}

 53%|█████▎    | 7015/13317 [1:12:19<1:04:59,  1.62it/s]
 53%|█████▎    | 7016/13317 [1:12:20<1:04:53,  1.62it/s]
                                                        
{'loss': 0.1893, 'grad_norm': 0.4473143219947815, 'learning_rate': 5.403890437087487e-06, 'epoch': 1.58}

 53%|█████▎    | 7016/13317 [1:12:20<1:04:53,  1.62it/s]
 53%|█████▎    | 7017/13317 [1:12:20<1:04:45,  1.62it/s]
                                                        
{'loss': 0.204, 'grad_norm': 0.5507468581199646, 'learning_rate': 5.402584071004427e-06, 'epoch': 1.58}

 53%|█████▎    | 7017/13317 [1:12:20<1:04:45,  1.62it/s]
 53%|█████▎    | 7018/13317 [1:12:21<1:04:37,  1.62it/s]
                                                        
{'loss': 0.2076, 'grad_norm': 0.5298529863357544, 'learning_rate': 5.401277677259581e-06, 'epoch': 1.58}

 53%|█████▎    | 7018/13317 [1:12:21<1:04:37,  1.62it/s]
 53%|█████▎    | 7019/13317 [1:12:22<1:04:39,  1.62it/s]
                                                        
{'loss': 0.2077, 'grad_norm': 0.493604838848114, 'learning_rate': 5.399971255942708e-06, 'epoch': 1.58}

 53%|█████▎    | 7019/13317 [1:12:22<1:04:39,  1.62it/s]
 53%|█████▎    | 7020/13317 [1:12:22<1:04:35,  1.62it/s]
                                                        
{'loss': 0.1965, 'grad_norm': 0.5600207448005676, 'learning_rate': 5.398664807143577e-06, 'epoch': 1.58}

 53%|█████▎    | 7020/13317 [1:12:22<1:04:35,  1.62it/s]
 53%|█████▎    | 7021/13317 [1:12:23<1:04:39,  1.62it/s]
                                                        
{'loss': 0.1875, 'grad_norm': 0.4775163531303406, 'learning_rate': 5.39735833095195e-06, 'epoch': 1.58}

 53%|█████▎    | 7021/13317 [1:12:23<1:04:39,  1.62it/s]
 53%|█████▎    | 7022/13317 [1:12:23<1:04:35,  1.62it/s]
                                                        
{'loss': 0.2416, 'grad_norm': 0.6449854969978333, 'learning_rate': 5.3960518274576e-06, 'epoch': 1.58}

 53%|█████▎    | 7022/13317 [1:12:24<1:04:35,  1.62it/s]
 53%|█████▎    | 7023/13317 [1:12:24<1:04:29,  1.63it/s]
                                                        
{'loss': 0.226, 'grad_norm': 0.5418733954429626, 'learning_rate': 5.394745296750297e-06, 'epoch': 1.58}

 53%|█████▎    | 7023/13317 [1:12:24<1:04:29,  1.63it/s]
 53%|█████▎    | 7024/13317 [1:12:25<1:04:32,  1.62it/s]
                                                        
{'loss': 0.2151, 'grad_norm': 0.5803484916687012, 'learning_rate': 5.3934387389198105e-06, 'epoch': 1.58}

 53%|█████▎    | 7024/13317 [1:12:25<1:04:32,  1.62it/s]
 53%|█████▎    | 7025/13317 [1:12:25<1:04:29,  1.63it/s]
                                                        
{'loss': 0.1879, 'grad_norm': 0.553011953830719, 'learning_rate': 5.392132154055918e-06, 'epoch': 1.58}

 53%|█████▎    | 7025/13317 [1:12:25<1:04:29,  1.63it/s]
 53%|█████▎    | 7026/13317 [1:12:26<1:04:26,  1.63it/s]
                                                        
{'loss': 0.2276, 'grad_norm': 0.5378259420394897, 'learning_rate': 5.390825542248393e-06, 'epoch': 1.58}

 53%|█████▎    | 7026/13317 [1:12:26<1:04:26,  1.63it/s]
 53%|█████▎    | 7027/13317 [1:12:26<1:04:24,  1.63it/s]
                                                        
{'loss': 0.1452, 'grad_norm': 0.4225045442581177, 'learning_rate': 5.389518903587016e-06, 'epoch': 1.58}

 53%|█████▎    | 7027/13317 [1:12:27<1:04:24,  1.63it/s]
 53%|█████▎    | 7028/13317 [1:12:27<1:04:22,  1.63it/s]
                                                        
{'loss': 0.1996, 'grad_norm': 0.5903758406639099, 'learning_rate': 5.388212238161567e-06, 'epoch': 1.58}

 53%|█████▎    | 7028/13317 [1:12:27<1:04:22,  1.63it/s]
 53%|█████▎    | 7029/13317 [1:12:28<1:04:26,  1.63it/s]
                                                        
{'loss': 0.2045, 'grad_norm': 0.5270705819129944, 'learning_rate': 5.386905546061827e-06, 'epoch': 1.58}

 53%|█████▎    | 7029/13317 [1:12:28<1:04:26,  1.63it/s]
 53%|█████▎    | 7030/13317 [1:12:28<1:04:24,  1.63it/s]
                                                        
{'loss': 0.1728, 'grad_norm': 0.45359036326408386, 'learning_rate': 5.385598827377578e-06, 'epoch': 1.58}

 53%|█████▎    | 7030/13317 [1:12:28<1:04:24,  1.63it/s]
 53%|█████▎    | 7031/13317 [1:12:29<1:04:22,  1.63it/s]
                                                        
{'loss': 0.1874, 'grad_norm': 0.44838741421699524, 'learning_rate': 5.384292082198608e-06, 'epoch': 1.58}

 53%|█████▎    | 7031/13317 [1:12:29<1:04:22,  1.63it/s]
 53%|█████▎    | 7032/13317 [1:12:30<1:04:19,  1.63it/s]
                                                        
{'loss': 0.2002, 'grad_norm': 0.522225022315979, 'learning_rate': 5.382985310614701e-06, 'epoch': 1.58}

 53%|█████▎    | 7032/13317 [1:12:30<1:04:19,  1.63it/s]
 53%|█████▎    | 7033/13317 [1:12:30<1:04:19,  1.63it/s]
                                                        
{'loss': 0.2269, 'grad_norm': 0.529026210308075, 'learning_rate': 5.38167851271565e-06, 'epoch': 1.58}

 53%|█████▎    | 7033/13317 [1:12:30<1:04:19,  1.63it/s]
 53%|█████▎    | 7034/13317 [1:12:31<1:04:22,  1.63it/s]
                                                        
{'loss': 0.3064, 'grad_norm': 0.5525975227355957, 'learning_rate': 5.3803716885912425e-06, 'epoch': 1.58}

 53%|█████▎    | 7034/13317 [1:12:31<1:04:22,  1.63it/s]
 53%|█████▎    | 7035/13317 [1:12:31<1:04:21,  1.63it/s]
                                                        
{'loss': 0.1822, 'grad_norm': 0.4741194546222687, 'learning_rate': 5.3790648383312725e-06, 'epoch': 1.58}

 53%|█████▎    | 7035/13317 [1:12:31<1:04:21,  1.63it/s]
 53%|█████▎    | 7036/13317 [1:12:32<1:04:19,  1.63it/s]
                                                        
{'loss': 0.2123, 'grad_norm': 0.5054454207420349, 'learning_rate': 5.377757962025537e-06, 'epoch': 1.59}

 53%|█████▎    | 7036/13317 [1:12:32<1:04:19,  1.63it/s]
 53%|█████▎    | 7037/13317 [1:12:33<1:04:21,  1.63it/s]
                                                        
{'loss': 0.1768, 'grad_norm': 0.4794967770576477, 'learning_rate': 5.376451059763826e-06, 'epoch': 1.59}

 53%|█████▎    | 7037/13317 [1:12:33<1:04:21,  1.63it/s]
 53%|█████▎    | 7038/13317 [1:12:33<1:04:18,  1.63it/s]
                                                        
{'loss': 0.2012, 'grad_norm': 0.5524386167526245, 'learning_rate': 5.375144131635944e-06, 'epoch': 1.59}

 53%|█████▎    | 7038/13317 [1:12:33<1:04:18,  1.63it/s]
 53%|█████▎    | 7039/13317 [1:12:34<1:04:19,  1.63it/s]
                                                        
{'loss': 0.1586, 'grad_norm': 0.45063552260398865, 'learning_rate': 5.373837177731688e-06, 'epoch': 1.59}

 53%|█████▎    | 7039/13317 [1:12:34<1:04:19,  1.63it/s]
 53%|█████▎    | 7040/13317 [1:12:34<1:04:15,  1.63it/s]
                                                        
{'loss': 0.2303, 'grad_norm': 0.6140782833099365, 'learning_rate': 5.3725301981408585e-06, 'epoch': 1.59}

 53%|█████▎    | 7040/13317 [1:12:35<1:04:15,  1.63it/s]
 53%|█████▎    | 7041/13317 [1:12:35<1:04:18,  1.63it/s]
                                                        
{'loss': 0.1515, 'grad_norm': 0.46948865056037903, 'learning_rate': 5.371223192953261e-06, 'epoch': 1.59}

 53%|█████▎    | 7041/13317 [1:12:35<1:04:18,  1.63it/s]
 53%|█████▎    | 7042/13317 [1:12:36<1:04:17,  1.63it/s]
                                                        
{'loss': 0.1786, 'grad_norm': 0.5069290399551392, 'learning_rate': 5.369916162258698e-06, 'epoch': 1.59}

 53%|█████▎    | 7042/13317 [1:12:36<1:04:17,  1.63it/s]
 53%|█████▎    | 7043/13317 [1:12:36<1:04:13,  1.63it/s]
                                                        
{'loss': 0.1872, 'grad_norm': 0.5318316221237183, 'learning_rate': 5.36860910614698e-06, 'epoch': 1.59}

 53%|█████▎    | 7043/13317 [1:12:36<1:04:13,  1.63it/s]
 53%|█████▎    | 7044/13317 [1:12:37<1:04:15,  1.63it/s]
                                                        
{'loss': 0.2306, 'grad_norm': 0.6466231346130371, 'learning_rate': 5.367302024707911e-06, 'epoch': 1.59}

 53%|█████▎    | 7044/13317 [1:12:37<1:04:15,  1.63it/s]
 53%|█████▎    | 7045/13317 [1:12:38<1:04:11,  1.63it/s]
                                                        
{'loss': 0.1881, 'grad_norm': 0.5173802375793457, 'learning_rate': 5.365994918031307e-06, 'epoch': 1.59}

 53%|█████▎    | 7045/13317 [1:12:38<1:04:11,  1.63it/s]
 53%|█████▎    | 7046/13317 [1:12:38<1:04:10,  1.63it/s]
                                                        
{'loss': 0.2641, 'grad_norm': 0.634314775466919, 'learning_rate': 5.364687786206974e-06, 'epoch': 1.59}

 53%|█████▎    | 7046/13317 [1:12:38<1:04:10,  1.63it/s]
 53%|█████▎    | 7047/13317 [1:12:39<1:04:10,  1.63it/s]
                                                        
{'loss': 0.1914, 'grad_norm': 0.5248712301254272, 'learning_rate': 5.3633806293247305e-06, 'epoch': 1.59}

 53%|█████▎    | 7047/13317 [1:12:39<1:04:10,  1.63it/s]
 53%|█████▎    | 7048/13317 [1:12:39<1:04:07,  1.63it/s]
                                                        
{'loss': 0.2177, 'grad_norm': 0.5424090623855591, 'learning_rate': 5.362073447474388e-06, 'epoch': 1.59}

 53%|█████▎    | 7048/13317 [1:12:39<1:04:07,  1.63it/s]
 53%|█████▎    | 7049/13317 [1:12:40<1:04:14,  1.63it/s]
                                                        
{'loss': 0.1752, 'grad_norm': 0.4759724736213684, 'learning_rate': 5.360766240745768e-06, 'epoch': 1.59}

 53%|█████▎    | 7049/13317 [1:12:40<1:04:14,  1.63it/s]
 53%|█████▎    | 7050/13317 [1:12:41<1:04:09,  1.63it/s]
                                                        
{'loss': 0.1936, 'grad_norm': 0.5365151762962341, 'learning_rate': 5.359459009228688e-06, 'epoch': 1.59}

 53%|█████▎    | 7050/13317 [1:12:41<1:04:09,  1.63it/s]
 53%|█████▎    | 7051/13317 [1:12:41<1:04:12,  1.63it/s]
                                                        
{'loss': 0.2648, 'grad_norm': 0.47000738978385925, 'learning_rate': 5.358151753012967e-06, 'epoch': 1.59}

 53%|█████▎    | 7051/13317 [1:12:41<1:04:12,  1.63it/s]
 53%|█████▎    | 7052/13317 [1:12:42<1:04:13,  1.63it/s]
                                                        
{'loss': 0.2151, 'grad_norm': 0.4998511075973511, 'learning_rate': 5.356844472188428e-06, 'epoch': 1.59}

 53%|█████▎    | 7052/13317 [1:12:42<1:04:13,  1.63it/s]
 53%|█████▎    | 7053/13317 [1:12:42<1:04:07,  1.63it/s]
                                                        
{'loss': 0.1715, 'grad_norm': 0.5801327228546143, 'learning_rate': 5.355537166844896e-06, 'epoch': 1.59}

 53%|█████▎    | 7053/13317 [1:12:43<1:04:07,  1.63it/s]
 53%|█████▎    | 7054/13317 [1:12:43<1:04:09,  1.63it/s]
                                                        
{'loss': 0.2061, 'grad_norm': 0.49257200956344604, 'learning_rate': 5.354229837072196e-06, 'epoch': 1.59}

 53%|█████▎    | 7054/13317 [1:12:43<1:04:09,  1.63it/s]
 53%|█████▎    | 7055/13317 [1:12:44<1:04:09,  1.63it/s]
                                                        
{'loss': 0.2006, 'grad_norm': 0.5139102935791016, 'learning_rate': 5.352922482960156e-06, 'epoch': 1.59}

 53%|█████▎    | 7055/13317 [1:12:44<1:04:09,  1.63it/s]
 53%|█████▎    | 7056/13317 [1:12:44<1:04:09,  1.63it/s]
                                                        
{'loss': 0.1644, 'grad_norm': 0.49044525623321533, 'learning_rate': 5.351615104598603e-06, 'epoch': 1.59}

 53%|█████▎    | 7056/13317 [1:12:44<1:04:09,  1.63it/s]
 53%|█████▎    | 7057/13317 [1:12:45<1:04:13,  1.62it/s]
                                                        
{'loss': 0.2184, 'grad_norm': 0.5710557103157043, 'learning_rate': 5.350307702077371e-06, 'epoch': 1.59}

 53%|█████▎    | 7057/13317 [1:12:45<1:04:13,  1.62it/s]
 53%|█████▎    | 7058/13317 [1:12:45<1:04:08,  1.63it/s]
                                                        
{'loss': 0.1555, 'grad_norm': 0.431494802236557, 'learning_rate': 5.349000275486291e-06, 'epoch': 1.59}

 53%|█████▎    | 7058/13317 [1:12:46<1:04:08,  1.63it/s]
 53%|█████▎    | 7059/13317 [1:12:46<1:04:26,  1.62it/s]
                                                        
{'loss': 0.1744, 'grad_norm': 0.4732770621776581, 'learning_rate': 5.347692824915196e-06, 'epoch': 1.59}

 53%|█████▎    | 7059/13317 [1:12:46<1:04:26,  1.62it/s]
 53%|█████▎    | 7060/13317 [1:12:47<1:04:17,  1.62it/s]
                                                        
{'loss': 0.1726, 'grad_norm': 0.5089778304100037, 'learning_rate': 5.346385350453922e-06, 'epoch': 1.59}

 53%|█████▎    | 7060/13317 [1:12:47<1:04:17,  1.62it/s]
 53%|█████▎    | 7061/13317 [1:12:47<1:04:15,  1.62it/s]
                                                        
{'loss': 0.1973, 'grad_norm': 0.5378944873809814, 'learning_rate': 5.345077852192307e-06, 'epoch': 1.59}

 53%|█████▎    | 7061/13317 [1:12:47<1:04:15,  1.62it/s]
 53%|█████▎    | 7062/13317 [1:12:48<1:04:13,  1.62it/s]
                                                        
{'loss': 0.2187, 'grad_norm': 0.6032872796058655, 'learning_rate': 5.343770330220191e-06, 'epoch': 1.59}

 53%|█████▎    | 7062/13317 [1:12:48<1:04:13,  1.62it/s]
 53%|█████▎    | 7063/13317 [1:12:49<1:04:06,  1.63it/s]
                                                        
{'loss': 0.2059, 'grad_norm': 0.5692281723022461, 'learning_rate': 5.342462784627412e-06, 'epoch': 1.59}

 53%|█████▎    | 7063/13317 [1:12:49<1:04:06,  1.63it/s]
 53%|█████▎    | 7064/13317 [1:12:49<1:04:09,  1.62it/s]
                                                        
{'loss': 0.1947, 'grad_norm': 0.5177699327468872, 'learning_rate': 5.341155215503814e-06, 'epoch': 1.59}

 53%|█████▎    | 7064/13317 [1:12:49<1:04:09,  1.62it/s]
 53%|█████▎    | 7065/13317 [1:12:50<1:04:05,  1.63it/s]
                                                        
{'loss': 0.1886, 'grad_norm': 0.46660861372947693, 'learning_rate': 5.339847622939241e-06, 'epoch': 1.59}

 53%|█████▎    | 7065/13317 [1:12:50<1:04:05,  1.63it/s]
 53%|█████▎    | 7066/13317 [1:12:50<1:04:09,  1.62it/s]
                                                        
{'loss': 0.2082, 'grad_norm': 0.5252666473388672, 'learning_rate': 5.338540007023538e-06, 'epoch': 1.59}

 53%|█████▎    | 7066/13317 [1:12:51<1:04:09,  1.62it/s]
 53%|█████▎    | 7067/13317 [1:12:51<1:04:08,  1.62it/s]
                                                        
{'loss': 0.2181, 'grad_norm': 0.5837341547012329, 'learning_rate': 5.337232367846552e-06, 'epoch': 1.59}

 53%|█████▎    | 7067/13317 [1:12:51<1:04:08,  1.62it/s]
 53%|█████▎    | 7068/13317 [1:12:52<1:04:02,  1.63it/s]
                                                        
{'loss': 0.1938, 'grad_norm': 0.5076927542686462, 'learning_rate': 5.335924705498131e-06, 'epoch': 1.59}

 53%|█████▎    | 7068/13317 [1:12:52<1:04:02,  1.63it/s]
 53%|█████▎    | 7069/13317 [1:12:52<1:04:05,  1.62it/s]
                                                        
{'loss': 0.2472, 'grad_norm': 0.617479145526886, 'learning_rate': 5.334617020068126e-06, 'epoch': 1.59}

 53%|█████▎    | 7069/13317 [1:12:52<1:04:05,  1.62it/s]
 53%|█████▎    | 7070/13317 [1:12:53<1:04:01,  1.63it/s]
                                                        
{'loss': 0.2676, 'grad_norm': 0.571104109287262, 'learning_rate': 5.333309311646389e-06, 'epoch': 1.59}

 53%|█████▎    | 7070/13317 [1:12:53<1:04:01,  1.63it/s]
 53%|█████▎    | 7071/13317 [1:12:53<1:04:03,  1.63it/s]
                                                        
{'loss': 0.1857, 'grad_norm': 0.5364582538604736, 'learning_rate': 5.332001580322773e-06, 'epoch': 1.59}

 53%|█████▎    | 7071/13317 [1:12:54<1:04:03,  1.63it/s]
 53%|█████▎    | 7072/13317 [1:12:54<1:04:00,  1.63it/s]
                                                        
{'loss': 0.2384, 'grad_norm': 0.5031870007514954, 'learning_rate': 5.330693826187134e-06, 'epoch': 1.59}

 53%|█████▎    | 7072/13317 [1:12:54<1:04:00,  1.63it/s]
 53%|█████▎    | 7073/13317 [1:12:55<1:03:54,  1.63it/s]
                                                        
{'loss': 0.2026, 'grad_norm': 0.5622990131378174, 'learning_rate': 5.329386049329327e-06, 'epoch': 1.59}

 53%|█████▎    | 7073/13317 [1:12:55<1:03:54,  1.63it/s]
 53%|█████▎    | 7074/13317 [1:12:55<1:03:56,  1.63it/s]
                                                        
{'loss': 0.169, 'grad_norm': 0.5038515329360962, 'learning_rate': 5.328078249839211e-06, 'epoch': 1.59}

 53%|█████▎    | 7074/13317 [1:12:55<1:03:56,  1.63it/s]
 53%|█████▎    | 7075/13317 [1:12:56<1:03:54,  1.63it/s]
                                                        
{'loss': 0.227, 'grad_norm': 0.6585351824760437, 'learning_rate': 5.326770427806645e-06, 'epoch': 1.59}

 53%|█████▎    | 7075/13317 [1:12:56<1:03:54,  1.63it/s]
 53%|█████▎    | 7076/13317 [1:12:57<1:03:56,  1.63it/s]
                                                        
{'loss': 0.1965, 'grad_norm': 0.5509839057922363, 'learning_rate': 5.32546258332149e-06, 'epoch': 1.59}

 53%|█████▎    | 7076/13317 [1:12:57<1:03:56,  1.63it/s]
 53%|█████▎    | 7077/13317 [1:12:57<1:03:53,  1.63it/s]
                                                        
{'loss': 0.2043, 'grad_norm': 0.45985499024391174, 'learning_rate': 5.324154716473612e-06, 'epoch': 1.59}

 53%|█████▎    | 7077/13317 [1:12:57<1:03:53,  1.63it/s]
 53%|█████▎    | 7078/13317 [1:12:58<1:03:51,  1.63it/s]
                                                        
{'loss': 0.2349, 'grad_norm': 0.6036573648452759, 'learning_rate': 5.32284682735287e-06, 'epoch': 1.59}

 53%|█████▎    | 7078/13317 [1:12:58<1:03:51,  1.63it/s]
 53%|█████▎    | 7079/13317 [1:12:58<1:03:57,  1.63it/s]
                                                        
{'loss': 0.1735, 'grad_norm': 0.5403165817260742, 'learning_rate': 5.321538916049135e-06, 'epoch': 1.59}

 53%|█████▎    | 7079/13317 [1:12:59<1:03:57,  1.63it/s]
 53%|█████▎    | 7080/13317 [1:12:59<1:03:53,  1.63it/s]
                                                        
{'loss': 0.1926, 'grad_norm': 0.5107944011688232, 'learning_rate': 5.32023098265227e-06, 'epoch': 1.59}

 53%|█████▎    | 7080/13317 [1:12:59<1:03:53,  1.63it/s]
 53%|█████▎    | 7081/13317 [1:13:00<1:03:55,  1.63it/s]
                                                        
{'loss': 0.1924, 'grad_norm': 0.522731602191925, 'learning_rate': 5.318923027252146e-06, 'epoch': 1.6}

 53%|█████▎    | 7081/13317 [1:13:00<1:03:55,  1.63it/s]
 53%|█████▎    | 7082/13317 [1:13:00<1:03:51,  1.63it/s]
                                                        
{'loss': 0.1991, 'grad_norm': 0.5081942081451416, 'learning_rate': 5.317615049938635e-06, 'epoch': 1.6}

 53%|█████▎    | 7082/13317 [1:13:00<1:03:51,  1.63it/s]
 53%|█████▎    | 7083/13317 [1:13:01<1:03:49,  1.63it/s]
                                                        
{'loss': 0.263, 'grad_norm': 0.5451730489730835, 'learning_rate': 5.316307050801606e-06, 'epoch': 1.6}

 53%|█████▎    | 7083/13317 [1:13:01<1:03:49,  1.63it/s]
 53%|█████▎    | 7084/13317 [1:13:01<1:03:53,  1.63it/s]
                                                        
{'loss': 0.2853, 'grad_norm': 0.6687849164009094, 'learning_rate': 5.314999029930933e-06, 'epoch': 1.6}

 53%|█████▎    | 7084/13317 [1:13:02<1:03:53,  1.63it/s]
 53%|█████▎    | 7085/13317 [1:13:02<1:03:49,  1.63it/s]
                                                        
{'loss': 0.2259, 'grad_norm': 0.5731313824653625, 'learning_rate': 5.313690987416492e-06, 'epoch': 1.6}

 53%|█████▎    | 7085/13317 [1:13:02<1:03:49,  1.63it/s]
 53%|█████▎    | 7086/13317 [1:13:03<1:03:48,  1.63it/s]
                                                        
{'loss': 0.2059, 'grad_norm': 0.5196858048439026, 'learning_rate': 5.312382923348159e-06, 'epoch': 1.6}

 53%|█████▎    | 7086/13317 [1:13:03<1:03:48,  1.63it/s]
 53%|█████▎    | 7087/13317 [1:13:03<1:03:49,  1.63it/s]
                                                        
{'loss': 0.1793, 'grad_norm': 0.48271551728248596, 'learning_rate': 5.31107483781581e-06, 'epoch': 1.6}

 53%|█████▎    | 7087/13317 [1:13:03<1:03:49,  1.63it/s]
 53%|█████▎    | 7088/13317 [1:13:04<1:03:47,  1.63it/s]
                                                        
{'loss': 0.1959, 'grad_norm': 0.48834940791130066, 'learning_rate': 5.309766730909328e-06, 'epoch': 1.6}

 53%|█████▎    | 7088/13317 [1:13:04<1:03:47,  1.63it/s]
 53%|█████▎    | 7089/13317 [1:13:05<1:03:50,  1.63it/s]
                                                        
{'loss': 0.2929, 'grad_norm': 0.5845345854759216, 'learning_rate': 5.30845860271859e-06, 'epoch': 1.6}

 53%|█████▎    | 7089/13317 [1:13:05<1:03:50,  1.63it/s]
 53%|█████▎    | 7090/13317 [1:13:05<1:03:49,  1.63it/s]
                                                        
{'loss': 0.2202, 'grad_norm': 0.5705515146255493, 'learning_rate': 5.307150453333483e-06, 'epoch': 1.6}

 53%|█████▎    | 7090/13317 [1:13:05<1:03:49,  1.63it/s]
 53%|█████▎    | 7091/13317 [1:13:06<1:03:48,  1.63it/s]
                                                        
{'loss': 0.1665, 'grad_norm': 0.4791051745414734, 'learning_rate': 5.305842282843884e-06, 'epoch': 1.6}

 53%|█████▎    | 7091/13317 [1:13:06<1:03:48,  1.63it/s]
 53%|█████▎    | 7092/13317 [1:13:06<1:03:46,  1.63it/s]
                                                        
{'loss': 0.2194, 'grad_norm': 0.5745377540588379, 'learning_rate': 5.304534091339684e-06, 'epoch': 1.6}

 53%|█████▎    | 7092/13317 [1:13:07<1:03:46,  1.63it/s]
 53%|█████▎    | 7093/13317 [1:13:07<1:03:43,  1.63it/s]
                                                        
{'loss': 0.1627, 'grad_norm': 0.4274597465991974, 'learning_rate': 5.303225878910766e-06, 'epoch': 1.6}

 53%|█████▎    | 7093/13317 [1:13:07<1:03:43,  1.63it/s]
 53%|█████▎    | 7094/13317 [1:13:08<1:03:48,  1.63it/s]
                                                        
{'loss': 0.2005, 'grad_norm': 0.5245261192321777, 'learning_rate': 5.301917645647021e-06, 'epoch': 1.6}

 53%|█████▎    | 7094/13317 [1:13:08<1:03:48,  1.63it/s]
 53%|█████▎    | 7095/13317 [1:13:08<1:03:43,  1.63it/s]
                                                        
{'loss': 0.2767, 'grad_norm': 0.6359529495239258, 'learning_rate': 5.300609391638336e-06, 'epoch': 1.6}

 53%|█████▎    | 7095/13317 [1:13:08<1:03:43,  1.63it/s]
 53%|█████▎    | 7096/13317 [1:13:09<1:03:43,  1.63it/s]
                                                        
{'loss': 0.188, 'grad_norm': 0.48313310742378235, 'learning_rate': 5.299301116974603e-06, 'epoch': 1.6}

 53%|█████▎    | 7096/13317 [1:13:09<1:03:43,  1.63it/s]
 53%|█████▎    | 7097/13317 [1:13:09<1:03:46,  1.63it/s]
                                                        
{'loss': 0.2127, 'grad_norm': 1.691925048828125, 'learning_rate': 5.297992821745715e-06, 'epoch': 1.6}

 53%|█████▎    | 7097/13317 [1:13:10<1:03:46,  1.63it/s]
 53%|█████▎    | 7098/13317 [1:13:10<1:03:42,  1.63it/s]
                                                        
{'loss': 0.199, 'grad_norm': 0.5107685327529907, 'learning_rate': 5.296684506041564e-06, 'epoch': 1.6}

 53%|█████▎    | 7098/13317 [1:13:10<1:03:42,  1.63it/s]
 53%|█████▎    | 7099/13317 [1:13:11<1:03:46,  1.62it/s]
                                                        
{'loss': 0.24, 'grad_norm': 0.4912128448486328, 'learning_rate': 5.295376169952047e-06, 'epoch': 1.6}

 53%|█████▎    | 7099/13317 [1:13:11<1:03:46,  1.62it/s]
 53%|█████▎    | 7100/13317 [1:13:11<1:03:45,  1.63it/s]
                                                        
{'loss': 0.1613, 'grad_norm': 0.43949684500694275, 'learning_rate': 5.294067813567058e-06, 'epoch': 1.6}

 53%|█████▎    | 7100/13317 [1:13:11<1:03:45,  1.63it/s]
 53%|█████▎    | 7101/13317 [1:13:12<1:03:57,  1.62it/s]
                                                        
{'loss': 0.176, 'grad_norm': 0.4595821499824524, 'learning_rate': 5.292759436976499e-06, 'epoch': 1.6}

 53%|█████▎    | 7101/13317 [1:13:12<1:03:57,  1.62it/s]
 53%|█████▎    | 7102/13317 [1:13:13<1:03:52,  1.62it/s]
                                                        
{'loss': 0.1558, 'grad_norm': 0.4163760542869568, 'learning_rate': 5.291451040270265e-06, 'epoch': 1.6}

 53%|█████▎    | 7102/13317 [1:13:13<1:03:52,  1.62it/s]
 53%|█████▎    | 7103/13317 [1:13:13<1:03:46,  1.62it/s]
                                                        
{'loss': 0.1555, 'grad_norm': 0.42114391922950745, 'learning_rate': 5.2901426235382595e-06, 'epoch': 1.6}

 53%|█████▎    | 7103/13317 [1:13:13<1:03:46,  1.62it/s]
 53%|█████▎    | 7104/13317 [1:13:14<1:03:46,  1.62it/s]
                                                        
{'loss': 0.1825, 'grad_norm': 0.5047652125358582, 'learning_rate': 5.288834186870384e-06, 'epoch': 1.6}

 53%|█████▎    | 7104/13317 [1:13:14<1:03:46,  1.62it/s]
 53%|█████▎    | 7105/13317 [1:13:14<1:03:42,  1.63it/s]
                                                        
{'loss': 0.1932, 'grad_norm': 0.4163365662097931, 'learning_rate': 5.28752573035654e-06, 'epoch': 1.6}

 53%|█████▎    | 7105/13317 [1:13:15<1:03:42,  1.63it/s]
 53%|█████▎    | 7106/13317 [1:13:15<1:03:43,  1.62it/s]
                                                        
{'loss': 0.2101, 'grad_norm': 0.537196159362793, 'learning_rate': 5.286217254086635e-06, 'epoch': 1.6}

 53%|█████▎    | 7106/13317 [1:13:15<1:03:43,  1.62it/s]
 53%|█████▎    | 7107/13317 [1:13:16<1:03:39,  1.63it/s]
                                                        
{'loss': 0.1934, 'grad_norm': 0.48114392161369324, 'learning_rate': 5.284908758150574e-06, 'epoch': 1.6}

 53%|█████▎    | 7107/13317 [1:13:16<1:03:39,  1.63it/s]
 53%|█████▎    | 7108/13317 [1:13:16<1:03:35,  1.63it/s]
                                                        
{'loss': 0.2509, 'grad_norm': 0.5978216528892517, 'learning_rate': 5.283600242638265e-06, 'epoch': 1.6}

 53%|█████▎    | 7108/13317 [1:13:16<1:03:35,  1.63it/s]
 53%|█████▎    | 7109/13317 [1:13:17<1:03:39,  1.63it/s]
                                                        
{'loss': 0.1835, 'grad_norm': 0.5358274579048157, 'learning_rate': 5.2822917076396165e-06, 'epoch': 1.6}

 53%|█████▎    | 7109/13317 [1:13:17<1:03:39,  1.63it/s]
 53%|█████▎    | 7110/13317 [1:13:17<1:03:34,  1.63it/s]
                                                        
{'loss': 0.2093, 'grad_norm': 0.5239014029502869, 'learning_rate': 5.280983153244538e-06, 'epoch': 1.6}

 53%|█████▎    | 7110/13317 [1:13:18<1:03:34,  1.63it/s]
 53%|█████▎    | 7111/13317 [1:13:18<1:03:36,  1.63it/s]
                                                        
{'loss': 0.2182, 'grad_norm': 0.524803876876831, 'learning_rate': 5.279674579542942e-06, 'epoch': 1.6}

 53%|█████▎    | 7111/13317 [1:13:18<1:03:36,  1.63it/s]
 53%|█████▎    | 7112/13317 [1:13:19<1:03:34,  1.63it/s]
                                                        
{'loss': 0.2236, 'grad_norm': 0.6092434525489807, 'learning_rate': 5.278365986624743e-06, 'epoch': 1.6}

 53%|█████▎    | 7112/13317 [1:13:19<1:03:34,  1.63it/s]
 53%|█████▎    | 7113/13317 [1:13:19<1:03:31,  1.63it/s]
                                                        
{'loss': 0.1951, 'grad_norm': 0.5483732223510742, 'learning_rate': 5.27705737457985e-06, 'epoch': 1.6}

 53%|█████▎    | 7113/13317 [1:13:19<1:03:31,  1.63it/s]
 53%|█████▎    | 7114/13317 [1:13:20<1:03:34,  1.63it/s]
                                                        
{'loss': 0.1788, 'grad_norm': 0.4755924642086029, 'learning_rate': 5.275748743498185e-06, 'epoch': 1.6}

 53%|█████▎    | 7114/13317 [1:13:20<1:03:34,  1.63it/s]
 53%|█████▎    | 7115/13317 [1:13:21<1:03:31,  1.63it/s]
                                                        
{'loss': 0.209, 'grad_norm': 0.45615750551223755, 'learning_rate': 5.274440093469659e-06, 'epoch': 1.6}

 53%|█████▎    | 7115/13317 [1:13:21<1:03:31,  1.63it/s]
 53%|█████▎    | 7116/13317 [1:13:21<1:03:35,  1.63it/s]
                                                        
{'loss': 0.1422, 'grad_norm': 0.4379142224788666, 'learning_rate': 5.273131424584194e-06, 'epoch': 1.6}

 53%|█████▎    | 7116/13317 [1:13:21<1:03:35,  1.63it/s]
 53%|█████▎    | 7117/13317 [1:13:22<1:03:33,  1.63it/s]
                                                        
{'loss': 0.2191, 'grad_norm': 0.5993815660476685, 'learning_rate': 5.271822736931708e-06, 'epoch': 1.6}

 53%|█████▎    | 7117/13317 [1:13:22<1:03:33,  1.63it/s]
 53%|█████▎    | 7118/13317 [1:13:22<1:03:31,  1.63it/s]
                                                        
{'loss': 0.2561, 'grad_norm': 0.5883170366287231, 'learning_rate': 5.270514030602122e-06, 'epoch': 1.6}

 53%|█████▎    | 7118/13317 [1:13:23<1:03:31,  1.63it/s]
 53%|█████▎    | 7119/13317 [1:13:23<1:03:35,  1.62it/s]
                                                        
{'loss': 0.2126, 'grad_norm': 0.5870510339736938, 'learning_rate': 5.2692053056853585e-06, 'epoch': 1.6}

 53%|█████▎    | 7119/13317 [1:13:23<1:03:35,  1.62it/s]
 53%|█████▎    | 7120/13317 [1:13:24<1:03:36,  1.62it/s]
                                                        
{'loss': 0.2719, 'grad_norm': 0.8217763900756836, 'learning_rate': 5.267896562271339e-06, 'epoch': 1.6}

 53%|█████▎    | 7120/13317 [1:13:24<1:03:36,  1.62it/s]
 53%|█████▎    | 7121/13317 [1:13:24<1:03:36,  1.62it/s]
                                                        
{'loss': 0.2236, 'grad_norm': 0.5588961243629456, 'learning_rate': 5.266587800449991e-06, 'epoch': 1.6}

 53%|█████▎    | 7121/13317 [1:13:24<1:03:36,  1.62it/s]
 53%|█████▎    | 7122/13317 [1:13:25<1:03:33,  1.62it/s]
                                                        
{'loss': 0.2462, 'grad_norm': 0.6093043684959412, 'learning_rate': 5.265279020311236e-06, 'epoch': 1.6}

 53%|█████▎    | 7122/13317 [1:13:25<1:03:33,  1.62it/s]
 53%|█████▎    | 7123/13317 [1:13:25<1:03:32,  1.62it/s]
                                                        
{'loss': 0.1707, 'grad_norm': 0.4327956438064575, 'learning_rate': 5.263970221945004e-06, 'epoch': 1.6}

 53%|█████▎    | 7123/13317 [1:13:26<1:03:32,  1.62it/s]
 53%|█████▎    | 7124/13317 [1:13:26<1:03:36,  1.62it/s]
                                                        
{'loss': 0.1874, 'grad_norm': 0.5021454691886902, 'learning_rate': 5.262661405441225e-06, 'epoch': 1.6}

 53%|█████▎    | 7124/13317 [1:13:26<1:03:36,  1.62it/s]
 54%|█████▎    | 7125/13317 [1:13:27<1:03:31,  1.62it/s]
                                                        
{'loss': 0.203, 'grad_norm': 0.5354024767875671, 'learning_rate': 5.261352570889823e-06, 'epoch': 1.61}

 54%|█████▎    | 7125/13317 [1:13:27<1:03:31,  1.62it/s]
 54%|█████▎    | 7126/13317 [1:13:27<1:03:33,  1.62it/s]
                                                        
{'loss': 0.1506, 'grad_norm': 0.4316437244415283, 'learning_rate': 5.260043718380735e-06, 'epoch': 1.61}

 54%|█████▎    | 7126/13317 [1:13:27<1:03:33,  1.62it/s]
 54%|█████▎    | 7127/13317 [1:13:28<1:03:33,  1.62it/s]
                                                        
{'loss': 0.1916, 'grad_norm': 0.46964773535728455, 'learning_rate': 5.258734848003888e-06, 'epoch': 1.61}

 54%|█████▎    | 7127/13317 [1:13:28<1:03:33,  1.62it/s]
 54%|█████▎    | 7128/13317 [1:13:29<1:03:29,  1.62it/s]
                                                        
{'loss': 0.1521, 'grad_norm': 0.43710026144981384, 'learning_rate': 5.257425959849218e-06, 'epoch': 1.61}

 54%|█████▎    | 7128/13317 [1:13:29<1:03:29,  1.62it/s]
 54%|█████▎    | 7129/13317 [1:13:29<1:03:28,  1.62it/s]
                                                        
{'loss': 0.2013, 'grad_norm': 0.5554503202438354, 'learning_rate': 5.25611705400666e-06, 'epoch': 1.61}

 54%|█████▎    | 7129/13317 [1:13:29<1:03:28,  1.62it/s]
 54%|█████▎    | 7130/13317 [1:13:30<1:03:24,  1.63it/s]
                                                        
{'loss': 0.2685, 'grad_norm': 0.5813581347465515, 'learning_rate': 5.254808130566148e-06, 'epoch': 1.61}

 54%|█████▎    | 7130/13317 [1:13:30<1:03:24,  1.63it/s]
 54%|█████▎    | 7131/13317 [1:13:30<1:03:22,  1.63it/s]
                                                        
{'loss': 0.216, 'grad_norm': 0.5264366865158081, 'learning_rate': 5.25349918961762e-06, 'epoch': 1.61}

 54%|█████▎    | 7131/13317 [1:13:31<1:03:22,  1.63it/s]
 54%|█████▎    | 7132/13317 [1:13:31<1:03:26,  1.62it/s]
                                                        
{'loss': 0.1779, 'grad_norm': 0.43524405360221863, 'learning_rate': 5.252190231251012e-06, 'epoch': 1.61}

 54%|█████▎    | 7132/13317 [1:13:31<1:03:26,  1.62it/s]
 54%|█████▎    | 7133/13317 [1:13:32<1:03:23,  1.63it/s]
                                                        
{'loss': 0.2141, 'grad_norm': 0.5060290694236755, 'learning_rate': 5.250881255556266e-06, 'epoch': 1.61}

 54%|█████▎    | 7133/13317 [1:13:32<1:03:23,  1.63it/s]
 54%|█████▎    | 7134/13317 [1:13:32<1:03:23,  1.63it/s]
                                                        
{'loss': 0.1607, 'grad_norm': 0.4492722451686859, 'learning_rate': 5.2495722626233215e-06, 'epoch': 1.61}

 54%|█████▎    | 7134/13317 [1:13:32<1:03:23,  1.63it/s]
 54%|█████▎    | 7135/13317 [1:13:33<1:03:20,  1.63it/s]
                                                        
{'loss': 0.1883, 'grad_norm': 0.5469435453414917, 'learning_rate': 5.248263252542119e-06, 'epoch': 1.61}

 54%|█████▎    | 7135/13317 [1:13:33<1:03:20,  1.63it/s]
 54%|█████▎    | 7136/13317 [1:13:33<1:03:27,  1.62it/s]
                                                        
{'loss': 0.1728, 'grad_norm': 0.45432355999946594, 'learning_rate': 5.246954225402603e-06, 'epoch': 1.61}

 54%|█████▎    | 7136/13317 [1:13:34<1:03:27,  1.62it/s]
 54%|█████▎    | 7137/13317 [1:13:34<1:03:27,  1.62it/s]
                                                        
{'loss': 0.2417, 'grad_norm': 0.6075946688652039, 'learning_rate': 5.245645181294717e-06, 'epoch': 1.61}

 54%|█████▎    | 7137/13317 [1:13:34<1:03:27,  1.62it/s]
 54%|█████▎    | 7138/13317 [1:13:35<1:03:21,  1.63it/s]
                                                        
{'loss': 0.2793, 'grad_norm': 0.7223010063171387, 'learning_rate': 5.244336120308406e-06, 'epoch': 1.61}

 54%|█████▎    | 7138/13317 [1:13:35<1:03:21,  1.63it/s]
 54%|█████▎    | 7139/13317 [1:13:35<1:03:23,  1.62it/s]
                                                        
{'loss': 0.2069, 'grad_norm': 0.5276742577552795, 'learning_rate': 5.2430270425336164e-06, 'epoch': 1.61}

 54%|█████▎    | 7139/13317 [1:13:35<1:03:23,  1.62it/s]
 54%|█████▎    | 7140/13317 [1:13:36<1:03:16,  1.63it/s]
                                                        
{'loss': 0.2231, 'grad_norm': 0.5439051985740662, 'learning_rate': 5.2417179480602965e-06, 'epoch': 1.61}

 54%|█████▎    | 7140/13317 [1:13:36<1:03:16,  1.63it/s]
 54%|█████▎    | 7141/13317 [1:13:37<1:03:16,  1.63it/s]
                                                        
{'loss': 0.173, 'grad_norm': 0.46780726313591003, 'learning_rate': 5.240408836978394e-06, 'epoch': 1.61}

 54%|█████▎    | 7141/13317 [1:13:37<1:03:16,  1.63it/s]
 54%|█████▎    | 7142/13317 [1:13:37<1:03:22,  1.62it/s]
                                                        
{'loss': 0.2341, 'grad_norm': 0.5093647241592407, 'learning_rate': 5.239099709377858e-06, 'epoch': 1.61}

 54%|█████▎    | 7142/13317 [1:13:37<1:03:22,  1.62it/s]
 54%|█████▎    | 7143/13317 [1:13:38<1:03:20,  1.62it/s]
                                                        
{'loss': 0.2305, 'grad_norm': 0.6043696403503418, 'learning_rate': 5.237790565348642e-06, 'epoch': 1.61}

 54%|█████▎    | 7143/13317 [1:13:38<1:03:20,  1.62it/s]
 54%|█████▎    | 7144/13317 [1:13:38<1:03:22,  1.62it/s]
                                                        
{'loss': 0.2728, 'grad_norm': 0.5972473621368408, 'learning_rate': 5.236481404980694e-06, 'epoch': 1.61}

 54%|█████▎    | 7144/13317 [1:13:39<1:03:22,  1.62it/s]
 54%|█████▎    | 7145/13317 [1:13:39<1:03:19,  1.62it/s]
                                                        
{'loss': 0.163, 'grad_norm': 0.45467787981033325, 'learning_rate': 5.235172228363971e-06, 'epoch': 1.61}

 54%|█████▎    | 7145/13317 [1:13:39<1:03:19,  1.62it/s]
 54%|█████▎    | 7146/13317 [1:13:40<1:03:18,  1.62it/s]
                                                        
{'loss': 0.2179, 'grad_norm': 0.5352340936660767, 'learning_rate': 5.233863035588427e-06, 'epoch': 1.61}

 54%|█████▎    | 7146/13317 [1:13:40<1:03:18,  1.62it/s]
 54%|█████▎    | 7147/13317 [1:13:40<1:03:16,  1.63it/s]
                                                        
{'loss': 0.2427, 'grad_norm': 0.709722638130188, 'learning_rate': 5.232553826744013e-06, 'epoch': 1.61}

 54%|█████▎    | 7147/13317 [1:13:40<1:03:16,  1.63it/s]
 54%|█████▎    | 7148/13317 [1:13:41<1:03:15,  1.63it/s]
                                                        
{'loss': 0.1869, 'grad_norm': 0.4545430541038513, 'learning_rate': 5.231244601920692e-06, 'epoch': 1.61}

 54%|█████▎    | 7148/13317 [1:13:41<1:03:15,  1.63it/s]
 54%|█████▎    | 7149/13317 [1:13:41<1:03:16,  1.62it/s]
                                                        
{'loss': 0.1888, 'grad_norm': 0.5553690195083618, 'learning_rate': 5.229935361208417e-06, 'epoch': 1.61}

 54%|█████▎    | 7149/13317 [1:13:42<1:03:16,  1.62it/s]
 54%|█████▎    | 7150/13317 [1:13:42<1:03:12,  1.63it/s]
                                                        
{'loss': 0.196, 'grad_norm': 0.5028542280197144, 'learning_rate': 5.228626104697149e-06, 'epoch': 1.61}

 54%|█████▎    | 7150/13317 [1:13:42<1:03:12,  1.63it/s]
 54%|█████▎    | 7151/13317 [1:13:43<1:03:11,  1.63it/s]
                                                        
{'loss': 0.2165, 'grad_norm': 0.5663976073265076, 'learning_rate': 5.2273168324768465e-06, 'epoch': 1.61}

 54%|█████▎    | 7151/13317 [1:13:43<1:03:11,  1.63it/s]
 54%|█████▎    | 7152/13317 [1:13:43<1:03:11,  1.63it/s]
                                                        
{'loss': 0.1773, 'grad_norm': 0.4701593518257141, 'learning_rate': 5.226007544637469e-06, 'epoch': 1.61}

 54%|█████▎    | 7152/13317 [1:13:43<1:03:11,  1.63it/s]
 54%|█████▎    | 7153/13317 [1:13:44<1:03:09,  1.63it/s]
                                                        
{'loss': 0.1847, 'grad_norm': 0.45466792583465576, 'learning_rate': 5.224698241268983e-06, 'epoch': 1.61}

 54%|█████▎    | 7153/13317 [1:13:44<1:03:09,  1.63it/s]
 54%|█████▎    | 7154/13317 [1:13:45<1:03:10,  1.63it/s]
                                                        
{'loss': 0.1272, 'grad_norm': 0.4202376902103424, 'learning_rate': 5.223388922461345e-06, 'epoch': 1.61}

 54%|█████▎    | 7154/13317 [1:13:45<1:03:10,  1.63it/s]
 54%|█████▎    | 7155/13317 [1:13:45<1:03:07,  1.63it/s]
                                                        
{'loss': 0.1991, 'grad_norm': 0.524358868598938, 'learning_rate': 5.222079588304527e-06, 'epoch': 1.61}

 54%|█████▎    | 7155/13317 [1:13:45<1:03:07,  1.63it/s]
 54%|█████▎    | 7156/13317 [1:13:46<1:03:07,  1.63it/s]
                                                        
{'loss': 0.189, 'grad_norm': 0.5058379769325256, 'learning_rate': 5.2207702388884875e-06, 'epoch': 1.61}

 54%|█████▎    | 7156/13317 [1:13:46<1:03:07,  1.63it/s]
 54%|█████▎    | 7157/13317 [1:13:46<1:03:06,  1.63it/s]
                                                        
{'loss': 0.2118, 'grad_norm': 0.5810631513595581, 'learning_rate': 5.2194608743031945e-06, 'epoch': 1.61}

 54%|█████▎    | 7157/13317 [1:13:47<1:03:06,  1.63it/s]
 54%|█████▍    | 7158/13317 [1:13:47<1:03:03,  1.63it/s]
                                                        
{'loss': 0.2266, 'grad_norm': 0.5505291819572449, 'learning_rate': 5.218151494638617e-06, 'epoch': 1.61}

 54%|█████▍    | 7158/13317 [1:13:47<1:03:03,  1.63it/s]
 54%|█████▍    | 7159/13317 [1:13:48<1:03:06,  1.63it/s]
                                                        
{'loss': 0.2485, 'grad_norm': 0.6022589206695557, 'learning_rate': 5.216842099984722e-06, 'epoch': 1.61}

 54%|█████▍    | 7159/13317 [1:13:48<1:03:06,  1.63it/s]
 54%|█████▍    | 7160/13317 [1:13:48<1:03:04,  1.63it/s]
                                                        
{'loss': 0.1919, 'grad_norm': 0.5642301440238953, 'learning_rate': 5.2155326904314795e-06, 'epoch': 1.61}

 54%|█████▍    | 7160/13317 [1:13:48<1:03:04,  1.63it/s]
 54%|█████▍    | 7161/13317 [1:13:49<1:03:07,  1.63it/s]
                                                        
{'loss': 0.2186, 'grad_norm': 0.5679441094398499, 'learning_rate': 5.2142232660688565e-06, 'epoch': 1.61}

 54%|█████▍    | 7161/13317 [1:13:49<1:03:07,  1.63it/s]
 54%|█████▍    | 7162/13317 [1:13:49<1:03:09,  1.62it/s]
                                                        
{'loss': 0.2123, 'grad_norm': 0.5824337005615234, 'learning_rate': 5.2129138269868294e-06, 'epoch': 1.61}

 54%|█████▍    | 7162/13317 [1:13:50<1:03:09,  1.62it/s]
 54%|█████▍    | 7163/13317 [1:13:50<1:03:04,  1.63it/s]
                                                        
{'loss': 0.2389, 'grad_norm': 0.6084970831871033, 'learning_rate': 5.211604373275366e-06, 'epoch': 1.61}

 54%|█████▍    | 7163/13317 [1:13:50<1:03:04,  1.63it/s]
 54%|█████▍    | 7164/13317 [1:13:51<1:03:07,  1.62it/s]
                                                        
{'loss': 0.1999, 'grad_norm': 0.5613682270050049, 'learning_rate': 5.210294905024443e-06, 'epoch': 1.61}

 54%|█████▍    | 7164/13317 [1:13:51<1:03:07,  1.62it/s]
 54%|█████▍    | 7165/13317 [1:13:51<1:03:02,  1.63it/s]
                                                        
{'loss': 0.198, 'grad_norm': 0.4926520586013794, 'learning_rate': 5.208985422324034e-06, 'epoch': 1.61}

 54%|█████▍    | 7165/13317 [1:13:51<1:03:02,  1.63it/s]
 54%|█████▍    | 7166/13317 [1:13:52<1:03:04,  1.63it/s]
                                                        
{'loss': 0.2152, 'grad_norm': 0.5610690712928772, 'learning_rate': 5.207675925264112e-06, 'epoch': 1.61}

 54%|█████▍    | 7166/13317 [1:13:52<1:03:04,  1.63it/s]
 54%|█████▍    | 7167/13317 [1:13:53<1:03:04,  1.63it/s]
                                                        
{'loss': 0.2261, 'grad_norm': 0.6094096302986145, 'learning_rate': 5.206366413934656e-06, 'epoch': 1.61}

 54%|█████▍    | 7167/13317 [1:13:53<1:03:04,  1.63it/s]
 54%|█████▍    | 7168/13317 [1:13:53<1:02:59,  1.63it/s]
                                                        
{'loss': 0.2204, 'grad_norm': 0.5930729508399963, 'learning_rate': 5.205056888425642e-06, 'epoch': 1.61}

 54%|█████▍    | 7168/13317 [1:13:53<1:02:59,  1.63it/s]
 54%|█████▍    | 7169/13317 [1:13:54<1:03:02,  1.63it/s]
                                                        
{'loss': 0.235, 'grad_norm': 0.6054964065551758, 'learning_rate': 5.203747348827048e-06, 'epoch': 1.62}

 54%|█████▍    | 7169/13317 [1:13:54<1:03:02,  1.63it/s]
 54%|█████▍    | 7170/13317 [1:13:54<1:02:59,  1.63it/s]
                                                        
{'loss': 0.16, 'grad_norm': 0.42776209115982056, 'learning_rate': 5.202437795228855e-06, 'epoch': 1.62}

 54%|█████▍    | 7170/13317 [1:13:55<1:02:59,  1.63it/s]
 54%|█████▍    | 7171/13317 [1:13:55<1:03:01,  1.63it/s]
                                                        
{'loss': 0.2093, 'grad_norm': 0.47901150584220886, 'learning_rate': 5.201128227721042e-06, 'epoch': 1.62}

 54%|█████▍    | 7171/13317 [1:13:55<1:03:01,  1.63it/s]
 54%|█████▍    | 7172/13317 [1:13:56<1:02:59,  1.63it/s]
                                                        
{'loss': 0.1964, 'grad_norm': 0.48341408371925354, 'learning_rate': 5.19981864639359e-06, 'epoch': 1.62}

 54%|█████▍    | 7172/13317 [1:13:56<1:02:59,  1.63it/s]
 54%|█████▍    | 7173/13317 [1:13:56<1:02:55,  1.63it/s]
                                                        
{'loss': 0.1848, 'grad_norm': 0.47751206159591675, 'learning_rate': 5.1985090513364815e-06, 'epoch': 1.62}

 54%|█████▍    | 7173/13317 [1:13:56<1:02:55,  1.63it/s]
 54%|█████▍    | 7174/13317 [1:13:57<1:02:59,  1.63it/s]
                                                        
{'loss': 0.24, 'grad_norm': 0.5295319557189941, 'learning_rate': 5.197199442639697e-06, 'epoch': 1.62}

 54%|█████▍    | 7174/13317 [1:13:57<1:02:59,  1.63it/s]
 54%|█████▍    | 7175/13317 [1:13:57<1:02:57,  1.63it/s]
                                                        
{'loss': 0.1968, 'grad_norm': 0.5134786367416382, 'learning_rate': 5.195889820393226e-06, 'epoch': 1.62}

 54%|█████▍    | 7175/13317 [1:13:58<1:02:57,  1.63it/s]
 54%|█████▍    | 7176/13317 [1:13:58<1:02:55,  1.63it/s]
                                                        
{'loss': 0.1457, 'grad_norm': 0.4126206040382385, 'learning_rate': 5.194580184687048e-06, 'epoch': 1.62}

 54%|█████▍    | 7176/13317 [1:13:58<1:02:55,  1.63it/s]
 54%|█████▍    | 7177/13317 [1:13:59<1:02:55,  1.63it/s]
                                                        
{'loss': 0.2236, 'grad_norm': 0.5224302411079407, 'learning_rate': 5.193270535611152e-06, 'epoch': 1.62}

 54%|█████▍    | 7177/13317 [1:13:59<1:02:55,  1.63it/s]
 54%|█████▍    | 7178/13317 [1:13:59<1:02:55,  1.63it/s]
                                                        
{'loss': 0.1677, 'grad_norm': 0.49476882815361023, 'learning_rate': 5.191960873255525e-06, 'epoch': 1.62}

 54%|█████▍    | 7178/13317 [1:13:59<1:02:55,  1.63it/s]
 54%|█████▍    | 7179/13317 [1:14:00<1:02:57,  1.62it/s]
                                                        
{'loss': 0.2371, 'grad_norm': 0.5927494764328003, 'learning_rate': 5.190651197710151e-06, 'epoch': 1.62}

 54%|█████▍    | 7179/13317 [1:14:00<1:02:57,  1.62it/s]
 54%|█████▍    | 7180/13317 [1:14:01<1:02:55,  1.63it/s]
                                                        
{'loss': 0.1647, 'grad_norm': 0.4560144543647766, 'learning_rate': 5.1893415090650235e-06, 'epoch': 1.62}

 54%|█████▍    | 7180/13317 [1:14:01<1:02:55,  1.63it/s]
 54%|█████▍    | 7181/13317 [1:14:01<1:03:11,  1.62it/s]
                                                        
{'loss': 0.1711, 'grad_norm': 0.5195112228393555, 'learning_rate': 5.1880318074101266e-06, 'epoch': 1.62}

 54%|█████▍    | 7181/13317 [1:14:01<1:03:11,  1.62it/s]
 54%|█████▍    | 7182/13317 [1:14:02<1:03:08,  1.62it/s]
                                                        
{'loss': 0.2319, 'grad_norm': 0.6437726616859436, 'learning_rate': 5.1867220928354565e-06, 'epoch': 1.62}

 54%|█████▍    | 7182/13317 [1:14:02<1:03:08,  1.62it/s]
 54%|█████▍    | 7183/13317 [1:14:02<1:03:00,  1.62it/s]
                                                        
{'loss': 0.1915, 'grad_norm': 0.48259642720222473, 'learning_rate': 5.185412365430999e-06, 'epoch': 1.62}

 54%|█████▍    | 7183/13317 [1:14:03<1:03:00,  1.62it/s]
 54%|█████▍    | 7184/13317 [1:14:03<1:03:00,  1.62it/s]
                                                        
{'loss': 0.1973, 'grad_norm': 0.5316976308822632, 'learning_rate': 5.184102625286751e-06, 'epoch': 1.62}

 54%|█████▍    | 7184/13317 [1:14:03<1:03:00,  1.62it/s]
 54%|█████▍    | 7185/13317 [1:14:04<1:02:56,  1.62it/s]
                                                        
{'loss': 0.1593, 'grad_norm': 0.4620870351791382, 'learning_rate': 5.182792872492702e-06, 'epoch': 1.62}

 54%|█████▍    | 7185/13317 [1:14:04<1:02:56,  1.62it/s]
 54%|█████▍    | 7186/13317 [1:14:04<1:02:55,  1.62it/s]
                                                        
{'loss': 0.1798, 'grad_norm': 0.522649347782135, 'learning_rate': 5.181483107138848e-06, 'epoch': 1.62}

 54%|█████▍    | 7186/13317 [1:14:04<1:02:55,  1.62it/s]
 54%|█████▍    | 7187/13317 [1:14:05<1:02:56,  1.62it/s]
                                                        
{'loss': 0.1366, 'grad_norm': 0.39712953567504883, 'learning_rate': 5.1801733293151844e-06, 'epoch': 1.62}

 54%|█████▍    | 7187/13317 [1:14:05<1:02:56,  1.62it/s]
 54%|█████▍    | 7188/13317 [1:14:05<1:02:50,  1.63it/s]
                                                        
{'loss': 0.2117, 'grad_norm': 0.5564481019973755, 'learning_rate': 5.178863539111703e-06, 'epoch': 1.62}

 54%|█████▍    | 7188/13317 [1:14:06<1:02:50,  1.63it/s]
 54%|█████▍    | 7189/13317 [1:14:06<1:02:50,  1.63it/s]
                                                        
{'loss': 0.2046, 'grad_norm': 0.44400596618652344, 'learning_rate': 5.177553736618404e-06, 'epoch': 1.62}

 54%|█████▍    | 7189/13317 [1:14:06<1:02:50,  1.63it/s]
 54%|█████▍    | 7190/13317 [1:14:07<1:02:47,  1.63it/s]
                                                        
{'loss': 0.2587, 'grad_norm': 0.5285660028457642, 'learning_rate': 5.176243921925283e-06, 'epoch': 1.62}

 54%|█████▍    | 7190/13317 [1:14:07<1:02:47,  1.63it/s]
 54%|█████▍    | 7191/13317 [1:14:07<1:02:49,  1.63it/s]
                                                        
{'loss': 0.2116, 'grad_norm': 0.5392860770225525, 'learning_rate': 5.174934095122337e-06, 'epoch': 1.62}

 54%|█████▍    | 7191/13317 [1:14:07<1:02:49,  1.63it/s]
 54%|█████▍    | 7192/13317 [1:14:08<1:02:45,  1.63it/s]
                                                        
{'loss': 0.2003, 'grad_norm': 0.5684996843338013, 'learning_rate': 5.173624256299567e-06, 'epoch': 1.62}

 54%|█████▍    | 7192/13317 [1:14:08<1:02:45,  1.63it/s]
 54%|█████▍    | 7193/13317 [1:14:09<1:02:43,  1.63it/s]
                                                        
{'loss': 0.1797, 'grad_norm': 0.5192857980728149, 'learning_rate': 5.172314405546973e-06, 'epoch': 1.62}

 54%|█████▍    | 7193/13317 [1:14:09<1:02:43,  1.63it/s]
 54%|█████▍    | 7194/13317 [1:14:09<1:02:46,  1.63it/s]
                                                        
{'loss': 0.1556, 'grad_norm': 0.4497133195400238, 'learning_rate': 5.171004542954555e-06, 'epoch': 1.62}

 54%|█████▍    | 7194/13317 [1:14:09<1:02:46,  1.63it/s]
 54%|█████▍    | 7195/13317 [1:14:10<1:02:43,  1.63it/s]
                                                        
{'loss': 0.185, 'grad_norm': 0.4923328161239624, 'learning_rate': 5.169694668612315e-06, 'epoch': 1.62}

 54%|█████▍    | 7195/13317 [1:14:10<1:02:43,  1.63it/s]
 54%|█████▍    | 7196/13317 [1:14:10<1:02:45,  1.63it/s]
                                                        
{'loss': 0.2092, 'grad_norm': 0.49469366669654846, 'learning_rate': 5.168384782610253e-06, 'epoch': 1.62}

 54%|█████▍    | 7196/13317 [1:14:11<1:02:45,  1.63it/s]
 54%|█████▍    | 7197/13317 [1:14:11<1:02:43,  1.63it/s]
                                                        
{'loss': 0.2269, 'grad_norm': 0.5551508069038391, 'learning_rate': 5.1670748850383734e-06, 'epoch': 1.62}

 54%|█████▍    | 7197/13317 [1:14:11<1:02:43,  1.63it/s]
 54%|█████▍    | 7198/13317 [1:14:12<1:02:39,  1.63it/s]
                                                        
{'loss': 0.1794, 'grad_norm': 0.5338966250419617, 'learning_rate': 5.165764975986681e-06, 'epoch': 1.62}

 54%|█████▍    | 7198/13317 [1:14:12<1:02:39,  1.63it/s]
 54%|█████▍    | 7199/13317 [1:14:12<1:02:41,  1.63it/s]
                                                        
{'loss': 0.2101, 'grad_norm': 0.5146652460098267, 'learning_rate': 5.164455055545179e-06, 'epoch': 1.62}

 54%|█████▍    | 7199/13317 [1:14:12<1:02:41,  1.63it/s]
 54%|█████▍    | 7200/13317 [1:14:13<1:02:37,  1.63it/s]
                                                        
{'loss': 0.1993, 'grad_norm': 0.4920101761817932, 'learning_rate': 5.163145123803873e-06, 'epoch': 1.62}

 54%|█████▍    | 7200/13317 [1:14:13<1:02:37,  1.63it/s]
 54%|█████▍    | 7201/13317 [1:14:13<1:02:39,  1.63it/s]
                                                        
{'loss': 0.2314, 'grad_norm': 0.5603706240653992, 'learning_rate': 5.161835180852771e-06, 'epoch': 1.62}

 54%|█████▍    | 7201/13317 [1:14:14<1:02:39,  1.63it/s]
 54%|█████▍    | 7202/13317 [1:14:14<1:02:36,  1.63it/s]
                                                        
{'loss': 0.2601, 'grad_norm': 0.6629812717437744, 'learning_rate': 5.160525226781878e-06, 'epoch': 1.62}

 54%|█████▍    | 7202/13317 [1:14:14<1:02:36,  1.63it/s]
 54%|█████▍    | 7203/13317 [1:14:15<1:02:36,  1.63it/s]
                                                        
{'loss': 0.1627, 'grad_norm': 0.4280727505683899, 'learning_rate': 5.159215261681203e-06, 'epoch': 1.62}

 54%|█████▍    | 7203/13317 [1:14:15<1:02:36,  1.63it/s]
 54%|█████▍    | 7204/13317 [1:14:15<1:02:38,  1.63it/s]
                                                        
{'loss': 0.1678, 'grad_norm': 0.4548726975917816, 'learning_rate': 5.157905285640753e-06, 'epoch': 1.62}

 54%|█████▍    | 7204/13317 [1:14:15<1:02:38,  1.63it/s]
 54%|█████▍    | 7205/13317 [1:14:16<1:02:35,  1.63it/s]
                                                        
{'loss': 0.1971, 'grad_norm': 0.5074963569641113, 'learning_rate': 5.156595298750538e-06, 'epoch': 1.62}

 54%|█████▍    | 7205/13317 [1:14:16<1:02:35,  1.63it/s]
 54%|█████▍    | 7206/13317 [1:14:17<1:02:35,  1.63it/s]
                                                        
{'loss': 0.173, 'grad_norm': 0.4738781452178955, 'learning_rate': 5.155285301100567e-06, 'epoch': 1.62}

 54%|█████▍    | 7206/13317 [1:14:17<1:02:35,  1.63it/s]
 54%|█████▍    | 7207/13317 [1:14:17<1:02:36,  1.63it/s]
                                                        
{'loss': 0.1909, 'grad_norm': 0.5052748322486877, 'learning_rate': 5.153975292780852e-06, 'epoch': 1.62}

 54%|█████▍    | 7207/13317 [1:14:17<1:02:36,  1.63it/s]
 54%|█████▍    | 7208/13317 [1:14:18<1:02:33,  1.63it/s]
                                                        
{'loss': 0.2073, 'grad_norm': 0.5581414103507996, 'learning_rate': 5.152665273881404e-06, 'epoch': 1.62}

 54%|█████▍    | 7208/13317 [1:14:18<1:02:33,  1.63it/s]
 54%|█████▍    | 7209/13317 [1:14:18<1:02:33,  1.63it/s]
                                                        
{'loss': 0.2247, 'grad_norm': 0.575028121471405, 'learning_rate': 5.1513552444922365e-06, 'epoch': 1.62}

 54%|█████▍    | 7209/13317 [1:14:19<1:02:33,  1.63it/s]
 54%|█████▍    | 7210/13317 [1:14:19<1:02:30,  1.63it/s]
                                                        
{'loss': 0.1818, 'grad_norm': 0.4687754213809967, 'learning_rate': 5.150045204703361e-06, 'epoch': 1.62}

 54%|█████▍    | 7210/13317 [1:14:19<1:02:30,  1.63it/s]
 54%|█████▍    | 7211/13317 [1:14:20<1:02:33,  1.63it/s]
                                                        
{'loss': 0.1929, 'grad_norm': 0.5487077832221985, 'learning_rate': 5.148735154604791e-06, 'epoch': 1.62}

 54%|█████▍    | 7211/13317 [1:14:20<1:02:33,  1.63it/s]
 54%|█████▍    | 7212/13317 [1:14:20<1:04:09,  1.59it/s]
                                                        
{'loss': 0.2065, 'grad_norm': 0.530825138092041, 'learning_rate': 5.14742509428654e-06, 'epoch': 1.62}

 54%|█████▍    | 7212/13317 [1:14:20<1:04:09,  1.59it/s]
 54%|█████▍    | 7213/13317 [1:14:21<1:03:37,  1.60it/s]
                                                        
{'loss': 0.1763, 'grad_norm': 0.523227334022522, 'learning_rate': 5.146115023838623e-06, 'epoch': 1.62}

 54%|█████▍    | 7213/13317 [1:14:21<1:03:37,  1.60it/s]
 54%|█████▍    | 7214/13317 [1:14:22<1:03:21,  1.61it/s]
                                                        
{'loss': 0.2016, 'grad_norm': 0.5106335878372192, 'learning_rate': 5.1448049433510605e-06, 'epoch': 1.63}

 54%|█████▍    | 7214/13317 [1:14:22<1:03:21,  1.61it/s]
 54%|█████▍    | 7215/13317 [1:14:22<1:03:02,  1.61it/s]
                                                        
{'loss': 0.1777, 'grad_norm': 0.4678870737552643, 'learning_rate': 5.143494852913862e-06, 'epoch': 1.63}

 54%|█████▍    | 7215/13317 [1:14:22<1:03:02,  1.61it/s]
 54%|█████▍    | 7216/13317 [1:14:23<1:02:52,  1.62it/s]
                                                        
{'loss': 0.2491, 'grad_norm': 0.6182467341423035, 'learning_rate': 5.14218475261705e-06, 'epoch': 1.63}

 54%|█████▍    | 7216/13317 [1:14:23<1:02:52,  1.62it/s]
 54%|█████▍    | 7217/13317 [1:14:23<1:02:47,  1.62it/s]
                                                        
{'loss': 0.2317, 'grad_norm': 0.580965518951416, 'learning_rate': 5.140874642550639e-06, 'epoch': 1.63}

 54%|█████▍    | 7217/13317 [1:14:23<1:02:47,  1.62it/s]
 54%|█████▍    | 7218/13317 [1:14:24<1:02:39,  1.62it/s]
                                                        
{'loss': 0.1981, 'grad_norm': 0.5047381520271301, 'learning_rate': 5.139564522804648e-06, 'epoch': 1.63}

 54%|█████▍    | 7218/13317 [1:14:24<1:02:39,  1.62it/s]
 54%|█████▍    | 7219/13317 [1:14:25<1:02:39,  1.62it/s]
                                                        
{'loss': 0.1585, 'grad_norm': 0.44610998034477234, 'learning_rate': 5.138254393469097e-06, 'epoch': 1.63}

 54%|█████▍    | 7219/13317 [1:14:25<1:02:39,  1.62it/s]
 54%|█████▍    | 7220/13317 [1:14:25<1:02:34,  1.62it/s]
                                                        
{'loss': 0.1854, 'grad_norm': 0.4255506992340088, 'learning_rate': 5.136944254634004e-06, 'epoch': 1.63}

 54%|█████▍    | 7220/13317 [1:14:25<1:02:34,  1.62it/s]
 54%|█████▍    | 7221/13317 [1:14:26<1:02:33,  1.62it/s]
                                                        
{'loss': 0.1719, 'grad_norm': 0.4995812773704529, 'learning_rate': 5.135634106389393e-06, 'epoch': 1.63}

 54%|█████▍    | 7221/13317 [1:14:26<1:02:33,  1.62it/s]
 54%|█████▍    | 7222/13317 [1:14:26<1:02:36,  1.62it/s]
                                                        
{'loss': 0.167, 'grad_norm': 0.479166179895401, 'learning_rate': 5.1343239488252805e-06, 'epoch': 1.63}

 54%|█████▍    | 7222/13317 [1:14:27<1:02:36,  1.62it/s]
 54%|█████▍    | 7223/13317 [1:14:27<1:02:29,  1.63it/s]
                                                        
{'loss': 0.1554, 'grad_norm': 0.4447009563446045, 'learning_rate': 5.133013782031693e-06, 'epoch': 1.63}

 54%|█████▍    | 7223/13317 [1:14:27<1:02:29,  1.63it/s]
 54%|█████▍    | 7224/13317 [1:14:28<1:02:35,  1.62it/s]
                                                        
{'loss': 0.2514, 'grad_norm': 0.5717443823814392, 'learning_rate': 5.131703606098648e-06, 'epoch': 1.63}

 54%|█████▍    | 7224/13317 [1:14:28<1:02:35,  1.62it/s]
 54%|█████▍    | 7225/13317 [1:14:28<1:02:29,  1.62it/s]
                                                        
{'loss': 0.2057, 'grad_norm': 0.539653480052948, 'learning_rate': 5.130393421116172e-06, 'epoch': 1.63}

 54%|█████▍    | 7225/13317 [1:14:28<1:02:29,  1.62it/s]
 54%|█████▍    | 7226/13317 [1:14:29<1:02:30,  1.62it/s]
                                                        
{'loss': 0.1541, 'grad_norm': 0.44991421699523926, 'learning_rate': 5.129083227174286e-06, 'epoch': 1.63}

 54%|█████▍    | 7226/13317 [1:14:29<1:02:30,  1.62it/s]
 54%|█████▍    | 7227/13317 [1:14:30<1:02:31,  1.62it/s]
                                                        
{'loss': 0.1891, 'grad_norm': 0.5568758845329285, 'learning_rate': 5.127773024363018e-06, 'epoch': 1.63}

 54%|█████▍    | 7227/13317 [1:14:30<1:02:31,  1.62it/s]
 54%|█████▍    | 7228/13317 [1:14:30<1:02:27,  1.62it/s]
                                                        
{'loss': 0.2551, 'grad_norm': 0.613732099533081, 'learning_rate': 5.126462812772388e-06, 'epoch': 1.63}

 54%|█████▍    | 7228/13317 [1:14:30<1:02:27,  1.62it/s]
 54%|█████▍    | 7229/13317 [1:14:31<1:02:29,  1.62it/s]
                                                        
{'loss': 0.1549, 'grad_norm': 0.4765521287918091, 'learning_rate': 5.1251525924924265e-06, 'epoch': 1.63}

 54%|█████▍    | 7229/13317 [1:14:31<1:02:29,  1.62it/s]
 54%|█████▍    | 7230/13317 [1:14:31<1:02:25,  1.63it/s]
                                                        
{'loss': 0.1799, 'grad_norm': 0.5293727517127991, 'learning_rate': 5.123842363613154e-06, 'epoch': 1.63}

 54%|█████▍    | 7230/13317 [1:14:31<1:02:25,  1.63it/s]
 54%|█████▍    | 7231/13317 [1:14:32<1:02:23,  1.63it/s]
                                                        
{'loss': 0.2523, 'grad_norm': 0.5623459815979004, 'learning_rate': 5.1225321262246005e-06, 'epoch': 1.63}

 54%|█████▍    | 7231/13317 [1:14:32<1:02:23,  1.63it/s]
 54%|█████▍    | 7232/13317 [1:14:33<1:02:26,  1.62it/s]
                                                        
{'loss': 0.14, 'grad_norm': 0.4117627739906311, 'learning_rate': 5.121221880416793e-06, 'epoch': 1.63}

 54%|█████▍    | 7232/13317 [1:14:33<1:02:26,  1.62it/s]
 54%|█████▍    | 7233/13317 [1:14:33<1:02:21,  1.63it/s]
                                                        
{'loss': 0.1824, 'grad_norm': 0.49093443155288696, 'learning_rate': 5.119911626279758e-06, 'epoch': 1.63}

 54%|█████▍    | 7233/13317 [1:14:33<1:02:21,  1.63it/s]
 54%|█████▍    | 7234/13317 [1:14:34<1:02:22,  1.63it/s]
                                                        
{'loss': 0.1499, 'grad_norm': 0.4704391658306122, 'learning_rate': 5.118601363903527e-06, 'epoch': 1.63}

 54%|█████▍    | 7234/13317 [1:14:34<1:02:22,  1.63it/s]
 54%|█████▍    | 7235/13317 [1:14:34<1:02:18,  1.63it/s]
                                                        
{'loss': 0.1964, 'grad_norm': 0.5474660396575928, 'learning_rate': 5.117291093378123e-06, 'epoch': 1.63}

 54%|█████▍    | 7235/13317 [1:14:35<1:02:18,  1.63it/s]
 54%|█████▍    | 7236/13317 [1:14:35<1:02:15,  1.63it/s]
                                                        
{'loss': 0.2154, 'grad_norm': 0.5910525321960449, 'learning_rate': 5.115980814793582e-06, 'epoch': 1.63}

 54%|█████▍    | 7236/13317 [1:14:35<1:02:15,  1.63it/s]
 54%|█████▍    | 7237/13317 [1:14:36<1:02:14,  1.63it/s]
                                                        
{'loss': 0.1565, 'grad_norm': 0.4529450833797455, 'learning_rate': 5.114670528239928e-06, 'epoch': 1.63}

 54%|█████▍    | 7237/13317 [1:14:36<1:02:14,  1.63it/s]
 54%|█████▍    | 7238/13317 [1:14:36<1:02:14,  1.63it/s]
                                                        
{'loss': 0.168, 'grad_norm': 0.5036755800247192, 'learning_rate': 5.113360233807197e-06, 'epoch': 1.63}

 54%|█████▍    | 7238/13317 [1:14:36<1:02:14,  1.63it/s]
 54%|█████▍    | 7239/13317 [1:14:37<1:02:18,  1.63it/s]
                                                        
{'loss': 0.1553, 'grad_norm': 0.477877140045166, 'learning_rate': 5.112049931585418e-06, 'epoch': 1.63}

 54%|█████▍    | 7239/13317 [1:14:37<1:02:18,  1.63it/s]
 54%|█████▍    | 7240/13317 [1:14:38<1:02:16,  1.63it/s]
                                                        
{'loss': 0.1629, 'grad_norm': 0.4608604311943054, 'learning_rate': 5.110739621664621e-06, 'epoch': 1.63}

 54%|█████▍    | 7240/13317 [1:14:38<1:02:16,  1.63it/s]
 54%|█████▍    | 7241/13317 [1:14:38<1:02:14,  1.63it/s]
                                                        
{'loss': 0.1484, 'grad_norm': 0.4633721113204956, 'learning_rate': 5.109429304134839e-06, 'epoch': 1.63}

 54%|█████▍    | 7241/13317 [1:14:38<1:02:14,  1.63it/s]
 54%|█████▍    | 7242/13317 [1:14:39<1:02:12,  1.63it/s]
                                                        
{'loss': 0.184, 'grad_norm': 0.49325940012931824, 'learning_rate': 5.108118979086107e-06, 'epoch': 1.63}

 54%|█████▍    | 7242/13317 [1:14:39<1:02:12,  1.63it/s]
 54%|█████▍    | 7243/13317 [1:14:39<1:02:10,  1.63it/s]
                                                        
{'loss': 0.1837, 'grad_norm': 0.48235055804252625, 'learning_rate': 5.106808646608455e-06, 'epoch': 1.63}

 54%|█████▍    | 7243/13317 [1:14:39<1:02:10,  1.63it/s]
 54%|█████▍    | 7244/13317 [1:14:40<1:02:14,  1.63it/s]
                                                        
{'loss': 0.2218, 'grad_norm': 0.671539306640625, 'learning_rate': 5.1054983067919174e-06, 'epoch': 1.63}

 54%|█████▍    | 7244/13317 [1:14:40<1:02:14,  1.63it/s]
 54%|█████▍    | 7245/13317 [1:14:41<1:02:10,  1.63it/s]
                                                        
{'loss': 0.2048, 'grad_norm': 0.54152911901474, 'learning_rate': 5.104187959726531e-06, 'epoch': 1.63}

 54%|█████▍    | 7245/13317 [1:14:41<1:02:10,  1.63it/s]
 54%|█████▍    | 7246/13317 [1:14:41<1:02:12,  1.63it/s]
                                                        
{'loss': 0.1802, 'grad_norm': 0.48746171593666077, 'learning_rate': 5.102877605502328e-06, 'epoch': 1.63}

 54%|█████▍    | 7246/13317 [1:14:41<1:02:12,  1.63it/s]
 54%|█████▍    | 7247/13317 [1:14:42<1:02:11,  1.63it/s]
                                                        
{'loss': 0.2487, 'grad_norm': 0.6938998699188232, 'learning_rate': 5.101567244209343e-06, 'epoch': 1.63}

 54%|█████▍    | 7247/13317 [1:14:42<1:02:11,  1.63it/s]
 54%|█████▍    | 7248/13317 [1:14:42<1:02:09,  1.63it/s]
                                                        
{'loss': 0.1773, 'grad_norm': 0.47949081659317017, 'learning_rate': 5.1002568759376134e-06, 'epoch': 1.63}

 54%|█████▍    | 7248/13317 [1:14:43<1:02:09,  1.63it/s]
 54%|█████▍    | 7249/13317 [1:14:43<1:02:12,  1.63it/s]
                                                        
{'loss': 0.2023, 'grad_norm': 0.5023109912872314, 'learning_rate': 5.098946500777176e-06, 'epoch': 1.63}

 54%|█████▍    | 7249/13317 [1:14:43<1:02:12,  1.63it/s]
 54%|█████▍    | 7250/13317 [1:14:44<1:02:10,  1.63it/s]
                                                        
{'loss': 0.22, 'grad_norm': 0.530929684638977, 'learning_rate': 5.097636118818065e-06, 'epoch': 1.63}

 54%|█████▍    | 7250/13317 [1:14:44<1:02:10,  1.63it/s]
 54%|█████▍    | 7251/13317 [1:14:44<1:02:12,  1.63it/s]
                                                        
{'loss': 0.157, 'grad_norm': 0.430742472410202, 'learning_rate': 5.09632573015032e-06, 'epoch': 1.63}

 54%|█████▍    | 7251/13317 [1:14:44<1:02:12,  1.63it/s]
 54%|█████▍    | 7252/13317 [1:14:45<1:02:10,  1.63it/s]
                                                        
{'loss': 0.1757, 'grad_norm': 0.43947944045066833, 'learning_rate': 5.0950153348639764e-06, 'epoch': 1.63}

 54%|█████▍    | 7252/13317 [1:14:45<1:02:10,  1.63it/s]
 54%|█████▍    | 7253/13317 [1:14:45<1:02:07,  1.63it/s]
                                                        
{'loss': 0.1723, 'grad_norm': 0.4455338418483734, 'learning_rate': 5.093704933049074e-06, 'epoch': 1.63}

 54%|█████▍    | 7253/13317 [1:14:46<1:02:07,  1.63it/s]
 54%|█████▍    | 7254/13317 [1:14:46<1:02:09,  1.63it/s]
                                                        
{'loss': 0.2062, 'grad_norm': 0.564237117767334, 'learning_rate': 5.09239452479565e-06, 'epoch': 1.63}

 54%|█████▍    | 7254/13317 [1:14:46<1:02:09,  1.63it/s]
 54%|█████▍    | 7255/13317 [1:14:47<1:02:07,  1.63it/s]
                                                        
{'loss': 0.161, 'grad_norm': 0.43022701144218445, 'learning_rate': 5.091084110193744e-06, 'epoch': 1.63}

 54%|█████▍    | 7255/13317 [1:14:47<1:02:07,  1.63it/s]
 54%|█████▍    | 7256/13317 [1:14:47<1:02:11,  1.62it/s]
                                                        
{'loss': 0.201, 'grad_norm': 0.4759797751903534, 'learning_rate': 5.089773689333396e-06, 'epoch': 1.63}

 54%|█████▍    | 7256/13317 [1:14:47<1:02:11,  1.62it/s]
 54%|█████▍    | 7257/13317 [1:14:48<1:02:07,  1.63it/s]
                                                        
{'loss': 0.1704, 'grad_norm': 0.47111621499061584, 'learning_rate': 5.088463262304645e-06, 'epoch': 1.63}

 54%|█████▍    | 7257/13317 [1:14:48<1:02:07,  1.63it/s]
 55%|█████▍    | 7258/13317 [1:14:49<1:02:05,  1.63it/s]
                                                        
{'loss': 0.3065, 'grad_norm': 0.7553569078445435, 'learning_rate': 5.087152829197531e-06, 'epoch': 1.64}

 55%|█████▍    | 7258/13317 [1:14:49<1:02:05,  1.63it/s]
 55%|█████▍    | 7259/13317 [1:14:49<1:02:07,  1.63it/s]
                                                        
{'loss': 0.1766, 'grad_norm': 0.4549739956855774, 'learning_rate': 5.085842390102092e-06, 'epoch': 1.64}

 55%|█████▍    | 7259/13317 [1:14:49<1:02:07,  1.63it/s]
 55%|█████▍    | 7260/13317 [1:14:50<1:02:02,  1.63it/s]
                                                        
{'loss': 0.2023, 'grad_norm': 0.5458799004554749, 'learning_rate': 5.084531945108375e-06, 'epoch': 1.64}

 55%|█████▍    | 7260/13317 [1:14:50<1:02:02,  1.63it/s]
 55%|█████▍    | 7261/13317 [1:14:50<1:02:03,  1.63it/s]
                                                        
{'loss': 0.2026, 'grad_norm': 0.4778237044811249, 'learning_rate': 5.0832214943064185e-06, 'epoch': 1.64}

 55%|█████▍    | 7261/13317 [1:14:51<1:02:03,  1.63it/s]
 55%|█████▍    | 7262/13317 [1:14:51<1:02:04,  1.63it/s]
                                                        
{'loss': 0.1813, 'grad_norm': 0.5056227445602417, 'learning_rate': 5.081911037786262e-06, 'epoch': 1.64}

 55%|█████▍    | 7262/13317 [1:14:51<1:02:04,  1.63it/s]
 55%|█████▍    | 7263/13317 [1:14:52<1:01:59,  1.63it/s]
                                                        
{'loss': 0.2293, 'grad_norm': 0.6579784154891968, 'learning_rate': 5.080600575637953e-06, 'epoch': 1.64}

 55%|█████▍    | 7263/13317 [1:14:52<1:01:59,  1.63it/s]
 55%|█████▍    | 7264/13317 [1:14:52<1:01:59,  1.63it/s]
                                                        
{'loss': 0.2036, 'grad_norm': 0.5232561230659485, 'learning_rate': 5.079290107951528e-06, 'epoch': 1.64}

 55%|█████▍    | 7264/13317 [1:14:52<1:01:59,  1.63it/s]
 55%|█████▍    | 7265/13317 [1:14:53<1:01:56,  1.63it/s]
                                                        
{'loss': 0.2386, 'grad_norm': 0.5993547439575195, 'learning_rate': 5.077979634817034e-06, 'epoch': 1.64}

 55%|█████▍    | 7265/13317 [1:14:53<1:01:56,  1.63it/s]
 55%|█████▍    | 7266/13317 [1:14:53<1:01:58,  1.63it/s]
                                                        
{'loss': 0.2122, 'grad_norm': 0.5713189840316772, 'learning_rate': 5.076669156324514e-06, 'epoch': 1.64}

 55%|█████▍    | 7266/13317 [1:14:54<1:01:58,  1.63it/s]
 55%|█████▍    | 7267/13317 [1:14:54<1:01:56,  1.63it/s]
                                                        
{'loss': 0.1542, 'grad_norm': 0.40440091490745544, 'learning_rate': 5.07535867256401e-06, 'epoch': 1.64}

 55%|█████▍    | 7267/13317 [1:14:54<1:01:56,  1.63it/s]
 55%|█████▍    | 7268/13317 [1:14:55<1:01:53,  1.63it/s]
                                                        
{'loss': 0.1737, 'grad_norm': 0.5225034356117249, 'learning_rate': 5.074048183625569e-06, 'epoch': 1.64}

 55%|█████▍    | 7268/13317 [1:14:55<1:01:53,  1.63it/s]
 55%|█████▍    | 7269/13317 [1:14:55<1:01:59,  1.63it/s]
                                                        
{'loss': 0.1868, 'grad_norm': 0.5021220445632935, 'learning_rate': 5.072737689599233e-06, 'epoch': 1.64}

 55%|█████▍    | 7269/13317 [1:14:55<1:01:59,  1.63it/s]
 55%|█████▍    | 7270/13317 [1:14:56<1:02:04,  1.62it/s]
                                                        
{'loss': 0.1652, 'grad_norm': 0.42319226264953613, 'learning_rate': 5.071427190575048e-06, 'epoch': 1.64}

 55%|█████▍    | 7270/13317 [1:14:56<1:02:04,  1.62it/s]
 55%|█████▍    | 7271/13317 [1:14:57<1:02:18,  1.62it/s]
                                                        
{'loss': 0.1874, 'grad_norm': 0.487346887588501, 'learning_rate': 5.070116686643057e-06, 'epoch': 1.64}

 55%|█████▍    | 7271/13317 [1:14:57<1:02:18,  1.62it/s]
 55%|█████▍    | 7272/13317 [1:14:57<1:02:14,  1.62it/s]
                                                        
{'loss': 0.1947, 'grad_norm': 0.5171376466751099, 'learning_rate': 5.068806177893309e-06, 'epoch': 1.64}

 55%|█████▍    | 7272/13317 [1:14:57<1:02:14,  1.62it/s]
 55%|█████▍    | 7273/13317 [1:14:58<1:02:05,  1.62it/s]
                                                        
{'loss': 0.1692, 'grad_norm': 0.504727303981781, 'learning_rate': 5.067495664415848e-06, 'epoch': 1.64}

 55%|█████▍    | 7273/13317 [1:14:58<1:02:05,  1.62it/s]
 55%|█████▍    | 7274/13317 [1:14:58<1:02:04,  1.62it/s]
                                                        
{'loss': 0.216, 'grad_norm': 0.5209782719612122, 'learning_rate': 5.0661851463007196e-06, 'epoch': 1.64}

 55%|█████▍    | 7274/13317 [1:14:59<1:02:04,  1.62it/s]
 55%|█████▍    | 7275/13317 [1:14:59<1:02:00,  1.62it/s]
                                                        
{'loss': 0.2235, 'grad_norm': 0.5767617225646973, 'learning_rate': 5.064874623637971e-06, 'epoch': 1.64}

 55%|█████▍    | 7275/13317 [1:14:59<1:02:00,  1.62it/s]
 55%|█████▍    | 7276/13317 [1:15:00<1:02:02,  1.62it/s]
                                                        
{'loss': 0.173, 'grad_norm': 0.43967747688293457, 'learning_rate': 5.063564096517648e-06, 'epoch': 1.64}

 55%|█████▍    | 7276/13317 [1:15:00<1:02:02,  1.62it/s]
 55%|█████▍    | 7277/13317 [1:15:00<1:02:03,  1.62it/s]
                                                        
{'loss': 0.1912, 'grad_norm': 0.5011276006698608, 'learning_rate': 5.0622535650298e-06, 'epoch': 1.64}

 55%|█████▍    | 7277/13317 [1:15:00<1:02:03,  1.62it/s]
 55%|█████▍    | 7278/13317 [1:15:01<1:01:56,  1.62it/s]
                                                        
{'loss': 0.2, 'grad_norm': 0.5223530530929565, 'learning_rate': 5.060943029264473e-06, 'epoch': 1.64}

 55%|█████▍    | 7278/13317 [1:15:01<1:01:56,  1.62it/s]
 55%|█████▍    | 7279/13317 [1:15:01<1:01:59,  1.62it/s]
                                                        
{'loss': 0.2508, 'grad_norm': 0.6511743068695068, 'learning_rate': 5.0596324893117134e-06, 'epoch': 1.64}

 55%|█████▍    | 7279/13317 [1:15:02<1:01:59,  1.62it/s]
 55%|█████▍    | 7280/13317 [1:15:02<1:01:53,  1.63it/s]
                                                        
{'loss': 0.216, 'grad_norm': 0.5146664977073669, 'learning_rate': 5.05832194526157e-06, 'epoch': 1.64}

 55%|█████▍    | 7280/13317 [1:15:02<1:01:53,  1.63it/s]
 55%|█████▍    | 7281/13317 [1:15:03<1:01:53,  1.63it/s]
                                                        
{'loss': 0.1994, 'grad_norm': 0.5456100106239319, 'learning_rate': 5.057011397204092e-06, 'epoch': 1.64}

 55%|█████▍    | 7281/13317 [1:15:03<1:01:53,  1.63it/s]
 55%|█████▍    | 7282/13317 [1:15:03<1:01:52,  1.63it/s]
                                                        
{'loss': 0.2364, 'grad_norm': 0.5771104097366333, 'learning_rate': 5.0557008452293275e-06, 'epoch': 1.64}

 55%|█████▍    | 7282/13317 [1:15:03<1:01:52,  1.63it/s]
 55%|█████▍    | 7283/13317 [1:15:04<1:01:49,  1.63it/s]
                                                        
{'loss': 0.176, 'grad_norm': 0.42648690938949585, 'learning_rate': 5.054390289427326e-06, 'epoch': 1.64}

 55%|█████▍    | 7283/13317 [1:15:04<1:01:49,  1.63it/s]
 55%|█████▍    | 7284/13317 [1:15:05<1:01:49,  1.63it/s]
                                                        
{'loss': 0.2444, 'grad_norm': 0.5503426194190979, 'learning_rate': 5.0530797298881365e-06, 'epoch': 1.64}

 55%|█████▍    | 7284/13317 [1:15:05<1:01:49,  1.63it/s]
 55%|█████▍    | 7285/13317 [1:15:05<1:01:44,  1.63it/s]
                                                        
{'loss': 0.1918, 'grad_norm': 0.6121863126754761, 'learning_rate': 5.0517691667018056e-06, 'epoch': 1.64}

 55%|█████▍    | 7285/13317 [1:15:05<1:01:44,  1.63it/s]
 55%|█████▍    | 7286/13317 [1:15:06<1:01:46,  1.63it/s]
                                                        
{'loss': 0.1935, 'grad_norm': 0.4807908833026886, 'learning_rate': 5.050458599958384e-06, 'epoch': 1.64}

 55%|█████▍    | 7286/13317 [1:15:06<1:01:46,  1.63it/s]
 55%|█████▍    | 7287/13317 [1:15:06<1:01:47,  1.63it/s]
                                                        
{'loss': 0.1682, 'grad_norm': 0.4748699367046356, 'learning_rate': 5.049148029747924e-06, 'epoch': 1.64}

 55%|█████▍    | 7287/13317 [1:15:07<1:01:47,  1.63it/s]
 55%|█████▍    | 7288/13317 [1:15:07<1:01:46,  1.63it/s]
                                                        
{'loss': 0.2439, 'grad_norm': 0.5886806845664978, 'learning_rate': 5.0478374561604735e-06, 'epoch': 1.64}

 55%|█████▍    | 7288/13317 [1:15:07<1:01:46,  1.63it/s]
 55%|█████▍    | 7289/13317 [1:15:08<1:01:50,  1.62it/s]
                                                        
{'loss': 0.2046, 'grad_norm': 0.538898229598999, 'learning_rate': 5.046526879286083e-06, 'epoch': 1.64}

 55%|█████▍    | 7289/13317 [1:15:08<1:01:50,  1.62it/s]
 55%|█████▍    | 7290/13317 [1:15:08<1:01:42,  1.63it/s]
                                                        
{'loss': 0.1989, 'grad_norm': 0.5423792600631714, 'learning_rate': 5.0452162992148045e-06, 'epoch': 1.64}

 55%|█████▍    | 7290/13317 [1:15:08<1:01:42,  1.63it/s]
 55%|█████▍    | 7291/13317 [1:15:09<1:01:44,  1.63it/s]
                                                        
{'loss': 0.1343, 'grad_norm': 0.4201042354106903, 'learning_rate': 5.043905716036686e-06, 'epoch': 1.64}

 55%|█████▍    | 7291/13317 [1:15:09<1:01:44,  1.63it/s]
 55%|█████▍    | 7292/13317 [1:15:09<1:01:46,  1.63it/s]
                                                        
{'loss': 0.1687, 'grad_norm': 0.46846339106559753, 'learning_rate': 5.042595129841781e-06, 'epoch': 1.64}

 55%|█████▍    | 7292/13317 [1:15:10<1:01:46,  1.63it/s]
 55%|█████▍    | 7293/13317 [1:15:10<1:01:43,  1.63it/s]
                                                        
{'loss': 0.1723, 'grad_norm': 0.471351683139801, 'learning_rate': 5.0412845407201395e-06, 'epoch': 1.64}

 55%|█████▍    | 7293/13317 [1:15:10<1:01:43,  1.63it/s]
 55%|█████▍    | 7294/13317 [1:15:11<1:01:44,  1.63it/s]
                                                        
{'loss': 0.1761, 'grad_norm': 0.4801364839076996, 'learning_rate': 5.039973948761811e-06, 'epoch': 1.64}

 55%|█████▍    | 7294/13317 [1:15:11<1:01:44,  1.63it/s]
 55%|█████▍    | 7295/13317 [1:15:11<1:01:41,  1.63it/s]
                                                        
{'loss': 0.1498, 'grad_norm': 0.42127159237861633, 'learning_rate': 5.038663354056853e-06, 'epoch': 1.64}

 55%|█████▍    | 7295/13317 [1:15:11<1:01:41,  1.63it/s]
 55%|█████▍    | 7296/13317 [1:15:12<1:01:44,  1.63it/s]
                                                        
{'loss': 0.2784, 'grad_norm': 0.6620330810546875, 'learning_rate': 5.0373527566953086e-06, 'epoch': 1.64}

 55%|█████▍    | 7296/13317 [1:15:12<1:01:44,  1.63it/s]
 55%|█████▍    | 7297/13317 [1:15:13<1:01:43,  1.63it/s]
                                                        
{'loss': 0.1778, 'grad_norm': 0.4769168496131897, 'learning_rate': 5.036042156767238e-06, 'epoch': 1.64}

 55%|█████▍    | 7297/13317 [1:15:13<1:01:43,  1.63it/s]
 55%|█████▍    | 7298/13317 [1:15:13<1:01:38,  1.63it/s]
                                                        
{'loss': 0.1763, 'grad_norm': 0.5078997015953064, 'learning_rate': 5.034731554362686e-06, 'epoch': 1.64}

 55%|█████▍    | 7298/13317 [1:15:13<1:01:38,  1.63it/s]
 55%|█████▍    | 7299/13317 [1:15:14<1:01:44,  1.62it/s]
                                                        
{'loss': 0.1981, 'grad_norm': 0.5493348240852356, 'learning_rate': 5.033420949571712e-06, 'epoch': 1.64}

 55%|█████▍    | 7299/13317 [1:15:14<1:01:44,  1.62it/s]
 55%|█████▍    | 7300/13317 [1:15:14<1:01:40,  1.63it/s]
                                                        
{'loss': 0.2432, 'grad_norm': 0.5752531886100769, 'learning_rate': 5.032110342484363e-06, 'epoch': 1.64}

 55%|█████▍    | 7300/13317 [1:15:15<1:01:40,  1.63it/s]
 55%|█████▍    | 7301/13317 [1:15:15<1:01:43,  1.62it/s]
                                                        
{'loss': 0.2982, 'grad_norm': 0.6869374513626099, 'learning_rate': 5.030799733190694e-06, 'epoch': 1.64}

 55%|█████▍    | 7301/13317 [1:15:15<1:01:43,  1.62it/s]
 55%|█████▍    | 7302/13317 [1:15:16<1:01:39,  1.63it/s]
                                                        
{'loss': 0.2081, 'grad_norm': 0.5662198662757874, 'learning_rate': 5.029489121780757e-06, 'epoch': 1.64}

 55%|█████▍    | 7302/13317 [1:15:16<1:01:39,  1.63it/s]
 55%|█████▍    | 7303/13317 [1:15:16<1:01:34,  1.63it/s]
                                                        
{'loss': 0.1907, 'grad_norm': 0.4918937087059021, 'learning_rate': 5.028178508344604e-06, 'epoch': 1.65}

 55%|█████▍    | 7303/13317 [1:15:16<1:01:34,  1.63it/s]
 55%|█████▍    | 7304/13317 [1:15:17<1:01:38,  1.63it/s]
                                                        
{'loss': 0.2352, 'grad_norm': 0.6055728793144226, 'learning_rate': 5.026867892972292e-06, 'epoch': 1.65}

 55%|█████▍    | 7304/13317 [1:15:17<1:01:38,  1.63it/s]
 55%|█████▍    | 7305/13317 [1:15:17<1:01:34,  1.63it/s]
                                                        
{'loss': 0.1885, 'grad_norm': 0.4896051287651062, 'learning_rate': 5.0255572757538686e-06, 'epoch': 1.65}

 55%|█████▍    | 7305/13317 [1:15:18<1:01:34,  1.63it/s]
 55%|█████▍    | 7306/13317 [1:15:18<1:01:34,  1.63it/s]
                                                        
{'loss': 0.1777, 'grad_norm': 0.4908515214920044, 'learning_rate': 5.024246656779391e-06, 'epoch': 1.65}

 55%|█████▍    | 7306/13317 [1:15:18<1:01:34,  1.63it/s]
 55%|█████▍    | 7307/13317 [1:15:19<1:01:32,  1.63it/s]
                                                        
{'loss': 0.1769, 'grad_norm': 0.5196570754051208, 'learning_rate': 5.022936036138911e-06, 'epoch': 1.65}

 55%|█████▍    | 7307/13317 [1:15:19<1:01:32,  1.63it/s]
 55%|█████▍    | 7308/13317 [1:15:19<1:01:32,  1.63it/s]
                                                        
{'loss': 0.1722, 'grad_norm': 0.48504990339279175, 'learning_rate': 5.021625413922483e-06, 'epoch': 1.65}

 55%|█████▍    | 7308/13317 [1:15:19<1:01:32,  1.63it/s]
 55%|█████▍    | 7309/13317 [1:15:20<1:01:34,  1.63it/s]
                                                        
{'loss': 0.2058, 'grad_norm': 0.5549295544624329, 'learning_rate': 5.020314790220159e-06, 'epoch': 1.65}

 55%|█████▍    | 7309/13317 [1:15:20<1:01:34,  1.63it/s]
 55%|█████▍    | 7310/13317 [1:15:21<1:01:30,  1.63it/s]
                                                        
{'loss': 0.1768, 'grad_norm': 0.5357018709182739, 'learning_rate': 5.019004165121994e-06, 'epoch': 1.65}

 55%|█████▍    | 7310/13317 [1:15:21<1:01:30,  1.63it/s]
 55%|█████▍    | 7311/13317 [1:15:21<1:01:30,  1.63it/s]
                                                        
{'loss': 0.2658, 'grad_norm': 0.6551342606544495, 'learning_rate': 5.017693538718044e-06, 'epoch': 1.65}

 55%|█████▍    | 7311/13317 [1:15:21<1:01:30,  1.63it/s]
 55%|█████▍    | 7312/13317 [1:15:22<1:01:29,  1.63it/s]
                                                        
{'loss': 0.1721, 'grad_norm': 0.510963499546051, 'learning_rate': 5.016382911098358e-06, 'epoch': 1.65}

 55%|█████▍    | 7312/13317 [1:15:22<1:01:29,  1.63it/s]
 55%|█████▍    | 7313/13317 [1:15:22<1:01:29,  1.63it/s]
                                                        
{'loss': 0.1697, 'grad_norm': 0.4910082221031189, 'learning_rate': 5.015072282352993e-06, 'epoch': 1.65}

 55%|█████▍    | 7313/13317 [1:15:23<1:01:29,  1.63it/s]
 55%|█████▍    | 7314/13317 [1:15:23<1:01:34,  1.62it/s]
                                                        
{'loss': 0.1955, 'grad_norm': 0.6655313968658447, 'learning_rate': 5.013761652572003e-06, 'epoch': 1.65}

 55%|█████▍    | 7314/13317 [1:15:23<1:01:34,  1.62it/s]
 55%|█████▍    | 7315/13317 [1:15:24<1:01:35,  1.62it/s]
                                                        
{'loss': 0.1809, 'grad_norm': 0.4639987349510193, 'learning_rate': 5.012451021845441e-06, 'epoch': 1.65}

 55%|█████▍    | 7315/13317 [1:15:24<1:01:35,  1.62it/s]
 55%|█████▍    | 7316/13317 [1:15:24<1:01:36,  1.62it/s]
                                                        
{'loss': 0.2488, 'grad_norm': 0.6125690937042236, 'learning_rate': 5.011140390263362e-06, 'epoch': 1.65}

 55%|█████▍    | 7316/13317 [1:15:24<1:01:36,  1.62it/s]
 55%|█████▍    | 7317/13317 [1:15:25<1:01:33,  1.62it/s]
                                                        
{'loss': 0.1974, 'grad_norm': 0.5275874733924866, 'learning_rate': 5.0098297579158215e-06, 'epoch': 1.65}

 55%|█████▍    | 7317/13317 [1:15:25<1:01:33,  1.62it/s]
 55%|█████▍    | 7318/13317 [1:15:25<1:01:29,  1.63it/s]
                                                        
{'loss': 0.2661, 'grad_norm': 0.5554585456848145, 'learning_rate': 5.008519124892871e-06, 'epoch': 1.65}

 55%|█████▍    | 7318/13317 [1:15:26<1:01:29,  1.63it/s]
 55%|█████▍    | 7319/13317 [1:15:26<1:01:33,  1.62it/s]
                                                        
{'loss': 0.2191, 'grad_norm': 0.4941442012786865, 'learning_rate': 5.007208491284569e-06, 'epoch': 1.65}

 55%|█████▍    | 7319/13317 [1:15:26<1:01:33,  1.62it/s]
 55%|█████▍    | 7320/13317 [1:15:27<1:01:29,  1.63it/s]
                                                        
{'loss': 0.2384, 'grad_norm': 0.5622983574867249, 'learning_rate': 5.005897857180964e-06, 'epoch': 1.65}

 55%|█████▍    | 7320/13317 [1:15:27<1:01:29,  1.63it/s]
 55%|█████▍    | 7321/13317 [1:15:27<1:01:28,  1.63it/s]
                                                        
{'loss': 0.1898, 'grad_norm': 0.5128751397132874, 'learning_rate': 5.004587222672116e-06, 'epoch': 1.65}

 55%|█████▍    | 7321/13317 [1:15:27<1:01:28,  1.63it/s]
 55%|█████▍    | 7322/13317 [1:15:28<1:01:29,  1.62it/s]
                                                        
{'loss': 0.1995, 'grad_norm': 0.45150133967399597, 'learning_rate': 5.003276587848077e-06, 'epoch': 1.65}

 55%|█████▍    | 7322/13317 [1:15:28<1:01:29,  1.62it/s]
 55%|█████▍    | 7323/13317 [1:15:29<1:01:25,  1.63it/s]
                                                        
{'loss': 0.2439, 'grad_norm': 0.5657757520675659, 'learning_rate': 5.001965952798901e-06, 'epoch': 1.65}

 55%|█████▍    | 7323/13317 [1:15:29<1:01:25,  1.63it/s]
 55%|█████▍    | 7324/13317 [1:15:29<1:01:28,  1.62it/s]
                                                        
{'loss': 0.1801, 'grad_norm': 0.5062392354011536, 'learning_rate': 5.000655317614643e-06, 'epoch': 1.65}

 55%|█████▍    | 7324/13317 [1:15:29<1:01:28,  1.62it/s]
 55%|█████▌    | 7325/13317 [1:15:30<1:01:24,  1.63it/s]
                                                        
{'loss': 0.1849, 'grad_norm': 0.48072290420532227, 'learning_rate': 4.9993446823853574e-06, 'epoch': 1.65}

 55%|█████▌    | 7325/13317 [1:15:30<1:01:24,  1.63it/s]
 55%|█████▌    | 7326/13317 [1:15:30<1:01:25,  1.63it/s]
                                                        
{'loss': 0.2434, 'grad_norm': 0.5978647470474243, 'learning_rate': 4.9980340472011e-06, 'epoch': 1.65}

 55%|█████▌    | 7326/13317 [1:15:31<1:01:25,  1.63it/s]
 55%|█████▌    | 7327/13317 [1:15:31<1:01:26,  1.63it/s]
                                                        
{'loss': 0.1608, 'grad_norm': 0.4587480425834656, 'learning_rate': 4.996723412151925e-06, 'epoch': 1.65}

 55%|█████▌    | 7327/13317 [1:15:31<1:01:26,  1.63it/s]
 55%|█████▌    | 7328/13317 [1:15:32<1:01:22,  1.63it/s]
                                                        
{'loss': 0.2825, 'grad_norm': 0.5210952758789062, 'learning_rate': 4.995412777327885e-06, 'epoch': 1.65}

 55%|█████▌    | 7328/13317 [1:15:32<1:01:22,  1.63it/s]
 55%|█████▌    | 7329/13317 [1:15:32<1:01:25,  1.62it/s]
                                                        
{'loss': 0.2119, 'grad_norm': 0.505435049533844, 'learning_rate': 4.9941021428190365e-06, 'epoch': 1.65}

 55%|█████▌    | 7329/13317 [1:15:32<1:01:25,  1.62it/s]
 55%|█████▌    | 7330/13317 [1:15:33<1:01:21,  1.63it/s]
                                                        
{'loss': 0.1723, 'grad_norm': 0.5040183663368225, 'learning_rate': 4.9927915087154345e-06, 'epoch': 1.65}

 55%|█████▌    | 7330/13317 [1:15:33<1:01:21,  1.63it/s]
 55%|█████▌    | 7331/13317 [1:15:33<1:01:25,  1.62it/s]
                                                        
{'loss': 0.1609, 'grad_norm': 0.4177142381668091, 'learning_rate': 4.99148087510713e-06, 'epoch': 1.65}

 55%|█████▌    | 7331/13317 [1:15:34<1:01:25,  1.62it/s]
 55%|█████▌    | 7332/13317 [1:15:34<1:01:20,  1.63it/s]
                                                        
{'loss': 0.1877, 'grad_norm': 0.5034470558166504, 'learning_rate': 4.990170242084179e-06, 'epoch': 1.65}

 55%|█████▌    | 7332/13317 [1:15:34<1:01:20,  1.63it/s]
 55%|█████▌    | 7333/13317 [1:15:35<1:01:16,  1.63it/s]
                                                        
{'loss': 0.2217, 'grad_norm': 0.538724958896637, 'learning_rate': 4.98885960973664e-06, 'epoch': 1.65}

 55%|█████▌    | 7333/13317 [1:15:35<1:01:16,  1.63it/s]
 55%|█████▌    | 7334/13317 [1:15:35<1:01:18,  1.63it/s]
                                                        
{'loss': 0.2183, 'grad_norm': 0.5626336336135864, 'learning_rate': 4.98754897815456e-06, 'epoch': 1.65}

 55%|█████▌    | 7334/13317 [1:15:35<1:01:18,  1.63it/s]
 55%|█████▌    | 7335/13317 [1:15:36<1:01:16,  1.63it/s]
                                                        
{'loss': 0.2085, 'grad_norm': 0.586033046245575, 'learning_rate': 4.986238347427998e-06, 'epoch': 1.65}

 55%|█████▌    | 7335/13317 [1:15:36<1:01:16,  1.63it/s]
 55%|█████▌    | 7336/13317 [1:15:37<1:01:19,  1.63it/s]
                                                        
{'loss': 0.2138, 'grad_norm': 0.5693533420562744, 'learning_rate': 4.984927717647009e-06, 'epoch': 1.65}

 55%|█████▌    | 7336/13317 [1:15:37<1:01:19,  1.63it/s]
 55%|█████▌    | 7337/13317 [1:15:37<1:01:17,  1.63it/s]
                                                        
{'loss': 0.2272, 'grad_norm': 0.6419259905815125, 'learning_rate': 4.983617088901645e-06, 'epoch': 1.65}

 55%|█████▌    | 7337/13317 [1:15:37<1:01:17,  1.63it/s]
 55%|█████▌    | 7338/13317 [1:15:38<1:01:15,  1.63it/s]
                                                        
{'loss': 0.2121, 'grad_norm': 0.5594373345375061, 'learning_rate': 4.982306461281958e-06, 'epoch': 1.65}

 55%|█████▌    | 7338/13317 [1:15:38<1:01:15,  1.63it/s]
 55%|█████▌    | 7339/13317 [1:15:38<1:01:11,  1.63it/s]
                                                        
{'loss': 0.1803, 'grad_norm': 0.5168997049331665, 'learning_rate': 4.980995834878006e-06, 'epoch': 1.65}

 55%|█████▌    | 7339/13317 [1:15:39<1:01:11,  1.63it/s]
 55%|█████▌    | 7340/13317 [1:15:39<1:01:10,  1.63it/s]
                                                        
{'loss': 0.2019, 'grad_norm': 0.49258115887641907, 'learning_rate': 4.979685209779842e-06, 'epoch': 1.65}

 55%|█████▌    | 7340/13317 [1:15:39<1:01:10,  1.63it/s]
 55%|█████▌    | 7341/13317 [1:15:40<1:01:14,  1.63it/s]
                                                        
{'loss': 0.2146, 'grad_norm': 0.5659079551696777, 'learning_rate': 4.978374586077518e-06, 'epoch': 1.65}

 55%|█████▌    | 7341/13317 [1:15:40<1:01:14,  1.63it/s]
 55%|█████▌    | 7342/13317 [1:15:40<1:01:11,  1.63it/s]
                                                        
{'loss': 0.1906, 'grad_norm': 0.6336652636528015, 'learning_rate': 4.97706396386109e-06, 'epoch': 1.65}

 55%|█████▌    | 7342/13317 [1:15:40<1:01:11,  1.63it/s]
 55%|█████▌    | 7343/13317 [1:15:41<1:01:09,  1.63it/s]
                                                        
{'loss': 0.1601, 'grad_norm': 0.47945520281791687, 'learning_rate': 4.975753343220609e-06, 'epoch': 1.65}

 55%|█████▌    | 7343/13317 [1:15:41<1:01:09,  1.63it/s]
 55%|█████▌    | 7344/13317 [1:15:41<1:01:08,  1.63it/s]
                                                        
{'loss': 0.1904, 'grad_norm': 0.5196856260299683, 'learning_rate': 4.974442724246133e-06, 'epoch': 1.65}

 55%|█████▌    | 7344/13317 [1:15:42<1:01:08,  1.63it/s]
 55%|█████▌    | 7345/13317 [1:15:42<1:01:06,  1.63it/s]
                                                        
{'loss': 0.1809, 'grad_norm': 0.5920265913009644, 'learning_rate': 4.973132107027709e-06, 'epoch': 1.65}

 55%|█████▌    | 7345/13317 [1:15:42<1:01:06,  1.63it/s]
 55%|█████▌    | 7346/13317 [1:15:43<1:01:05,  1.63it/s]
                                                        
{'loss': 0.1798, 'grad_norm': 0.46386992931365967, 'learning_rate': 4.971821491655396e-06, 'epoch': 1.65}

 55%|█████▌    | 7346/13317 [1:15:43<1:01:05,  1.63it/s]
 55%|█████▌    | 7347/13317 [1:15:43<1:01:02,  1.63it/s]
                                                        
{'loss': 0.1944, 'grad_norm': 0.4766245186328888, 'learning_rate': 4.970510878219244e-06, 'epoch': 1.66}

 55%|█████▌    | 7347/13317 [1:15:43<1:01:02,  1.63it/s]
 55%|█████▌    | 7348/13317 [1:15:44<1:01:03,  1.63it/s]
                                                        
{'loss': 0.2472, 'grad_norm': 0.5536177754402161, 'learning_rate': 4.9692002668093075e-06, 'epoch': 1.66}

 55%|█████▌    | 7348/13317 [1:15:44<1:01:03,  1.63it/s]
 55%|█████▌    | 7349/13317 [1:15:45<1:01:10,  1.63it/s]
                                                        
{'loss': 0.1405, 'grad_norm': 0.4277435541152954, 'learning_rate': 4.9678896575156385e-06, 'epoch': 1.66}

 55%|█████▌    | 7349/13317 [1:15:45<1:01:10,  1.63it/s]
 55%|█████▌    | 7350/13317 [1:15:45<1:01:07,  1.63it/s]
                                                        
{'loss': 0.2292, 'grad_norm': 0.6293033361434937, 'learning_rate': 4.96657905042829e-06, 'epoch': 1.66}

 55%|█████▌    | 7350/13317 [1:15:45<1:01:07,  1.63it/s]
 55%|█████▌    | 7351/13317 [1:15:46<1:01:10,  1.63it/s]
                                                        
{'loss': 0.2664, 'grad_norm': 0.5479112863540649, 'learning_rate': 4.9652684456373155e-06, 'epoch': 1.66}

 55%|█████▌    | 7351/13317 [1:15:46<1:01:10,  1.63it/s]
 55%|█████▌    | 7352/13317 [1:15:46<1:01:08,  1.63it/s]
                                                        
{'loss': 0.2314, 'grad_norm': 0.5896051526069641, 'learning_rate': 4.963957843232764e-06, 'epoch': 1.66}

 55%|█████▌    | 7352/13317 [1:15:47<1:01:08,  1.63it/s]
 55%|█████▌    | 7353/13317 [1:15:47<1:01:04,  1.63it/s]
                                                        
{'loss': 0.2127, 'grad_norm': 0.5237513184547424, 'learning_rate': 4.962647243304691e-06, 'epoch': 1.66}

 55%|█████▌    | 7353/13317 [1:15:47<1:01:04,  1.63it/s]
 55%|█████▌    | 7354/13317 [1:15:48<1:01:03,  1.63it/s]
                                                        
{'loss': 0.2195, 'grad_norm': 0.6327598690986633, 'learning_rate': 4.96133664594315e-06, 'epoch': 1.66}

 55%|█████▌    | 7354/13317 [1:15:48<1:01:03,  1.63it/s]
 55%|█████▌    | 7355/13317 [1:15:48<1:01:02,  1.63it/s]
                                                        
{'loss': 0.2166, 'grad_norm': 0.5269202589988708, 'learning_rate': 4.96002605123819e-06, 'epoch': 1.66}

 55%|█████▌    | 7355/13317 [1:15:48<1:01:02,  1.63it/s]
 55%|█████▌    | 7356/13317 [1:15:49<1:01:06,  1.63it/s]
                                                        
{'loss': 0.2224, 'grad_norm': 0.6184138655662537, 'learning_rate': 4.958715459279862e-06, 'epoch': 1.66}

 55%|█████▌    | 7356/13317 [1:15:49<1:01:06,  1.63it/s]
 55%|█████▌    | 7357/13317 [1:15:49<1:01:05,  1.63it/s]
                                                        
{'loss': 0.2342, 'grad_norm': 0.5869218707084656, 'learning_rate': 4.957404870158221e-06, 'epoch': 1.66}

 55%|█████▌    | 7357/13317 [1:15:50<1:01:05,  1.63it/s]
 55%|█████▌    | 7358/13317 [1:15:50<1:01:01,  1.63it/s]
                                                        
{'loss': 0.2002, 'grad_norm': 0.49209147691726685, 'learning_rate': 4.956094283963316e-06, 'epoch': 1.66}

 55%|█████▌    | 7358/13317 [1:15:50<1:01:01,  1.63it/s]
 55%|█████▌    | 7359/13317 [1:15:51<1:01:04,  1.63it/s]
                                                        
{'loss': 0.227, 'grad_norm': 0.5504924058914185, 'learning_rate': 4.954783700785196e-06, 'epoch': 1.66}

 55%|█████▌    | 7359/13317 [1:15:51<1:01:04,  1.63it/s]
 55%|█████▌    | 7360/13317 [1:15:51<1:01:01,  1.63it/s]
                                                        
{'loss': 0.3084, 'grad_norm': 0.6197952628135681, 'learning_rate': 4.953473120713917e-06, 'epoch': 1.66}

 55%|█████▌    | 7360/13317 [1:15:51<1:01:01,  1.63it/s]
 55%|█████▌    | 7361/13317 [1:15:52<1:01:01,  1.63it/s]
                                                        
{'loss': 0.2233, 'grad_norm': 0.582594096660614, 'learning_rate': 4.952162543839527e-06, 'epoch': 1.66}

 55%|█████▌    | 7361/13317 [1:15:52<1:01:01,  1.63it/s]
 55%|█████▌    | 7362/13317 [1:15:53<1:00:58,  1.63it/s]
                                                        
{'loss': 0.1899, 'grad_norm': 0.5609330534934998, 'learning_rate': 4.9508519702520765e-06, 'epoch': 1.66}

 55%|█████▌    | 7362/13317 [1:15:53<1:00:58,  1.63it/s]
 55%|█████▌    | 7363/13317 [1:15:53<1:00:57,  1.63it/s]
                                                        
{'loss': 0.227, 'grad_norm': 0.5888752937316895, 'learning_rate': 4.9495414000416165e-06, 'epoch': 1.66}

 55%|█████▌    | 7363/13317 [1:15:53<1:00:57,  1.63it/s]
 55%|█████▌    | 7364/13317 [1:15:54<1:00:59,  1.63it/s]
                                                        
{'loss': 0.1956, 'grad_norm': 0.5286619663238525, 'learning_rate': 4.948230833298197e-06, 'epoch': 1.66}

 55%|█████▌    | 7364/13317 [1:15:54<1:00:59,  1.63it/s]
 55%|█████▌    | 7365/13317 [1:15:54<1:00:58,  1.63it/s]
                                                        
{'loss': 0.1926, 'grad_norm': 0.5292000770568848, 'learning_rate': 4.946920270111866e-06, 'epoch': 1.66}

 55%|█████▌    | 7365/13317 [1:15:55<1:00:58,  1.63it/s]
 55%|█████▌    | 7366/13317 [1:15:55<1:01:02,  1.62it/s]
                                                        
{'loss': 0.283, 'grad_norm': 0.6979288458824158, 'learning_rate': 4.945609710572674e-06, 'epoch': 1.66}

 55%|█████▌    | 7366/13317 [1:15:55<1:01:02,  1.62it/s]
 55%|█████▌    | 7367/13317 [1:15:56<1:00:58,  1.63it/s]
                                                        
{'loss': 0.2484, 'grad_norm': 0.6305752992630005, 'learning_rate': 4.944299154770673e-06, 'epoch': 1.66}

 55%|█████▌    | 7367/13317 [1:15:56<1:00:58,  1.63it/s]
 55%|█████▌    | 7368/13317 [1:15:56<1:00:54,  1.63it/s]
                                                        
{'loss': 0.167, 'grad_norm': 0.43057674169540405, 'learning_rate': 4.942988602795909e-06, 'epoch': 1.66}

 55%|█████▌    | 7368/13317 [1:15:56<1:00:54,  1.63it/s]
 55%|█████▌    | 7369/13317 [1:15:57<1:00:57,  1.63it/s]
                                                        
{'loss': 0.1877, 'grad_norm': 0.493143767118454, 'learning_rate': 4.9416780547384305e-06, 'epoch': 1.66}

 55%|█████▌    | 7369/13317 [1:15:57<1:00:57,  1.63it/s]
 55%|█████▌    | 7370/13317 [1:15:57<1:00:56,  1.63it/s]
                                                        
{'loss': 0.2159, 'grad_norm': 0.6031073331832886, 'learning_rate': 4.940367510688287e-06, 'epoch': 1.66}

 55%|█████▌    | 7370/13317 [1:15:58<1:00:56,  1.63it/s]
 55%|█████▌    | 7371/13317 [1:15:58<1:00:55,  1.63it/s]
                                                        
{'loss': 0.1779, 'grad_norm': 0.4692228436470032, 'learning_rate': 4.93905697073553e-06, 'epoch': 1.66}

 55%|█████▌    | 7371/13317 [1:15:58<1:00:55,  1.63it/s]
 55%|█████▌    | 7372/13317 [1:15:59<1:00:54,  1.63it/s]
                                                        
{'loss': 0.2474, 'grad_norm': 0.5493068695068359, 'learning_rate': 4.937746434970201e-06, 'epoch': 1.66}

 55%|█████▌    | 7372/13317 [1:15:59<1:00:54,  1.63it/s]
 55%|█████▌    | 7373/13317 [1:15:59<1:00:50,  1.63it/s]
                                                        
{'loss': 0.24, 'grad_norm': 0.5738704204559326, 'learning_rate': 4.936435903482352e-06, 'epoch': 1.66}

 55%|█████▌    | 7373/13317 [1:15:59<1:00:50,  1.63it/s]
 55%|█████▌    | 7374/13317 [1:16:00<1:00:53,  1.63it/s]
                                                        
{'loss': 0.2647, 'grad_norm': 0.556842029094696, 'learning_rate': 4.935125376362031e-06, 'epoch': 1.66}

 55%|█████▌    | 7374/13317 [1:16:00<1:00:53,  1.63it/s]
 55%|█████▌    | 7375/13317 [1:16:01<1:00:51,  1.63it/s]
                                                        
{'loss': 0.1939, 'grad_norm': 0.47426173090934753, 'learning_rate': 4.933814853699282e-06, 'epoch': 1.66}

 55%|█████▌    | 7375/13317 [1:16:01<1:00:51,  1.63it/s]
 55%|█████▌    | 7376/13317 [1:16:01<1:00:53,  1.63it/s]
                                                        
{'loss': 0.2115, 'grad_norm': 0.5303113460540771, 'learning_rate': 4.932504335584154e-06, 'epoch': 1.66}

 55%|█████▌    | 7376/13317 [1:16:01<1:00:53,  1.63it/s]
 55%|█████▌    | 7377/13317 [1:16:02<1:00:51,  1.63it/s]
                                                        
{'loss': 0.2547, 'grad_norm': 0.673291802406311, 'learning_rate': 4.931193822106691e-06, 'epoch': 1.66}

 55%|█████▌    | 7377/13317 [1:16:02<1:00:51,  1.63it/s]
 55%|█████▌    | 7378/13317 [1:16:02<1:00:49,  1.63it/s]
                                                        
{'loss': 0.1781, 'grad_norm': 0.45875582098960876, 'learning_rate': 4.929883313356945e-06, 'epoch': 1.66}

 55%|█████▌    | 7378/13317 [1:16:02<1:00:49,  1.63it/s]
 55%|█████▌    | 7379/13317 [1:16:03<1:00:51,  1.63it/s]
                                                        
{'loss': 0.2083, 'grad_norm': 0.542975664138794, 'learning_rate': 4.9285728094249535e-06, 'epoch': 1.66}

 55%|█████▌    | 7379/13317 [1:16:03<1:00:51,  1.63it/s]
 55%|█████▌    | 7380/13317 [1:16:04<1:00:46,  1.63it/s]
                                                        
{'loss': 0.1906, 'grad_norm': 0.508345365524292, 'learning_rate': 4.927262310400768e-06, 'epoch': 1.66}

 55%|█████▌    | 7380/13317 [1:16:04<1:00:46,  1.63it/s]
 55%|█████▌    | 7381/13317 [1:16:04<1:00:50,  1.63it/s]
                                                        
{'loss': 0.241, 'grad_norm': 0.6139123439788818, 'learning_rate': 4.925951816374432e-06, 'epoch': 1.66}

 55%|█████▌    | 7381/13317 [1:16:04<1:00:50,  1.63it/s]
 55%|█████▌    | 7382/13317 [1:16:05<1:00:58,  1.62it/s]
                                                        
{'loss': 0.2252, 'grad_norm': 0.5082869529724121, 'learning_rate': 4.924641327435991e-06, 'epoch': 1.66}

 55%|█████▌    | 7382/13317 [1:16:05<1:00:58,  1.62it/s]
 55%|█████▌    | 7383/13317 [1:16:05<1:00:53,  1.62it/s]
                                                        
{'loss': 0.2195, 'grad_norm': 0.6423043608665466, 'learning_rate': 4.923330843675488e-06, 'epoch': 1.66}

 55%|█████▌    | 7383/13317 [1:16:06<1:00:53,  1.62it/s]
 55%|█████▌    | 7384/13317 [1:16:06<1:00:49,  1.63it/s]
                                                        
{'loss': 0.192, 'grad_norm': 0.5253596901893616, 'learning_rate': 4.922020365182968e-06, 'epoch': 1.66}

 55%|█████▌    | 7384/13317 [1:16:06<1:00:49,  1.63it/s]
 55%|█████▌    | 7385/13317 [1:16:07<1:00:47,  1.63it/s]
                                                        
{'loss': 0.1838, 'grad_norm': 0.513199508190155, 'learning_rate': 4.920709892048474e-06, 'epoch': 1.66}

 55%|█████▌    | 7385/13317 [1:16:07<1:00:47,  1.63it/s]
 55%|█████▌    | 7386/13317 [1:16:07<1:01:01,  1.62it/s]
                                                        
{'loss': 0.21, 'grad_norm': 0.515286386013031, 'learning_rate': 4.919399424362049e-06, 'epoch': 1.66}

 55%|█████▌    | 7386/13317 [1:16:07<1:01:01,  1.62it/s]
 55%|█████▌    | 7387/13317 [1:16:08<1:00:55,  1.62it/s]
                                                        
{'loss': 0.1891, 'grad_norm': 0.5098074078559875, 'learning_rate': 4.918088962213738e-06, 'epoch': 1.66}

 55%|█████▌    | 7387/13317 [1:16:08<1:00:55,  1.62it/s]
 55%|█████▌    | 7388/13317 [1:16:09<1:00:50,  1.62it/s]
                                                        
{'loss': 0.2084, 'grad_norm': 0.5375636219978333, 'learning_rate': 4.916778505693584e-06, 'epoch': 1.66}

 55%|█████▌    | 7388/13317 [1:16:09<1:00:50,  1.62it/s]
 55%|█████▌    | 7389/13317 [1:16:09<1:00:50,  1.62it/s]
                                                        
{'loss': 0.2251, 'grad_norm': 0.5966562628746033, 'learning_rate': 4.915468054891626e-06, 'epoch': 1.66}

 55%|█████▌    | 7389/13317 [1:16:09<1:00:50,  1.62it/s]
 55%|█████▌    | 7390/13317 [1:16:10<1:00:46,  1.63it/s]
                                                        
{'loss': 0.1889, 'grad_norm': 0.46886691451072693, 'learning_rate': 4.9141576098979085e-06, 'epoch': 1.66}

 55%|█████▌    | 7390/13317 [1:16:10<1:00:46,  1.63it/s]
 56%|█████▌    | 7391/13317 [1:16:10<1:00:49,  1.62it/s]
                                                        
{'loss': 0.1871, 'grad_norm': 0.4698266386985779, 'learning_rate': 4.912847170802473e-06, 'epoch': 1.67}

 56%|█████▌    | 7391/13317 [1:16:10<1:00:49,  1.62it/s]
 56%|█████▌    | 7392/13317 [1:16:11<1:00:44,  1.63it/s]
                                                        
{'loss': 0.2128, 'grad_norm': 0.5274454951286316, 'learning_rate': 4.911536737695357e-06, 'epoch': 1.67}

 56%|█████▌    | 7392/13317 [1:16:11<1:00:44,  1.63it/s]
 56%|█████▌    | 7393/13317 [1:16:12<1:00:39,  1.63it/s]
                                                        
{'loss': 0.1937, 'grad_norm': 0.5342180132865906, 'learning_rate': 4.9102263106666056e-06, 'epoch': 1.67}

 56%|█████▌    | 7393/13317 [1:16:12<1:00:39,  1.63it/s]
 56%|█████▌    | 7394/13317 [1:16:12<1:00:43,  1.63it/s]
                                                        
{'loss': 0.1611, 'grad_norm': 0.4666421413421631, 'learning_rate': 4.908915889806257e-06, 'epoch': 1.67}

 56%|█████▌    | 7394/13317 [1:16:12<1:00:43,  1.63it/s]
 56%|█████▌    | 7395/13317 [1:16:13<1:00:41,  1.63it/s]
                                                        
{'loss': 0.1781, 'grad_norm': 0.4306219816207886, 'learning_rate': 4.907605475204352e-06, 'epoch': 1.67}

 56%|█████▌    | 7395/13317 [1:16:13<1:00:41,  1.63it/s]
 56%|█████▌    | 7396/13317 [1:16:13<1:00:40,  1.63it/s]
                                                        
{'loss': 0.1533, 'grad_norm': 0.3880946636199951, 'learning_rate': 4.906295066950927e-06, 'epoch': 1.67}

 56%|█████▌    | 7396/13317 [1:16:14<1:00:40,  1.63it/s]
 56%|█████▌    | 7397/13317 [1:16:14<1:00:37,  1.63it/s]
                                                        
{'loss': 0.1831, 'grad_norm': 0.4920230805873871, 'learning_rate': 4.9049846651360235e-06, 'epoch': 1.67}

 56%|█████▌    | 7397/13317 [1:16:14<1:00:37,  1.63it/s]
 56%|█████▌    | 7398/13317 [1:16:15<1:00:36,  1.63it/s]
                                                        
{'loss': 0.2341, 'grad_norm': 0.5458349585533142, 'learning_rate': 4.903674269849682e-06, 'epoch': 1.67}

 56%|█████▌    | 7398/13317 [1:16:15<1:00:36,  1.63it/s]
 56%|█████▌    | 7399/13317 [1:16:15<1:00:41,  1.63it/s]
                                                        
{'loss': 0.1822, 'grad_norm': 0.4926226735115051, 'learning_rate': 4.9023638811819355e-06, 'epoch': 1.67}

 56%|█████▌    | 7399/13317 [1:16:15<1:00:41,  1.63it/s]
 56%|█████▌    | 7400/13317 [1:16:16<1:00:38,  1.63it/s]
                                                        
{'loss': 0.1863, 'grad_norm': 0.4503163695335388, 'learning_rate': 4.901053499222825e-06, 'epoch': 1.67}

 56%|█████▌    | 7400/13317 [1:16:16<1:00:38,  1.63it/s]
 56%|█████▌    | 7401/13317 [1:16:17<1:00:40,  1.62it/s]
                                                        
{'loss': 0.1644, 'grad_norm': 0.5005050897598267, 'learning_rate': 4.899743124062387e-06, 'epoch': 1.67}

 56%|█████▌    | 7401/13317 [1:16:17<1:00:40,  1.62it/s]
 56%|█████▌    | 7402/13317 [1:16:17<1:00:39,  1.63it/s]
                                                        
{'loss': 0.2572, 'grad_norm': 0.6818132996559143, 'learning_rate': 4.8984327557906586e-06, 'epoch': 1.67}

 56%|█████▌    | 7402/13317 [1:16:17<1:00:39,  1.63it/s]
 56%|█████▌    | 7403/13317 [1:16:18<1:00:35,  1.63it/s]
                                                        
{'loss': 0.2196, 'grad_norm': 0.6109175682067871, 'learning_rate': 4.897122394497674e-06, 'epoch': 1.67}

 56%|█████▌    | 7403/13317 [1:16:18<1:00:35,  1.63it/s]
 56%|█████▌    | 7404/13317 [1:16:18<1:00:39,  1.62it/s]
                                                        
{'loss': 0.1855, 'grad_norm': 0.5210086703300476, 'learning_rate': 4.895812040273469e-06, 'epoch': 1.67}

 56%|█████▌    | 7404/13317 [1:16:18<1:00:39,  1.62it/s]
 56%|█████▌    | 7405/13317 [1:16:19<1:00:34,  1.63it/s]
                                                        
{'loss': 0.2203, 'grad_norm': 0.5855316519737244, 'learning_rate': 4.894501693208084e-06, 'epoch': 1.67}

 56%|█████▌    | 7405/13317 [1:16:19<1:00:34,  1.63it/s]
 56%|█████▌    | 7406/13317 [1:16:20<1:00:37,  1.62it/s]
                                                        
{'loss': 0.2075, 'grad_norm': 0.4849469065666199, 'learning_rate': 4.893191353391546e-06, 'epoch': 1.67}

 56%|█████▌    | 7406/13317 [1:16:20<1:00:37,  1.62it/s]
 56%|█████▌    | 7407/13317 [1:16:20<1:00:34,  1.63it/s]
                                                        
{'loss': 0.2105, 'grad_norm': 0.5440137982368469, 'learning_rate': 4.891881020913894e-06, 'epoch': 1.67}

 56%|█████▌    | 7407/13317 [1:16:20<1:00:34,  1.63it/s]
 56%|█████▌    | 7408/13317 [1:16:21<1:00:30,  1.63it/s]
                                                        
{'loss': 0.2029, 'grad_norm': 0.5252029895782471, 'learning_rate': 4.890570695865162e-06, 'epoch': 1.67}

 56%|█████▌    | 7408/13317 [1:16:21<1:00:30,  1.63it/s]
 56%|█████▌    | 7409/13317 [1:16:21<1:00:34,  1.63it/s]
                                                        
{'loss': 0.2149, 'grad_norm': 0.527904212474823, 'learning_rate': 4.889260378335381e-06, 'epoch': 1.67}

 56%|█████▌    | 7409/13317 [1:16:22<1:00:34,  1.63it/s]
 56%|█████▌    | 7410/13317 [1:16:22<1:00:30,  1.63it/s]
                                                        
{'loss': 0.1634, 'grad_norm': 0.48494964838027954, 'learning_rate': 4.887950068414584e-06, 'epoch': 1.67}

 56%|█████▌    | 7410/13317 [1:16:22<1:00:30,  1.63it/s]
 56%|█████▌    | 7411/13317 [1:16:23<1:00:30,  1.63it/s]
                                                        
{'loss': 0.1711, 'grad_norm': 0.46878781914711, 'learning_rate': 4.886639766192805e-06, 'epoch': 1.67}

 56%|█████▌    | 7411/13317 [1:16:23<1:00:30,  1.63it/s]
 56%|█████▌    | 7412/13317 [1:16:23<1:00:27,  1.63it/s]
                                                        
{'loss': 0.193, 'grad_norm': 0.47652870416641235, 'learning_rate': 4.885329471760073e-06, 'epoch': 1.67}

 56%|█████▌    | 7412/13317 [1:16:23<1:00:27,  1.63it/s]
 56%|█████▌    | 7413/13317 [1:16:24<1:00:25,  1.63it/s]
                                                        
{'loss': 0.2032, 'grad_norm': 0.5160307288169861, 'learning_rate': 4.884019185206419e-06, 'epoch': 1.67}

 56%|█████▌    | 7413/13317 [1:16:24<1:00:25,  1.63it/s]
 56%|█████▌    | 7414/13317 [1:16:25<1:00:29,  1.63it/s]
                                                        
{'loss': 0.1881, 'grad_norm': 0.5671024322509766, 'learning_rate': 4.882708906621878e-06, 'epoch': 1.67}

 56%|█████▌    | 7414/13317 [1:16:25<1:00:29,  1.63it/s]
 56%|█████▌    | 7415/13317 [1:16:25<1:00:28,  1.63it/s]
                                                        
{'loss': 0.3095, 'grad_norm': 0.7108439803123474, 'learning_rate': 4.881398636096476e-06, 'epoch': 1.67}

 56%|█████▌    | 7415/13317 [1:16:25<1:00:28,  1.63it/s]
 56%|█████▌    | 7416/13317 [1:16:26<1:00:25,  1.63it/s]
                                                        
{'loss': 0.1562, 'grad_norm': 0.4332064390182495, 'learning_rate': 4.880088373720243e-06, 'epoch': 1.67}

 56%|█████▌    | 7416/13317 [1:16:26<1:00:25,  1.63it/s]
 56%|█████▌    | 7417/13317 [1:16:26<1:00:29,  1.63it/s]
                                                        
{'loss': 0.2247, 'grad_norm': 0.5234014987945557, 'learning_rate': 4.878778119583207e-06, 'epoch': 1.67}

 56%|█████▌    | 7417/13317 [1:16:26<1:00:29,  1.63it/s]
 56%|█████▌    | 7418/13317 [1:16:27<1:00:25,  1.63it/s]
                                                        
{'loss': 0.2015, 'grad_norm': 0.5172388553619385, 'learning_rate': 4.877467873775402e-06, 'epoch': 1.67}

 56%|█████▌    | 7418/13317 [1:16:27<1:00:25,  1.63it/s]
 56%|█████▌    | 7419/13317 [1:16:28<1:00:27,  1.63it/s]
                                                        
{'loss': 0.1615, 'grad_norm': 0.4622945785522461, 'learning_rate': 4.876157636386848e-06, 'epoch': 1.67}

 56%|█████▌    | 7419/13317 [1:16:28<1:00:27,  1.63it/s]
 56%|█████▌    | 7420/13317 [1:16:28<1:00:24,  1.63it/s]
                                                        
{'loss': 0.1673, 'grad_norm': 0.4544826149940491, 'learning_rate': 4.874847407507576e-06, 'epoch': 1.67}

 56%|█████▌    | 7420/13317 [1:16:28<1:00:24,  1.63it/s]
 56%|█████▌    | 7421/13317 [1:16:29<1:00:22,  1.63it/s]
                                                        
{'loss': 0.1617, 'grad_norm': 0.45356637239456177, 'learning_rate': 4.8735371872276125e-06, 'epoch': 1.67}

 56%|█████▌    | 7421/13317 [1:16:29<1:00:22,  1.63it/s]
 56%|█████▌    | 7422/13317 [1:16:29<1:00:24,  1.63it/s]
                                                        
{'loss': 0.3156, 'grad_norm': 0.6731104254722595, 'learning_rate': 4.872226975636985e-06, 'epoch': 1.67}

 56%|█████▌    | 7422/13317 [1:16:30<1:00:24,  1.63it/s]
 56%|█████▌    | 7423/13317 [1:16:30<1:00:22,  1.63it/s]
                                                        
{'loss': 0.1802, 'grad_norm': 0.537318766117096, 'learning_rate': 4.8709167728257145e-06, 'epoch': 1.67}

 56%|█████▌    | 7423/13317 [1:16:30<1:00:22,  1.63it/s]
 56%|█████▌    | 7424/13317 [1:16:31<1:00:18,  1.63it/s]
                                                        
{'loss': 0.2524, 'grad_norm': 0.5417853593826294, 'learning_rate': 4.869606578883829e-06, 'epoch': 1.67}

 56%|█████▌    | 7424/13317 [1:16:31<1:00:18,  1.63it/s]
 56%|█████▌    | 7425/13317 [1:16:31<1:00:18,  1.63it/s]
                                                        
{'loss': 0.1769, 'grad_norm': 0.46745699644088745, 'learning_rate': 4.868296393901354e-06, 'epoch': 1.67}

 56%|█████▌    | 7425/13317 [1:16:31<1:00:18,  1.63it/s]
 56%|█████▌    | 7426/13317 [1:16:32<1:00:19,  1.63it/s]
                                                        
{'loss': 0.2437, 'grad_norm': 0.5687146186828613, 'learning_rate': 4.866986217968309e-06, 'epoch': 1.67}

 56%|█████▌    | 7426/13317 [1:16:32<1:00:19,  1.63it/s]
 56%|█████▌    | 7427/13317 [1:16:32<1:00:21,  1.63it/s]
                                                        
{'loss': 0.1993, 'grad_norm': 0.5009932518005371, 'learning_rate': 4.8656760511747195e-06, 'epoch': 1.67}

 56%|█████▌    | 7427/13317 [1:16:33<1:00:21,  1.63it/s]
 56%|█████▌    | 7428/13317 [1:16:33<1:00:19,  1.63it/s]
                                                        
{'loss': 0.2423, 'grad_norm': 0.6019724607467651, 'learning_rate': 4.864365893610609e-06, 'epoch': 1.67}

 56%|█████▌    | 7428/13317 [1:16:33<1:00:19,  1.63it/s]
 56%|█████▌    | 7429/13317 [1:16:34<1:00:21,  1.63it/s]
                                                        
{'loss': 0.1892, 'grad_norm': 0.46448227763175964, 'learning_rate': 4.863055745365997e-06, 'epoch': 1.67}

 56%|█████▌    | 7429/13317 [1:16:34<1:00:21,  1.63it/s]
 56%|█████▌    | 7430/13317 [1:16:34<1:00:17,  1.63it/s]
                                                        
{'loss': 0.152, 'grad_norm': 0.4005320370197296, 'learning_rate': 4.861745606530905e-06, 'epoch': 1.67}

 56%|█████▌    | 7430/13317 [1:16:34<1:00:17,  1.63it/s]
 56%|█████▌    | 7431/13317 [1:16:35<1:00:19,  1.63it/s]
                                                        
{'loss': 0.2114, 'grad_norm': 0.4809303283691406, 'learning_rate': 4.860435477195353e-06, 'epoch': 1.67}

 56%|█████▌    | 7431/13317 [1:16:35<1:00:19,  1.63it/s]
 56%|█████▌    | 7432/13317 [1:16:36<1:00:15,  1.63it/s]
                                                        
{'loss': 0.1527, 'grad_norm': 0.45671185851097107, 'learning_rate': 4.859125357449364e-06, 'epoch': 1.67}

 56%|█████▌    | 7432/13317 [1:16:36<1:00:15,  1.63it/s]
 56%|█████▌    | 7433/13317 [1:16:36<1:00:11,  1.63it/s]
                                                        
{'loss': 0.1903, 'grad_norm': 0.47993379831314087, 'learning_rate': 4.857815247382952e-06, 'epoch': 1.67}

 56%|█████▌    | 7433/13317 [1:16:36<1:00:11,  1.63it/s]
 56%|█████▌    | 7434/13317 [1:16:37<1:00:19,  1.63it/s]
                                                        
{'loss': 0.1586, 'grad_norm': 0.45561036467552185, 'learning_rate': 4.856505147086138e-06, 'epoch': 1.67}

 56%|█████▌    | 7434/13317 [1:16:37<1:00:19,  1.63it/s]
 56%|█████▌    | 7435/13317 [1:16:37<1:00:17,  1.63it/s]
                                                        
{'loss': 0.1884, 'grad_norm': 0.5478459596633911, 'learning_rate': 4.855195056648942e-06, 'epoch': 1.67}

 56%|█████▌    | 7435/13317 [1:16:38<1:00:17,  1.63it/s]
 56%|█████▌    | 7436/13317 [1:16:38<1:00:13,  1.63it/s]
                                                        
{'loss': 0.2524, 'grad_norm': 0.6250699162483215, 'learning_rate': 4.8538849761613775e-06, 'epoch': 1.68}

 56%|█████▌    | 7436/13317 [1:16:38<1:00:13,  1.63it/s]
 56%|█████▌    | 7437/13317 [1:16:39<1:00:11,  1.63it/s]
                                                        
{'loss': 0.169, 'grad_norm': 0.4821113646030426, 'learning_rate': 4.852574905713462e-06, 'epoch': 1.68}

 56%|█████▌    | 7437/13317 [1:16:39<1:00:11,  1.63it/s]
 56%|█████▌    | 7438/13317 [1:16:39<1:00:09,  1.63it/s]
                                                        
{'loss': 0.1865, 'grad_norm': 0.4911457896232605, 'learning_rate': 4.851264845395213e-06, 'epoch': 1.68}

 56%|█████▌    | 7438/13317 [1:16:39<1:00:09,  1.63it/s]
 56%|█████▌    | 7439/13317 [1:16:40<1:00:12,  1.63it/s]
                                                        
{'loss': 0.1615, 'grad_norm': 0.45533236861228943, 'learning_rate': 4.849954795296641e-06, 'epoch': 1.68}

 56%|█████▌    | 7439/13317 [1:16:40<1:00:12,  1.63it/s]
 56%|█████▌    | 7440/13317 [1:16:40<1:00:13,  1.63it/s]
                                                        
{'loss': 0.1442, 'grad_norm': 0.4177001118659973, 'learning_rate': 4.848644755507764e-06, 'epoch': 1.68}

 56%|█████▌    | 7440/13317 [1:16:41<1:00:13,  1.63it/s]
 56%|█████▌    | 7441/13317 [1:16:41<1:00:11,  1.63it/s]
                                                        
{'loss': 0.2411, 'grad_norm': 0.5439503192901611, 'learning_rate': 4.847334726118596e-06, 'epoch': 1.68}

 56%|█████▌    | 7441/13317 [1:16:41<1:00:11,  1.63it/s]
 56%|█████▌    | 7442/13317 [1:16:42<1:00:11,  1.63it/s]
                                                        
{'loss': 0.1361, 'grad_norm': 0.3924458920955658, 'learning_rate': 4.846024707219149e-06, 'epoch': 1.68}

 56%|█████▌    | 7442/13317 [1:16:42<1:00:11,  1.63it/s]
 56%|█████▌    | 7443/13317 [1:16:42<1:00:13,  1.63it/s]
                                                        
{'loss': 0.2137, 'grad_norm': 0.6059247255325317, 'learning_rate': 4.844714698899435e-06, 'epoch': 1.68}

 56%|█████▌    | 7443/13317 [1:16:42<1:00:13,  1.63it/s]
 56%|█████▌    | 7444/13317 [1:16:43<1:00:13,  1.63it/s]
                                                        
{'loss': 0.2002, 'grad_norm': 0.4935179054737091, 'learning_rate': 4.843404701249463e-06, 'epoch': 1.68}

 56%|█████▌    | 7444/13317 [1:16:43<1:00:13,  1.63it/s]
 56%|█████▌    | 7445/13317 [1:16:44<1:00:09,  1.63it/s]
                                                        
{'loss': 0.2226, 'grad_norm': 0.5355343222618103, 'learning_rate': 4.84209471435925e-06, 'epoch': 1.68}

 56%|█████▌    | 7445/13317 [1:16:44<1:00:09,  1.63it/s]
 56%|█████▌    | 7446/13317 [1:16:44<1:00:27,  1.62it/s]
                                                        
{'loss': 0.1913, 'grad_norm': 0.5228269100189209, 'learning_rate': 4.8407847383187984e-06, 'epoch': 1.68}

 56%|█████▌    | 7446/13317 [1:16:44<1:00:27,  1.62it/s]
 56%|█████▌    | 7447/13317 [1:16:45<1:00:22,  1.62it/s]
                                                        
{'loss': 0.1695, 'grad_norm': 0.43011870980262756, 'learning_rate': 4.839474773218123e-06, 'epoch': 1.68}

 56%|█████▌    | 7447/13317 [1:16:45<1:00:22,  1.62it/s]
 56%|█████▌    | 7448/13317 [1:16:45<1:00:15,  1.62it/s]
                                                        
{'loss': 0.1944, 'grad_norm': 0.4819645881652832, 'learning_rate': 4.838164819147229e-06, 'epoch': 1.68}

 56%|█████▌    | 7448/13317 [1:16:46<1:00:15,  1.62it/s]
 56%|█████▌    | 7449/13317 [1:16:46<1:00:13,  1.62it/s]
                                                        
{'loss': 0.1991, 'grad_norm': 0.5334817171096802, 'learning_rate': 4.8368548761961275e-06, 'epoch': 1.68}

 56%|█████▌    | 7449/13317 [1:16:46<1:00:13,  1.62it/s]
 56%|█████▌    | 7450/13317 [1:16:47<1:00:09,  1.63it/s]
                                                        
{'loss': 0.184, 'grad_norm': 0.5173267722129822, 'learning_rate': 4.835544944454822e-06, 'epoch': 1.68}

 56%|█████▌    | 7450/13317 [1:16:47<1:00:09,  1.63it/s]
 56%|█████▌    | 7451/13317 [1:16:47<1:00:10,  1.62it/s]
                                                        
{'loss': 0.2039, 'grad_norm': 0.48495274782180786, 'learning_rate': 4.83423502401332e-06, 'epoch': 1.68}

 56%|█████▌    | 7451/13317 [1:16:47<1:00:10,  1.62it/s]
 56%|█████▌    | 7452/13317 [1:16:48<1:00:08,  1.63it/s]
                                                        
{'loss': 0.2044, 'grad_norm': 0.5019245147705078, 'learning_rate': 4.832925114961629e-06, 'epoch': 1.68}

 56%|█████▌    | 7452/13317 [1:16:48<1:00:08,  1.63it/s]
 56%|█████▌    | 7453/13317 [1:16:48<1:00:06,  1.63it/s]
                                                        
{'loss': 0.228, 'grad_norm': 0.5866246819496155, 'learning_rate': 4.831615217389749e-06, 'epoch': 1.68}

 56%|█████▌    | 7453/13317 [1:16:49<1:00:06,  1.63it/s]
 56%|█████▌    | 7454/13317 [1:16:49<1:00:05,  1.63it/s]
                                                        
{'loss': 0.1721, 'grad_norm': 0.471868097782135, 'learning_rate': 4.830305331387687e-06, 'epoch': 1.68}

 56%|█████▌    | 7454/13317 [1:16:49<1:00:05,  1.63it/s]
 56%|█████▌    | 7455/13317 [1:16:50<1:00:05,  1.63it/s]
                                                        
{'loss': 0.2198, 'grad_norm': 0.5727577805519104, 'learning_rate': 4.8289954570454465e-06, 'epoch': 1.68}

 56%|█████▌    | 7455/13317 [1:16:50<1:00:05,  1.63it/s]
 56%|█████▌    | 7456/13317 [1:16:50<1:00:03,  1.63it/s]
                                                        
{'loss': 0.21, 'grad_norm': 0.4907388985157013, 'learning_rate': 4.827685594453028e-06, 'epoch': 1.68}

 56%|█████▌    | 7456/13317 [1:16:50<1:00:03,  1.63it/s]
 56%|█████▌    | 7457/13317 [1:16:51<1:00:06,  1.62it/s]
                                                        
{'loss': 0.1491, 'grad_norm': 0.3998325765132904, 'learning_rate': 4.826375743700433e-06, 'epoch': 1.68}

 56%|█████▌    | 7457/13317 [1:16:51<1:00:06,  1.62it/s]
 56%|█████▌    | 7458/13317 [1:16:52<1:00:05,  1.63it/s]
                                                        
{'loss': 0.2179, 'grad_norm': 0.5842616558074951, 'learning_rate': 4.825065904877663e-06, 'epoch': 1.68}

 56%|█████▌    | 7458/13317 [1:16:52<1:00:05,  1.63it/s]
 56%|█████▌    | 7459/13317 [1:16:52<1:00:07,  1.62it/s]
                                                        
{'loss': 0.2963, 'grad_norm': 0.6050488352775574, 'learning_rate': 4.823756078074721e-06, 'epoch': 1.68}

 56%|█████▌    | 7459/13317 [1:16:52<1:00:07,  1.62it/s]
 56%|█████▌    | 7460/13317 [1:16:53<1:00:02,  1.63it/s]
                                                        
{'loss': 0.1585, 'grad_norm': 0.43263864517211914, 'learning_rate': 4.822446263381598e-06, 'epoch': 1.68}

 56%|█████▌    | 7460/13317 [1:16:53<1:00:02,  1.63it/s]
 56%|█████▌    | 7461/13317 [1:16:53<1:00:02,  1.63it/s]
                                                        
{'loss': 0.1622, 'grad_norm': 0.48754173517227173, 'learning_rate': 4.821136460888298e-06, 'epoch': 1.68}

 56%|█████▌    | 7461/13317 [1:16:54<1:00:02,  1.63it/s]
 56%|█████▌    | 7462/13317 [1:16:54<1:00:03,  1.62it/s]
                                                        
{'loss': 0.2766, 'grad_norm': 0.706787645816803, 'learning_rate': 4.819826670684818e-06, 'epoch': 1.68}

 56%|█████▌    | 7462/13317 [1:16:54<1:00:03,  1.62it/s]
 56%|█████▌    | 7463/13317 [1:16:55<1:00:00,  1.63it/s]
                                                        
{'loss': 0.2101, 'grad_norm': 0.5204013586044312, 'learning_rate': 4.818516892861153e-06, 'epoch': 1.68}

 56%|█████▌    | 7463/13317 [1:16:55<1:00:00,  1.63it/s]
 56%|█████▌    | 7464/13317 [1:16:55<1:00:02,  1.62it/s]
                                                        
{'loss': 0.1653, 'grad_norm': 0.43908318877220154, 'learning_rate': 4.817207127507299e-06, 'epoch': 1.68}

 56%|█████▌    | 7464/13317 [1:16:55<1:00:02,  1.62it/s]
 56%|█████▌    | 7465/13317 [1:16:56<59:57,  1.63it/s]  
                                                      
{'loss': 0.1995, 'grad_norm': 0.5247921943664551, 'learning_rate': 4.815897374713249e-06, 'epoch': 1.68}

 56%|█████▌    | 7465/13317 [1:16:56<59:57,  1.63it/s]
 56%|█████▌    | 7466/13317 [1:16:56<1:00:00,  1.63it/s]
                                                        
{'loss': 0.2303, 'grad_norm': 0.5865961909294128, 'learning_rate': 4.8145876345690014e-06, 'epoch': 1.68}

 56%|█████▌    | 7466/13317 [1:16:57<1:00:00,  1.63it/s]
 56%|█████▌    | 7467/13317 [1:16:57<59:56,  1.63it/s]  
                                                      
{'loss': 0.1964, 'grad_norm': 0.5797193646430969, 'learning_rate': 4.813277907164545e-06, 'epoch': 1.68}

 56%|█████▌    | 7467/13317 [1:16:57<59:56,  1.63it/s]
 56%|█████▌    | 7468/13317 [1:16:58<59:55,  1.63it/s]
                                                      
{'loss': 0.1841, 'grad_norm': 0.4730904698371887, 'learning_rate': 4.811968192589873e-06, 'epoch': 1.68}

 56%|█████▌    | 7468/13317 [1:16:58<59:55,  1.63it/s]
 56%|█████▌    | 7469/13317 [1:16:58<59:55,  1.63it/s]
                                                      
{'loss': 0.1777, 'grad_norm': 0.5162469744682312, 'learning_rate': 4.81065849093498e-06, 'epoch': 1.68}

 56%|█████▌    | 7469/13317 [1:16:58<59:55,  1.63it/s]
 56%|█████▌    | 7470/13317 [1:16:59<59:53,  1.63it/s]
                                                      
{'loss': 0.1751, 'grad_norm': 0.46745598316192627, 'learning_rate': 4.80934880228985e-06, 'epoch': 1.68}

 56%|█████▌    | 7470/13317 [1:16:59<59:53,  1.63it/s]
 56%|█████▌    | 7471/13317 [1:17:00<59:52,  1.63it/s]
                                                      
{'loss': 0.1692, 'grad_norm': 0.5014040470123291, 'learning_rate': 4.808039126744476e-06, 'epoch': 1.68}

 56%|█████▌    | 7471/13317 [1:17:00<59:52,  1.63it/s]
 56%|█████▌    | 7472/13317 [1:17:00<59:56,  1.63it/s]
                                                      
{'loss': 0.2525, 'grad_norm': 0.5718440413475037, 'learning_rate': 4.806729464388851e-06, 'epoch': 1.68}

 56%|█████▌    | 7472/13317 [1:17:00<59:56,  1.63it/s]
 56%|█████▌    | 7473/13317 [1:17:01<59:50,  1.63it/s]
                                                      
{'loss': 0.168, 'grad_norm': 0.4573882520198822, 'learning_rate': 4.805419815312953e-06, 'epoch': 1.68}

 56%|█████▌    | 7473/13317 [1:17:01<59:50,  1.63it/s]
 56%|█████▌    | 7474/13317 [1:17:01<59:52,  1.63it/s]
                                                      
{'loss': 0.2221, 'grad_norm': 0.5807312726974487, 'learning_rate': 4.8041101796067755e-06, 'epoch': 1.68}

 56%|█████▌    | 7474/13317 [1:17:02<59:52,  1.63it/s]
 56%|█████▌    | 7475/13317 [1:17:02<59:50,  1.63it/s]
                                                      
{'loss': 0.1838, 'grad_norm': 0.4901393949985504, 'learning_rate': 4.8028005573603035e-06, 'epoch': 1.68}

 56%|█████▌    | 7475/13317 [1:17:02<59:50,  1.63it/s]
 56%|█████▌    | 7476/13317 [1:17:03<59:52,  1.63it/s]
                                                      
{'loss': 0.2369, 'grad_norm': 0.5714371204376221, 'learning_rate': 4.801490948663521e-06, 'epoch': 1.68}

 56%|█████▌    | 7476/13317 [1:17:03<59:52,  1.63it/s]
 56%|█████▌    | 7477/13317 [1:17:03<59:52,  1.63it/s]
                                                      
{'loss': 0.2062, 'grad_norm': 0.4858803153038025, 'learning_rate': 4.800181353606413e-06, 'epoch': 1.68}

 56%|█████▌    | 7477/13317 [1:17:03<59:52,  1.63it/s]
 56%|█████▌    | 7478/13317 [1:17:04<59:50,  1.63it/s]
                                                      
{'loss': 0.1354, 'grad_norm': 0.391143798828125, 'learning_rate': 4.798871772278959e-06, 'epoch': 1.68}

 56%|█████▌    | 7478/13317 [1:17:04<59:50,  1.63it/s]
 56%|█████▌    | 7479/13317 [1:17:04<59:49,  1.63it/s]
                                                      
{'loss': 0.1744, 'grad_norm': 0.5308247208595276, 'learning_rate': 4.797562204771147e-06, 'epoch': 1.68}

 56%|█████▌    | 7479/13317 [1:17:05<59:49,  1.63it/s]
 56%|█████▌    | 7480/13317 [1:17:05<59:47,  1.63it/s]
                                                      
{'loss': 0.1801, 'grad_norm': 0.4854661226272583, 'learning_rate': 4.796252651172953e-06, 'epoch': 1.69}

 56%|█████▌    | 7480/13317 [1:17:05<59:47,  1.63it/s]
 56%|█████▌    | 7481/13317 [1:17:06<59:47,  1.63it/s]
                                                      
{'loss': 0.2257, 'grad_norm': 0.5673900842666626, 'learning_rate': 4.794943111574359e-06, 'epoch': 1.69}

 56%|█████▌    | 7481/13317 [1:17:06<59:47,  1.63it/s]
 56%|█████▌    | 7482/13317 [1:17:06<59:44,  1.63it/s]
                                                      
{'loss': 0.1901, 'grad_norm': 0.5783113241195679, 'learning_rate': 4.793633586065345e-06, 'epoch': 1.69}

 56%|█████▌    | 7482/13317 [1:17:06<59:44,  1.63it/s]
 56%|█████▌    | 7483/13317 [1:17:07<59:43,  1.63it/s]
                                                      
{'loss': 0.2182, 'grad_norm': 0.5253275036811829, 'learning_rate': 4.79232407473589e-06, 'epoch': 1.69}

 56%|█████▌    | 7483/13317 [1:17:07<59:43,  1.63it/s]
 56%|█████▌    | 7484/13317 [1:17:08<59:47,  1.63it/s]
                                                      
{'loss': 0.2111, 'grad_norm': 0.5165311098098755, 'learning_rate': 4.791014577675969e-06, 'epoch': 1.69}

 56%|█████▌    | 7484/13317 [1:17:08<59:47,  1.63it/s]
 56%|█████▌    | 7485/13317 [1:17:08<59:44,  1.63it/s]
                                                      
{'loss': 0.1599, 'grad_norm': 0.43214356899261475, 'learning_rate': 4.789705094975558e-06, 'epoch': 1.69}

 56%|█████▌    | 7485/13317 [1:17:08<59:44,  1.63it/s]
 56%|█████▌    | 7486/13317 [1:17:09<59:45,  1.63it/s]
                                                      
{'loss': 0.1686, 'grad_norm': 0.46970096230506897, 'learning_rate': 4.788395626724636e-06, 'epoch': 1.69}

 56%|█████▌    | 7486/13317 [1:17:09<59:45,  1.63it/s]
 56%|█████▌    | 7487/13317 [1:17:09<59:47,  1.63it/s]
                                                      
{'loss': 0.1794, 'grad_norm': 0.5480268597602844, 'learning_rate': 4.787086173013172e-06, 'epoch': 1.69}

 56%|█████▌    | 7487/13317 [1:17:10<59:47,  1.63it/s]
 56%|█████▌    | 7488/13317 [1:17:10<59:44,  1.63it/s]
                                                      
{'loss': 0.2051, 'grad_norm': 0.5511844754219055, 'learning_rate': 4.785776733931144e-06, 'epoch': 1.69}

 56%|█████▌    | 7488/13317 [1:17:10<59:44,  1.63it/s]
 56%|█████▌    | 7489/13317 [1:17:11<59:46,  1.62it/s]
                                                      
{'loss': 0.1776, 'grad_norm': 0.4457753300666809, 'learning_rate': 4.784467309568524e-06, 'epoch': 1.69}

 56%|█████▌    | 7489/13317 [1:17:11<59:46,  1.62it/s]
 56%|█████▌    | 7490/13317 [1:17:11<59:45,  1.63it/s]
                                                      
{'loss': 0.1841, 'grad_norm': 0.4662325382232666, 'learning_rate': 4.783157900015279e-06, 'epoch': 1.69}

 56%|█████▌    | 7490/13317 [1:17:11<59:45,  1.63it/s]
 56%|█████▋    | 7491/13317 [1:17:12<59:41,  1.63it/s]
                                                      
{'loss': 0.1829, 'grad_norm': 0.5236969590187073, 'learning_rate': 4.781848505361385e-06, 'epoch': 1.69}

 56%|█████▋    | 7491/13317 [1:17:12<59:41,  1.63it/s]
 56%|█████▋    | 7492/13317 [1:17:12<59:39,  1.63it/s]
                                                      
{'loss': 0.256, 'grad_norm': 0.5215657353401184, 'learning_rate': 4.7805391256968055e-06, 'epoch': 1.69}

 56%|█████▋    | 7492/13317 [1:17:13<59:39,  1.63it/s]
 56%|█████▋    | 7493/13317 [1:17:13<59:37,  1.63it/s]
                                                      
{'loss': 0.1602, 'grad_norm': 0.45547720789909363, 'learning_rate': 4.779229761111514e-06, 'epoch': 1.69}

 56%|█████▋    | 7493/13317 [1:17:13<59:37,  1.63it/s]
 56%|█████▋    | 7494/13317 [1:17:14<59:53,  1.62it/s]
                                                      
{'loss': 0.2421, 'grad_norm': 0.572311282157898, 'learning_rate': 4.777920411695475e-06, 'epoch': 1.69}

 56%|█████▋    | 7494/13317 [1:17:14<59:53,  1.62it/s]
 56%|█████▋    | 7495/13317 [1:17:14<59:44,  1.62it/s]
                                                      
{'loss': 0.1974, 'grad_norm': 0.435452401638031, 'learning_rate': 4.776611077538654e-06, 'epoch': 1.69}

 56%|█████▋    | 7495/13317 [1:17:14<59:44,  1.62it/s]
 56%|█████▋    | 7496/13317 [1:17:15<59:43,  1.62it/s]
                                                      
{'loss': 0.2086, 'grad_norm': 0.5335685610771179, 'learning_rate': 4.7753017587310196e-06, 'epoch': 1.69}

 56%|█████▋    | 7496/13317 [1:17:15<59:43,  1.62it/s]
 56%|█████▋    | 7497/13317 [1:17:16<59:38,  1.63it/s]
                                                      
{'loss': 0.1726, 'grad_norm': 0.43159425258636475, 'learning_rate': 4.7739924553625324e-06, 'epoch': 1.69}

 56%|█████▋    | 7497/13317 [1:17:16<59:38,  1.63it/s]
 56%|█████▋    | 7498/13317 [1:17:16<59:36,  1.63it/s]
                                                      
{'loss': 0.2013, 'grad_norm': 0.473072350025177, 'learning_rate': 4.772683167523155e-06, 'epoch': 1.69}

 56%|█████▋    | 7498/13317 [1:17:16<59:36,  1.63it/s]
 56%|█████▋    | 7499/13317 [1:17:17<1:02:40,  1.55it/s]
                                                        
{'loss': 0.1771, 'grad_norm': 0.4928896725177765, 'learning_rate': 4.7713738953028545e-06, 'epoch': 1.69}

 56%|█████▋    | 7499/13317 [1:17:17<1:02:40,  1.55it/s]
 56%|█████▋    | 7500/13317 [1:17:17<1:01:41,  1.57it/s]
                                                        
{'loss': 0.1942, 'grad_norm': 0.520151674747467, 'learning_rate': 4.770064638791584e-06, 'epoch': 1.69}

 56%|█████▋    | 7500/13317 [1:17:18<1:01:41,  1.57it/s]

  0%|          | 0/13 [00:00<?, ?it/s]

 15%|█▌        | 2/13 [00:00<00:00, 13.37it/s]

 31%|███       | 4/13 [00:00<00:01,  8.40it/s]

 38%|███▊      | 5/13 [00:00<00:01,  7.80it/s]

 46%|████▌     | 6/13 [00:00<00:00,  7.43it/s]

 54%|█████▍    | 7/13 [00:00<00:00,  7.18it/s]

 62%|██████▏   | 8/13 [00:01<00:00,  7.01it/s]

 69%|██████▉   | 9/13 [00:01<00:00,  6.90it/s]

 77%|███████▋  | 10/13 [00:01<00:00,  6.83it/s]

 85%|████████▍ | 11/13 [00:01<00:00,  6.78it/s]

 92%|█████████▏| 12/13 [00:01<00:00,  6.74it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]
                                                        

                                               
{'eval_loss': 0.5499239563941956, 'eval_runtime': 1.9528, 'eval_samples_per_second': 102.415, 'eval_steps_per_second': 6.657, 'epoch': 1.69}

 56%|█████▋    | 7500/13317 [1:17:20<1:01:41,  1.57it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]

                                               
 56%|█████▋    | 7501/13317 [1:17:20<1:57:58,  1.22s/it]
                                                        
{'loss': 0.225, 'grad_norm': 0.5510013699531555, 'learning_rate': 4.768755398079309e-06, 'epoch': 1.69}

 56%|█████▋    | 7501/13317 [1:17:20<1:57:58,  1.22s/it]
 56%|█████▋    | 7502/13317 [1:17:21<1:40:25,  1.04s/it]
                                                        
{'loss': 0.1963, 'grad_norm': 0.5301988124847412, 'learning_rate': 4.767446173255987e-06, 'epoch': 1.69}

 56%|█████▋    | 7502/13317 [1:17:21<1:40:25,  1.04s/it]
 56%|█████▋    | 7503/13317 [1:17:21<1:28:05,  1.10it/s]
                                                        
{'loss': 0.183, 'grad_norm': 0.44061988592147827, 'learning_rate': 4.766136964411576e-06, 'epoch': 1.69}

 56%|█████▋    | 7503/13317 [1:17:22<1:28:05,  1.10it/s]
 56%|█████▋    | 7504/13317 [1:17:22<1:22:32,  1.17it/s]
                                                        
{'loss': 0.2064, 'grad_norm': 0.5661514401435852, 'learning_rate': 4.76482777163603e-06, 'epoch': 1.69}

 56%|█████▋    | 7504/13317 [1:17:22<1:22:32,  1.17it/s]
 56%|█████▋    | 7505/13317 [1:17:23<1:15:36,  1.28it/s]
                                                        
{'loss': 0.2026, 'grad_norm': 0.5944200754165649, 'learning_rate': 4.763518595019306e-06, 'epoch': 1.69}

 56%|█████▋    | 7505/13317 [1:17:23<1:15:36,  1.28it/s]
 56%|█████▋    | 7506/13317 [1:17:23<1:10:43,  1.37it/s]
                                                        
{'loss': 0.1566, 'grad_norm': 0.4084780812263489, 'learning_rate': 4.762209434651362e-06, 'epoch': 1.69}

 56%|█████▋    | 7506/13317 [1:17:23<1:10:43,  1.37it/s]
 56%|█████▋    | 7507/13317 [1:17:24<1:07:21,  1.44it/s]
                                                        
{'loss': 0.188, 'grad_norm': 0.46387243270874023, 'learning_rate': 4.760900290622143e-06, 'epoch': 1.69}

 56%|█████▋    | 7507/13317 [1:17:24<1:07:21,  1.44it/s]
 56%|█████▋    | 7508/13317 [1:17:24<1:04:58,  1.49it/s]
                                                        
{'loss': 0.1732, 'grad_norm': 0.4588668644428253, 'learning_rate': 4.759591163021607e-06, 'epoch': 1.69}

 56%|█████▋    | 7508/13317 [1:17:25<1:04:58,  1.49it/s]
 56%|█████▋    | 7509/13317 [1:17:25<1:03:21,  1.53it/s]
                                                        
{'loss': 0.1995, 'grad_norm': 0.5245073437690735, 'learning_rate': 4.758282051939704e-06, 'epoch': 1.69}

 56%|█████▋    | 7509/13317 [1:17:25<1:03:21,  1.53it/s]
 56%|█████▋    | 7510/13317 [1:17:26<1:05:16,  1.48it/s]
                                                        
{'loss': 0.2394, 'grad_norm': 0.5330265760421753, 'learning_rate': 4.756972957466384e-06, 'epoch': 1.69}

 56%|█████▋    | 7510/13317 [1:17:26<1:05:16,  1.48it/s]
 56%|█████▋    | 7511/13317 [1:17:26<1:03:30,  1.52it/s]
                                                        
{'loss': 0.1773, 'grad_norm': 0.46300768852233887, 'learning_rate': 4.755663879691596e-06, 'epoch': 1.69}

 56%|█████▋    | 7511/13317 [1:17:27<1:03:30,  1.52it/s]
 56%|█████▋    | 7512/13317 [1:17:27<1:02:16,  1.55it/s]
                                                        
{'loss': 0.1823, 'grad_norm': 0.5048240423202515, 'learning_rate': 4.754354818705284e-06, 'epoch': 1.69}

 56%|█████▋    | 7512/13317 [1:17:27<1:02:16,  1.55it/s]
 56%|█████▋    | 7513/13317 [1:17:28<1:01:23,  1.58it/s]
                                                        
{'loss': 0.1919, 'grad_norm': 0.5140899419784546, 'learning_rate': 4.7530457745974e-06, 'epoch': 1.69}

 56%|█████▋    | 7513/13317 [1:17:28<1:01:23,  1.58it/s]
 56%|█████▋    | 7514/13317 [1:17:28<1:00:45,  1.59it/s]
                                                        
{'loss': 0.1986, 'grad_norm': 0.48651784658432007, 'learning_rate': 4.7517367474578825e-06, 'epoch': 1.69}

 56%|█████▋    | 7514/13317 [1:17:28<1:00:45,  1.59it/s]
 56%|█████▋    | 7515/13317 [1:17:29<1:00:20,  1.60it/s]
                                                        
{'loss': 0.1902, 'grad_norm': 0.5142074823379517, 'learning_rate': 4.75042773737668e-06, 'epoch': 1.69}

 56%|█████▋    | 7515/13317 [1:17:29<1:00:20,  1.60it/s]
 56%|█████▋    | 7516/13317 [1:17:29<1:00:03,  1.61it/s]
                                                        
{'loss': 0.2415, 'grad_norm': 0.540716290473938, 'learning_rate': 4.749118744443736e-06, 'epoch': 1.69}

 56%|█████▋    | 7516/13317 [1:17:30<1:00:03,  1.61it/s]
 56%|█████▋    | 7517/13317 [1:17:30<59:51,  1.61it/s]  
                                                      
{'loss': 0.1623, 'grad_norm': 0.47460347414016724, 'learning_rate': 4.74780976874899e-06, 'epoch': 1.69}

 56%|█████▋    | 7517/13317 [1:17:30<59:51,  1.61it/s]
 56%|█████▋    | 7518/13317 [1:17:31<59:42,  1.62it/s]
                                                      
{'loss': 0.2589, 'grad_norm': 0.6523288488388062, 'learning_rate': 4.7465008103823816e-06, 'epoch': 1.69}

 56%|█████▋    | 7518/13317 [1:17:31<59:42,  1.62it/s]
 56%|█████▋    | 7519/13317 [1:17:31<59:32,  1.62it/s]
                                                      
{'loss': 0.2118, 'grad_norm': 0.5543038845062256, 'learning_rate': 4.745191869433853e-06, 'epoch': 1.69}

 56%|█████▋    | 7519/13317 [1:17:31<59:32,  1.62it/s]
 56%|█████▋    | 7520/13317 [1:17:32<59:28,  1.62it/s]
                                                      
{'loss': 0.203, 'grad_norm': 0.5638868808746338, 'learning_rate': 4.743882945993342e-06, 'epoch': 1.69}

 56%|█████▋    | 7520/13317 [1:17:32<59:28,  1.62it/s]
 56%|█████▋    | 7521/13317 [1:17:33<59:25,  1.63it/s]
                                                      
{'loss': 0.1859, 'grad_norm': 0.6121731400489807, 'learning_rate': 4.742574040150783e-06, 'epoch': 1.69}

 56%|█████▋    | 7521/13317 [1:17:33<59:25,  1.63it/s]
 56%|█████▋    | 7522/13317 [1:17:33<59:21,  1.63it/s]
                                                      
{'loss': 0.2349, 'grad_norm': 0.6103223562240601, 'learning_rate': 4.741265151996113e-06, 'epoch': 1.69}

 56%|█████▋    | 7522/13317 [1:17:33<59:21,  1.63it/s]
 56%|█████▋    | 7523/13317 [1:17:34<59:19,  1.63it/s]
                                                      
{'loss': 0.3504, 'grad_norm': 0.976586103439331, 'learning_rate': 4.739956281619268e-06, 'epoch': 1.69}

 56%|█████▋    | 7523/13317 [1:17:34<59:19,  1.63it/s]
 56%|█████▋    | 7524/13317 [1:17:34<59:20,  1.63it/s]
                                                      
{'loss': 0.2299, 'grad_norm': 0.6121913194656372, 'learning_rate': 4.7386474291101785e-06, 'epoch': 1.69}

 56%|█████▋    | 7524/13317 [1:17:35<59:20,  1.63it/s]
 57%|█████▋    | 7525/13317 [1:17:35<59:18,  1.63it/s]
                                                      
{'loss': 0.2796, 'grad_norm': 0.6494869589805603, 'learning_rate': 4.737338594558777e-06, 'epoch': 1.7}

 57%|█████▋    | 7525/13317 [1:17:35<59:18,  1.63it/s]
 57%|█████▋    | 7526/13317 [1:17:36<59:18,  1.63it/s]
                                                      
{'loss': 0.1898, 'grad_norm': 0.515211284160614, 'learning_rate': 4.736029778054996e-06, 'epoch': 1.7}

 57%|█████▋    | 7526/13317 [1:17:36<59:18,  1.63it/s]
 57%|█████▋    | 7527/13317 [1:17:36<59:17,  1.63it/s]
                                                      
{'loss': 0.1757, 'grad_norm': 0.5424942970275879, 'learning_rate': 4.734720979688765e-06, 'epoch': 1.7}

 57%|█████▋    | 7527/13317 [1:17:36<59:17,  1.63it/s]
 57%|█████▋    | 7528/13317 [1:17:37<59:14,  1.63it/s]
                                                      
{'loss': 0.2212, 'grad_norm': 0.554440975189209, 'learning_rate': 4.733412199550012e-06, 'epoch': 1.7}

 57%|█████▋    | 7528/13317 [1:17:37<59:14,  1.63it/s]
 57%|█████▋    | 7529/13317 [1:17:37<59:13,  1.63it/s]
                                                      
{'loss': 0.1665, 'grad_norm': 0.47375452518463135, 'learning_rate': 4.732103437728661e-06, 'epoch': 1.7}

 57%|█████▋    | 7529/13317 [1:17:38<59:13,  1.63it/s]
 57%|█████▋    | 7530/13317 [1:17:38<59:11,  1.63it/s]
                                                      
{'loss': 0.2262, 'grad_norm': 0.6317847967147827, 'learning_rate': 4.730794694314644e-06, 'epoch': 1.7}

 57%|█████▋    | 7530/13317 [1:17:38<59:11,  1.63it/s]
 57%|█████▋    | 7531/13317 [1:17:39<59:10,  1.63it/s]
                                                      
{'loss': 0.155, 'grad_norm': 0.4602128565311432, 'learning_rate': 4.72948596939788e-06, 'epoch': 1.7}

 57%|█████▋    | 7531/13317 [1:17:39<59:10,  1.63it/s]
 57%|█████▋    | 7532/13317 [1:17:39<59:27,  1.62it/s]
                                                      
{'loss': 0.2224, 'grad_norm': 0.5565625429153442, 'learning_rate': 4.728177263068292e-06, 'epoch': 1.7}

 57%|█████▋    | 7532/13317 [1:17:39<59:27,  1.62it/s]
 57%|█████▋    | 7533/13317 [1:17:40<59:21,  1.62it/s]
                                                      
{'loss': 0.1992, 'grad_norm': 0.5095118284225464, 'learning_rate': 4.726868575415808e-06, 'epoch': 1.7}

 57%|█████▋    | 7533/13317 [1:17:40<59:21,  1.62it/s]
 57%|█████▋    | 7534/13317 [1:17:41<59:16,  1.63it/s]
                                                      
{'loss': 0.2043, 'grad_norm': 0.4963219463825226, 'learning_rate': 4.725559906530342e-06, 'epoch': 1.7}

 57%|█████▋    | 7534/13317 [1:17:41<59:16,  1.63it/s]
 57%|█████▋    | 7535/13317 [1:17:41<59:11,  1.63it/s]
                                                      
{'loss': 0.2359, 'grad_norm': 0.6764605045318604, 'learning_rate': 4.724251256501817e-06, 'epoch': 1.7}

 57%|█████▋    | 7535/13317 [1:17:41<59:11,  1.63it/s]
 57%|█████▋    | 7536/13317 [1:17:42<59:11,  1.63it/s]
                                                      
{'loss': 0.1629, 'grad_norm': 0.477983683347702, 'learning_rate': 4.7229426254201504e-06, 'epoch': 1.7}

 57%|█████▋    | 7536/13317 [1:17:42<59:11,  1.63it/s]
 57%|█████▋    | 7537/13317 [1:17:42<59:16,  1.63it/s]
                                                      
{'loss': 0.1442, 'grad_norm': 0.44885939359664917, 'learning_rate': 4.7216340133752604e-06, 'epoch': 1.7}

 57%|█████▋    | 7537/13317 [1:17:43<59:16,  1.63it/s]
 57%|█████▋    | 7538/13317 [1:17:43<59:12,  1.63it/s]
                                                      
{'loss': 0.2416, 'grad_norm': 0.5913354158401489, 'learning_rate': 4.72032542045706e-06, 'epoch': 1.7}

 57%|█████▋    | 7538/13317 [1:17:43<59:12,  1.63it/s]
 57%|█████▋    | 7539/13317 [1:17:44<59:08,  1.63it/s]
                                                      
{'loss': 0.2184, 'grad_norm': 0.5551013946533203, 'learning_rate': 4.719016846755462e-06, 'epoch': 1.7}

 57%|█████▋    | 7539/13317 [1:17:44<59:08,  1.63it/s]
 57%|█████▋    | 7540/13317 [1:17:44<59:06,  1.63it/s]
                                                      
{'loss': 0.1742, 'grad_norm': 0.44807368516921997, 'learning_rate': 4.717708292360386e-06, 'epoch': 1.7}

 57%|█████▋    | 7540/13317 [1:17:44<59:06,  1.63it/s]
 57%|█████▋    | 7541/13317 [1:17:45<59:04,  1.63it/s]
                                                      
{'loss': 0.2632, 'grad_norm': 0.6640194654464722, 'learning_rate': 4.7163997573617366e-06, 'epoch': 1.7}

 57%|█████▋    | 7541/13317 [1:17:45<59:04,  1.63it/s]
 57%|█████▋    | 7542/13317 [1:17:45<59:02,  1.63it/s]
                                                      
{'loss': 0.1983, 'grad_norm': 0.5275508761405945, 'learning_rate': 4.715091241849427e-06, 'epoch': 1.7}

 57%|█████▋    | 7542/13317 [1:17:46<59:02,  1.63it/s]
 57%|█████▋    | 7543/13317 [1:17:46<59:02,  1.63it/s]
                                                      
{'loss': 0.2124, 'grad_norm': 0.5851572155952454, 'learning_rate': 4.7137827459133676e-06, 'epoch': 1.7}

 57%|█████▋    | 7543/13317 [1:17:46<59:02,  1.63it/s]
 57%|█████▋    | 7544/13317 [1:17:47<59:00,  1.63it/s]
                                                      
{'loss': 0.1737, 'grad_norm': 0.48225298523902893, 'learning_rate': 4.712474269643462e-06, 'epoch': 1.7}

 57%|█████▋    | 7544/13317 [1:17:47<59:00,  1.63it/s]
 57%|█████▋    | 7545/13317 [1:17:47<58:59,  1.63it/s]
                                                      
{'loss': 0.141, 'grad_norm': 0.4398362934589386, 'learning_rate': 4.7111658131296175e-06, 'epoch': 1.7}

 57%|█████▋    | 7545/13317 [1:17:47<58:59,  1.63it/s]
 57%|█████▋    | 7546/13317 [1:17:48<59:02,  1.63it/s]
                                                      
{'loss': 0.1647, 'grad_norm': 0.510248064994812, 'learning_rate': 4.709857376461741e-06, 'epoch': 1.7}

 57%|█████▋    | 7546/13317 [1:17:48<59:02,  1.63it/s]
 57%|█████▋    | 7547/13317 [1:17:49<59:01,  1.63it/s]
                                                      
{'loss': 0.2233, 'grad_norm': 0.5977803468704224, 'learning_rate': 4.708548959729737e-06, 'epoch': 1.7}

 57%|█████▋    | 7547/13317 [1:17:49<59:01,  1.63it/s]
 57%|█████▋    | 7548/13317 [1:17:49<59:00,  1.63it/s]
                                                      
{'loss': 0.2852, 'grad_norm': 0.7419137358665466, 'learning_rate': 4.707240563023503e-06, 'epoch': 1.7}

 57%|█████▋    | 7548/13317 [1:17:49<59:00,  1.63it/s]
 57%|█████▋    | 7549/13317 [1:17:50<59:00,  1.63it/s]
                                                      
{'loss': 0.171, 'grad_norm': 0.45929908752441406, 'learning_rate': 4.705932186432943e-06, 'epoch': 1.7}

 57%|█████▋    | 7549/13317 [1:17:50<59:00,  1.63it/s]
 57%|█████▋    | 7550/13317 [1:17:50<58:58,  1.63it/s]
                                                      
{'loss': 0.2309, 'grad_norm': 0.5554916858673096, 'learning_rate': 4.704623830047956e-06, 'epoch': 1.7}

 57%|█████▋    | 7550/13317 [1:17:50<58:58,  1.63it/s]
 57%|█████▋    | 7551/13317 [1:17:51<58:59,  1.63it/s]
                                                      
{'loss': 0.1668, 'grad_norm': 0.4440583884716034, 'learning_rate': 4.703315493958438e-06, 'epoch': 1.7}

 57%|█████▋    | 7551/13317 [1:17:51<58:59,  1.63it/s]
 57%|█████▋    | 7552/13317 [1:17:52<58:57,  1.63it/s]
                                                      
{'loss': 0.1535, 'grad_norm': 0.425571084022522, 'learning_rate': 4.702007178254286e-06, 'epoch': 1.7}

 57%|█████▋    | 7552/13317 [1:17:52<58:57,  1.63it/s]
 57%|█████▋    | 7553/13317 [1:17:52<58:57,  1.63it/s]
                                                      
{'loss': 0.2188, 'grad_norm': 0.4992372691631317, 'learning_rate': 4.7006988830253975e-06, 'epoch': 1.7}

 57%|█████▋    | 7553/13317 [1:17:52<58:57,  1.63it/s]
 57%|█████▋    | 7554/13317 [1:17:53<58:56,  1.63it/s]
                                                      
{'loss': 0.1992, 'grad_norm': 0.5660333037376404, 'learning_rate': 4.699390608361665e-06, 'epoch': 1.7}

 57%|█████▋    | 7554/13317 [1:17:53<58:56,  1.63it/s]
 57%|█████▋    | 7555/13317 [1:17:53<58:56,  1.63it/s]
                                                      
{'loss': 0.1747, 'grad_norm': 0.467835396528244, 'learning_rate': 4.698082354352981e-06, 'epoch': 1.7}

 57%|█████▋    | 7555/13317 [1:17:54<58:56,  1.63it/s]
 57%|█████▋    | 7556/13317 [1:17:54<58:55,  1.63it/s]
                                                      
{'loss': 0.1987, 'grad_norm': 0.5181900262832642, 'learning_rate': 4.696774121089235e-06, 'epoch': 1.7}

 57%|█████▋    | 7556/13317 [1:17:54<58:55,  1.63it/s]
 57%|█████▋    | 7557/13317 [1:17:55<58:53,  1.63it/s]
                                                      
{'loss': 0.24, 'grad_norm': 0.6416700482368469, 'learning_rate': 4.695465908660319e-06, 'epoch': 1.7}

 57%|█████▋    | 7557/13317 [1:17:55<58:53,  1.63it/s]
 57%|█████▋    | 7558/13317 [1:17:55<58:52,  1.63it/s]
                                                      
{'loss': 0.1893, 'grad_norm': 0.49060195684432983, 'learning_rate': 4.694157717156118e-06, 'epoch': 1.7}

 57%|█████▋    | 7558/13317 [1:17:55<58:52,  1.63it/s]
 57%|█████▋    | 7559/13317 [1:17:56<58:50,  1.63it/s]
                                                      
{'loss': 0.1726, 'grad_norm': 0.5313233733177185, 'learning_rate': 4.692849546666519e-06, 'epoch': 1.7}

 57%|█████▋    | 7559/13317 [1:17:56<58:50,  1.63it/s]
 57%|█████▋    | 7560/13317 [1:17:57<58:51,  1.63it/s]
                                                      
{'loss': 0.2537, 'grad_norm': 0.6509518623352051, 'learning_rate': 4.691541397281412e-06, 'epoch': 1.7}

 57%|█████▋    | 7560/13317 [1:17:57<58:51,  1.63it/s]
 57%|█████▋    | 7561/13317 [1:17:57<58:52,  1.63it/s]
                                                      
{'loss': 0.1688, 'grad_norm': 0.43051889538764954, 'learning_rate': 4.690233269090674e-06, 'epoch': 1.7}

 57%|█████▋    | 7561/13317 [1:17:57<58:52,  1.63it/s]
 57%|█████▋    | 7562/13317 [1:17:58<58:51,  1.63it/s]
                                                      
{'loss': 0.254, 'grad_norm': 0.6419432759284973, 'learning_rate': 4.68892516218419e-06, 'epoch': 1.7}

 57%|█████▋    | 7562/13317 [1:17:58<58:51,  1.63it/s]
 57%|█████▋    | 7563/13317 [1:17:58<58:50,  1.63it/s]
                                                      
{'loss': 0.305, 'grad_norm': 0.6480019092559814, 'learning_rate': 4.687617076651842e-06, 'epoch': 1.7}

 57%|█████▋    | 7563/13317 [1:17:58<58:50,  1.63it/s]
 57%|█████▋    | 7564/13317 [1:17:59<58:49,  1.63it/s]
                                                      
{'loss': 0.2638, 'grad_norm': 0.6221380233764648, 'learning_rate': 4.68630901258351e-06, 'epoch': 1.7}

 57%|█████▋    | 7564/13317 [1:17:59<58:49,  1.63it/s]
 57%|█████▋    | 7565/13317 [1:18:00<58:48,  1.63it/s]
                                                      
{'loss': 0.1921, 'grad_norm': 0.5358341932296753, 'learning_rate': 4.685000970069069e-06, 'epoch': 1.7}

 57%|█████▋    | 7565/13317 [1:18:00<58:48,  1.63it/s]
 57%|█████▋    | 7566/13317 [1:18:00<58:50,  1.63it/s]
                                                      
{'loss': 0.1889, 'grad_norm': 0.4791495203971863, 'learning_rate': 4.683692949198395e-06, 'epoch': 1.7}

 57%|█████▋    | 7566/13317 [1:18:00<58:50,  1.63it/s]
 57%|█████▋    | 7567/13317 [1:18:01<58:50,  1.63it/s]
                                                      
{'loss': 0.1637, 'grad_norm': 0.4466150403022766, 'learning_rate': 4.682384950061367e-06, 'epoch': 1.7}

 57%|█████▋    | 7567/13317 [1:18:01<58:50,  1.63it/s]
 57%|█████▋    | 7568/13317 [1:18:01<58:49,  1.63it/s]
                                                      
{'loss': 0.176, 'grad_norm': 0.5260393619537354, 'learning_rate': 4.6810769727478546e-06, 'epoch': 1.7}

 57%|█████▋    | 7568/13317 [1:18:02<58:49,  1.63it/s]
 57%|█████▋    | 7569/13317 [1:18:02<58:47,  1.63it/s]
                                                      
{'loss': 0.1893, 'grad_norm': 0.5363708138465881, 'learning_rate': 4.679769017347731e-06, 'epoch': 1.71}

 57%|█████▋    | 7569/13317 [1:18:02<58:47,  1.63it/s]
 57%|█████▋    | 7570/13317 [1:18:03<58:47,  1.63it/s]
                                                      
{'loss': 0.1741, 'grad_norm': 0.47653743624687195, 'learning_rate': 4.678461083950867e-06, 'epoch': 1.71}

 57%|█████▋    | 7570/13317 [1:18:03<58:47,  1.63it/s]
 57%|█████▋    | 7571/13317 [1:18:03<58:44,  1.63it/s]
                                                      
{'loss': 0.2384, 'grad_norm': 0.5562980771064758, 'learning_rate': 4.677153172647131e-06, 'epoch': 1.71}

 57%|█████▋    | 7571/13317 [1:18:03<58:44,  1.63it/s]
 57%|█████▋    | 7572/13317 [1:18:04<58:47,  1.63it/s]
                                                      
{'loss': 0.1975, 'grad_norm': 0.5792875289916992, 'learning_rate': 4.67584528352639e-06, 'epoch': 1.71}

 57%|█████▋    | 7572/13317 [1:18:04<58:47,  1.63it/s]
 57%|█████▋    | 7573/13317 [1:18:04<58:46,  1.63it/s]
                                                      
{'loss': 0.175, 'grad_norm': 0.49497538805007935, 'learning_rate': 4.67453741667851e-06, 'epoch': 1.71}

 57%|█████▋    | 7573/13317 [1:18:05<58:46,  1.63it/s]
 57%|█████▋    | 7574/13317 [1:18:05<58:47,  1.63it/s]
                                                      
{'loss': 0.1784, 'grad_norm': 0.45896610617637634, 'learning_rate': 4.673229572193357e-06, 'epoch': 1.71}

 57%|█████▋    | 7574/13317 [1:18:05<58:47,  1.63it/s]
 57%|█████▋    | 7575/13317 [1:18:06<58:45,  1.63it/s]
                                                      
{'loss': 0.2049, 'grad_norm': 0.6015190482139587, 'learning_rate': 4.671921750160791e-06, 'epoch': 1.71}

 57%|█████▋    | 7575/13317 [1:18:06<58:45,  1.63it/s]
 57%|█████▋    | 7576/13317 [1:18:06<58:44,  1.63it/s]
                                                      
{'loss': 0.1785, 'grad_norm': 0.49465304613113403, 'learning_rate': 4.6706139506706746e-06, 'epoch': 1.71}

 57%|█████▋    | 7576/13317 [1:18:06<58:44,  1.63it/s]
 57%|█████▋    | 7577/13317 [1:18:07<58:43,  1.63it/s]
                                                      
{'loss': 0.1833, 'grad_norm': 0.5088003277778625, 'learning_rate': 4.6693061738128685e-06, 'epoch': 1.71}

 57%|█████▋    | 7577/13317 [1:18:07<58:43,  1.63it/s]
 57%|█████▋    | 7578/13317 [1:18:08<58:42,  1.63it/s]
                                                      
{'loss': 0.1691, 'grad_norm': 0.5019683241844177, 'learning_rate': 4.667998419677229e-06, 'epoch': 1.71}

 57%|█████▋    | 7578/13317 [1:18:08<58:42,  1.63it/s]
 57%|█████▋    | 7579/13317 [1:18:08<58:41,  1.63it/s]
                                                      
{'loss': 0.1682, 'grad_norm': 0.46237799525260925, 'learning_rate': 4.6666906883536115e-06, 'epoch': 1.71}

 57%|█████▋    | 7579/13317 [1:18:08<58:41,  1.63it/s]
 57%|█████▋    | 7580/13317 [1:18:09<58:40,  1.63it/s]
                                                      
{'loss': 0.2582, 'grad_norm': 0.6269586086273193, 'learning_rate': 4.665382979931875e-06, 'epoch': 1.71}

 57%|█████▋    | 7580/13317 [1:18:09<58:40,  1.63it/s]
 57%|█████▋    | 7581/13317 [1:18:09<58:38,  1.63it/s]
                                                      
{'loss': 0.2068, 'grad_norm': 0.5620818734169006, 'learning_rate': 4.664075294501871e-06, 'epoch': 1.71}

 57%|█████▋    | 7581/13317 [1:18:10<58:38,  1.63it/s]
 57%|█████▋    | 7582/13317 [1:18:10<58:38,  1.63it/s]
                                                      
{'loss': 0.201, 'grad_norm': 0.5608657598495483, 'learning_rate': 4.66276763215345e-06, 'epoch': 1.71}

 57%|█████▋    | 7582/13317 [1:18:10<58:38,  1.63it/s]
 57%|█████▋    | 7583/13317 [1:18:11<58:38,  1.63it/s]
                                                      
{'loss': 0.1882, 'grad_norm': 0.48427608609199524, 'learning_rate': 4.661459992976463e-06, 'epoch': 1.71}

 57%|█████▋    | 7583/13317 [1:18:11<58:38,  1.63it/s]
 57%|█████▋    | 7584/13317 [1:18:11<58:37,  1.63it/s]
                                                      
{'loss': 0.1471, 'grad_norm': 0.3734530806541443, 'learning_rate': 4.660152377060761e-06, 'epoch': 1.71}

 57%|█████▋    | 7584/13317 [1:18:11<58:37,  1.63it/s]
 57%|█████▋    | 7585/13317 [1:18:12<58:35,  1.63it/s]
                                                      
{'loss': 0.1472, 'grad_norm': 0.4121442139148712, 'learning_rate': 4.658844784496188e-06, 'epoch': 1.71}

 57%|█████▋    | 7585/13317 [1:18:12<58:35,  1.63it/s]
 57%|█████▋    | 7586/13317 [1:18:12<58:35,  1.63it/s]
                                                      
{'loss': 0.2128, 'grad_norm': 0.522624671459198, 'learning_rate': 4.6575372153725885e-06, 'epoch': 1.71}

 57%|█████▋    | 7586/13317 [1:18:13<58:35,  1.63it/s]
 57%|█████▋    | 7587/13317 [1:18:13<58:37,  1.63it/s]
                                                      
{'loss': 0.2146, 'grad_norm': 0.5344205498695374, 'learning_rate': 4.65622966977981e-06, 'epoch': 1.71}

 57%|█████▋    | 7587/13317 [1:18:13<58:37,  1.63it/s]
 57%|█████▋    | 7588/13317 [1:18:14<58:36,  1.63it/s]
                                                      
{'loss': 0.2029, 'grad_norm': 0.5268179774284363, 'learning_rate': 4.654922147807694e-06, 'epoch': 1.71}

 57%|█████▋    | 7588/13317 [1:18:14<58:36,  1.63it/s]
 57%|█████▋    | 7589/13317 [1:18:14<58:34,  1.63it/s]
                                                      
{'loss': 0.1936, 'grad_norm': 0.5148314833641052, 'learning_rate': 4.6536146495460795e-06, 'epoch': 1.71}

 57%|█████▋    | 7589/13317 [1:18:14<58:34,  1.63it/s]
 57%|█████▋    | 7590/13317 [1:18:15<58:35,  1.63it/s]
                                                      
{'loss': 0.1819, 'grad_norm': 0.5194433927536011, 'learning_rate': 4.6523071750848056e-06, 'epoch': 1.71}

 57%|█████▋    | 7590/13317 [1:18:15<58:35,  1.63it/s]
 57%|█████▋    | 7591/13317 [1:18:16<58:34,  1.63it/s]
                                                      
{'loss': 0.1879, 'grad_norm': 0.6030069589614868, 'learning_rate': 4.650999724513711e-06, 'epoch': 1.71}

 57%|█████▋    | 7591/13317 [1:18:16<58:34,  1.63it/s]
 57%|█████▋    | 7592/13317 [1:18:16<58:34,  1.63it/s]
                                                      
{'loss': 0.2656, 'grad_norm': 0.5734964609146118, 'learning_rate': 4.6496922979226315e-06, 'epoch': 1.71}

 57%|█████▋    | 7592/13317 [1:18:16<58:34,  1.63it/s]
 57%|█████▋    | 7593/13317 [1:18:17<58:33,  1.63it/s]
                                                      
{'loss': 0.2152, 'grad_norm': 0.5979860424995422, 'learning_rate': 4.648384895401397e-06, 'epoch': 1.71}

 57%|█████▋    | 7593/13317 [1:18:17<58:33,  1.63it/s]
 57%|█████▋    | 7594/13317 [1:18:17<58:29,  1.63it/s]
                                                      
{'loss': 0.217, 'grad_norm': 0.5590647459030151, 'learning_rate': 4.6470775170398466e-06, 'epoch': 1.71}

 57%|█████▋    | 7594/13317 [1:18:17<58:29,  1.63it/s]
 57%|█████▋    | 7595/13317 [1:18:18<58:29,  1.63it/s]
                                                      
{'loss': 0.1924, 'grad_norm': 0.507840633392334, 'learning_rate': 4.645770162927806e-06, 'epoch': 1.71}

 57%|█████▋    | 7595/13317 [1:18:18<58:29,  1.63it/s]
 57%|█████▋    | 7596/13317 [1:18:19<58:29,  1.63it/s]
                                                      
{'loss': 0.201, 'grad_norm': 0.5925747156143188, 'learning_rate': 4.644462833155106e-06, 'epoch': 1.71}

 57%|█████▋    | 7596/13317 [1:18:19<58:29,  1.63it/s]
 57%|█████▋    | 7597/13317 [1:18:19<58:29,  1.63it/s]
                                                      
{'loss': 0.1401, 'grad_norm': 0.4004266560077667, 'learning_rate': 4.643155527811572e-06, 'epoch': 1.71}

 57%|█████▋    | 7597/13317 [1:18:19<58:29,  1.63it/s]
 57%|█████▋    | 7598/13317 [1:18:20<58:30,  1.63it/s]
                                                      
{'loss': 0.2113, 'grad_norm': 0.558968722820282, 'learning_rate': 4.641848246987036e-06, 'epoch': 1.71}

 57%|█████▋    | 7598/13317 [1:18:20<58:30,  1.63it/s]
 57%|█████▋    | 7599/13317 [1:18:20<58:29,  1.63it/s]
                                                      
{'loss': 0.2048, 'grad_norm': 0.5068298578262329, 'learning_rate': 4.640540990771313e-06, 'epoch': 1.71}

 57%|█████▋    | 7599/13317 [1:18:21<58:29,  1.63it/s]
 57%|█████▋    | 7600/13317 [1:18:21<58:27,  1.63it/s]
                                                      
{'loss': 0.2034, 'grad_norm': 0.4968954622745514, 'learning_rate': 4.639233759254232e-06, 'epoch': 1.71}

 57%|█████▋    | 7600/13317 [1:18:21<58:27,  1.63it/s]
 57%|█████▋    | 7601/13317 [1:18:22<58:25,  1.63it/s]
                                                      
{'loss': 0.1488, 'grad_norm': 0.4349064528942108, 'learning_rate': 4.637926552525613e-06, 'epoch': 1.71}

 57%|█████▋    | 7601/13317 [1:18:22<58:25,  1.63it/s]
 57%|█████▋    | 7602/13317 [1:18:22<58:25,  1.63it/s]
                                                      
{'loss': 0.1573, 'grad_norm': 0.4875854253768921, 'learning_rate': 4.636619370675272e-06, 'epoch': 1.71}

 57%|█████▋    | 7602/13317 [1:18:22<58:25,  1.63it/s]
 57%|█████▋    | 7603/13317 [1:18:23<58:24,  1.63it/s]
                                                      
{'loss': 0.1773, 'grad_norm': 0.4725230634212494, 'learning_rate': 4.6353122137930275e-06, 'epoch': 1.71}

 57%|█████▋    | 7603/13317 [1:18:23<58:24,  1.63it/s]
 57%|█████▋    | 7604/13317 [1:18:24<58:24,  1.63it/s]
                                                      
{'loss': 0.1931, 'grad_norm': 0.4536490738391876, 'learning_rate': 4.634005081968697e-06, 'epoch': 1.71}

 57%|█████▋    | 7604/13317 [1:18:24<58:24,  1.63it/s]
 57%|█████▋    | 7605/13317 [1:18:24<58:24,  1.63it/s]
                                                      
{'loss': 0.1574, 'grad_norm': 0.4430537223815918, 'learning_rate': 4.6326979752920905e-06, 'epoch': 1.71}

 57%|█████▋    | 7605/13317 [1:18:24<58:24,  1.63it/s]
 57%|█████▋    | 7606/13317 [1:18:25<58:23,  1.63it/s]
                                                      
{'loss': 0.2157, 'grad_norm': 0.6023006439208984, 'learning_rate': 4.631390893853022e-06, 'epoch': 1.71}

 57%|█████▋    | 7606/13317 [1:18:25<58:23,  1.63it/s]
 57%|█████▋    | 7607/13317 [1:18:25<58:23,  1.63it/s]
                                                      
{'loss': 0.1646, 'grad_norm': 0.43545055389404297, 'learning_rate': 4.630083837741302e-06, 'epoch': 1.71}

 57%|█████▋    | 7607/13317 [1:18:25<58:23,  1.63it/s]
 57%|█████▋    | 7608/13317 [1:18:26<58:22,  1.63it/s]
                                                      
{'loss': 0.1454, 'grad_norm': 0.47355717420578003, 'learning_rate': 4.628776807046741e-06, 'epoch': 1.71}

 57%|█████▋    | 7608/13317 [1:18:26<58:22,  1.63it/s]
 57%|█████▋    | 7609/13317 [1:18:27<58:24,  1.63it/s]
                                                      
{'loss': 0.1938, 'grad_norm': 0.5836262106895447, 'learning_rate': 4.627469801859143e-06, 'epoch': 1.71}

 57%|█████▋    | 7609/13317 [1:18:27<58:24,  1.63it/s]
 57%|█████▋    | 7610/13317 [1:18:27<58:23,  1.63it/s]
                                                      
{'loss': 0.2283, 'grad_norm': 0.6124023199081421, 'learning_rate': 4.626162822268314e-06, 'epoch': 1.71}

 57%|█████▋    | 7610/13317 [1:18:27<58:23,  1.63it/s]
 57%|█████▋    | 7611/13317 [1:18:28<58:24,  1.63it/s]
                                                      
{'loss': 0.1748, 'grad_norm': 0.5229192972183228, 'learning_rate': 4.624855868364057e-06, 'epoch': 1.71}

 57%|█████▋    | 7611/13317 [1:18:28<58:24,  1.63it/s]
 57%|█████▋    | 7612/13317 [1:18:28<58:21,  1.63it/s]
                                                      
{'loss': 0.1491, 'grad_norm': 0.39992275834083557, 'learning_rate': 4.623548940236175e-06, 'epoch': 1.71}

 57%|█████▋    | 7612/13317 [1:18:29<58:21,  1.63it/s]
 57%|█████▋    | 7613/13317 [1:18:29<58:20,  1.63it/s]
                                                      
{'loss': 0.2464, 'grad_norm': 0.5567113161087036, 'learning_rate': 4.622242037974465e-06, 'epoch': 1.72}

 57%|█████▋    | 7613/13317 [1:18:29<58:20,  1.63it/s]
 57%|█████▋    | 7614/13317 [1:18:30<58:21,  1.63it/s]
                                                      
{'loss': 0.2391, 'grad_norm': 0.6036406755447388, 'learning_rate': 4.620935161668727e-06, 'epoch': 1.72}

 57%|█████▋    | 7614/13317 [1:18:30<58:21,  1.63it/s]
 57%|█████▋    | 7615/13317 [1:18:30<58:19,  1.63it/s]
                                                      
{'loss': 0.1933, 'grad_norm': 0.5423818230628967, 'learning_rate': 4.619628311408758e-06, 'epoch': 1.72}

 57%|█████▋    | 7615/13317 [1:18:30<58:19,  1.63it/s]
 57%|█████▋    | 7616/13317 [1:18:31<58:18,  1.63it/s]
                                                      
{'loss': 0.1498, 'grad_norm': 0.4154386818408966, 'learning_rate': 4.6183214872843515e-06, 'epoch': 1.72}

 57%|█████▋    | 7616/13317 [1:18:31<58:18,  1.63it/s]
 57%|█████▋    | 7617/13317 [1:18:31<58:17,  1.63it/s]
                                                      
{'loss': 0.1532, 'grad_norm': 0.4330771267414093, 'learning_rate': 4.6170146893852994e-06, 'epoch': 1.72}

 57%|█████▋    | 7617/13317 [1:18:32<58:17,  1.63it/s]
 57%|█████▋    | 7618/13317 [1:18:32<58:16,  1.63it/s]
                                                      
{'loss': 0.1934, 'grad_norm': 0.5876789093017578, 'learning_rate': 4.615707917801395e-06, 'epoch': 1.72}

 57%|█████▋    | 7618/13317 [1:18:32<58:16,  1.63it/s]
 57%|█████▋    | 7619/13317 [1:18:33<58:15,  1.63it/s]
                                                      
{'loss': 0.2186, 'grad_norm': 0.60025554895401, 'learning_rate': 4.614401172622423e-06, 'epoch': 1.72}

 57%|█████▋    | 7619/13317 [1:18:33<58:15,  1.63it/s]
 57%|█████▋    | 7620/13317 [1:18:33<58:14,  1.63it/s]
                                                      
{'loss': 0.2149, 'grad_norm': 0.4928227961063385, 'learning_rate': 4.6130944539381746e-06, 'epoch': 1.72}

 57%|█████▋    | 7620/13317 [1:18:33<58:14,  1.63it/s]
 57%|█████▋    | 7621/13317 [1:18:34<58:13,  1.63it/s]
                                                      
{'loss': 0.1349, 'grad_norm': 0.3910715878009796, 'learning_rate': 4.611787761838434e-06, 'epoch': 1.72}

 57%|█████▋    | 7621/13317 [1:18:34<58:13,  1.63it/s]
 57%|█████▋    | 7622/13317 [1:18:35<58:12,  1.63it/s]
                                                      
{'loss': 0.1663, 'grad_norm': 0.4435305893421173, 'learning_rate': 4.610481096412985e-06, 'epoch': 1.72}

 57%|█████▋    | 7622/13317 [1:18:35<58:12,  1.63it/s]
 57%|█████▋    | 7623/13317 [1:18:35<58:13,  1.63it/s]
                                                      
{'loss': 0.1673, 'grad_norm': 0.5134963393211365, 'learning_rate': 4.609174457751608e-06, 'epoch': 1.72}

 57%|█████▋    | 7623/13317 [1:18:35<58:13,  1.63it/s]
 57%|█████▋    | 7624/13317 [1:18:36<58:12,  1.63it/s]
                                                      
{'loss': 0.2364, 'grad_norm': 0.6621041297912598, 'learning_rate': 4.607867845944082e-06, 'epoch': 1.72}

 57%|█████▋    | 7624/13317 [1:18:36<58:12,  1.63it/s]
 57%|█████▋    | 7625/13317 [1:18:36<58:11,  1.63it/s]
                                                      
{'loss': 0.2025, 'grad_norm': 0.5060814023017883, 'learning_rate': 4.606561261080191e-06, 'epoch': 1.72}

 57%|█████▋    | 7625/13317 [1:18:37<58:11,  1.63it/s]
 57%|█████▋    | 7626/13317 [1:18:37<58:10,  1.63it/s]
                                                      
{'loss': 0.1643, 'grad_norm': 0.4579577147960663, 'learning_rate': 4.605254703249704e-06, 'epoch': 1.72}

 57%|█████▋    | 7626/13317 [1:18:37<58:10,  1.63it/s]
 57%|█████▋    | 7627/13317 [1:18:38<58:10,  1.63it/s]
                                                      
{'loss': 0.1917, 'grad_norm': 0.5055303573608398, 'learning_rate': 4.6039481725424e-06, 'epoch': 1.72}

 57%|█████▋    | 7627/13317 [1:18:38<58:10,  1.63it/s]
 57%|█████▋    | 7628/13317 [1:18:38<58:10,  1.63it/s]
                                                      
{'loss': 0.1995, 'grad_norm': 0.5129362940788269, 'learning_rate': 4.6026416690480505e-06, 'epoch': 1.72}

 57%|█████▋    | 7628/13317 [1:18:38<58:10,  1.63it/s]
 57%|█████▋    | 7629/13317 [1:18:39<58:10,  1.63it/s]
                                                      
{'loss': 0.1831, 'grad_norm': 0.4920901954174042, 'learning_rate': 4.601335192856425e-06, 'epoch': 1.72}

 57%|█████▋    | 7629/13317 [1:18:39<58:10,  1.63it/s]
 57%|█████▋    | 7630/13317 [1:18:39<58:07,  1.63it/s]
                                                      
{'loss': 0.1926, 'grad_norm': 0.496494859457016, 'learning_rate': 4.6000287440572925e-06, 'epoch': 1.72}

 57%|█████▋    | 7630/13317 [1:18:40<58:07,  1.63it/s]
 57%|█████▋    | 7631/13317 [1:18:40<58:09,  1.63it/s]
                                                      
{'loss': 0.189, 'grad_norm': 0.4437755048274994, 'learning_rate': 4.5987223227404194e-06, 'epoch': 1.72}

 57%|█████▋    | 7631/13317 [1:18:40<58:09,  1.63it/s]
 57%|█████▋    | 7632/13317 [1:18:41<58:07,  1.63it/s]
                                                      
{'loss': 0.2123, 'grad_norm': 0.6187020540237427, 'learning_rate': 4.597415928995574e-06, 'epoch': 1.72}

 57%|█████▋    | 7632/13317 [1:18:41<58:07,  1.63it/s]
 57%|█████▋    | 7633/13317 [1:18:41<58:06,  1.63it/s]
                                                      
{'loss': 0.2081, 'grad_norm': 0.592962920665741, 'learning_rate': 4.596109562912515e-06, 'epoch': 1.72}

 57%|█████▋    | 7633/13317 [1:18:41<58:06,  1.63it/s]
 57%|█████▋    | 7634/13317 [1:18:42<58:04,  1.63it/s]
                                                      
{'loss': 0.2396, 'grad_norm': 0.5902127027511597, 'learning_rate': 4.594803224581006e-06, 'epoch': 1.72}

 57%|█████▋    | 7634/13317 [1:18:42<58:04,  1.63it/s]
 57%|█████▋    | 7635/13317 [1:18:43<58:04,  1.63it/s]
                                                      
{'loss': 0.1638, 'grad_norm': 0.5358203053474426, 'learning_rate': 4.593496914090807e-06, 'epoch': 1.72}

 57%|█████▋    | 7635/13317 [1:18:43<58:04,  1.63it/s]
 57%|█████▋    | 7636/13317 [1:18:43<58:03,  1.63it/s]
                                                      
{'loss': 0.1812, 'grad_norm': 0.5090360641479492, 'learning_rate': 4.592190631531672e-06, 'epoch': 1.72}

 57%|█████▋    | 7636/13317 [1:18:43<58:03,  1.63it/s]
 57%|█████▋    | 7637/13317 [1:18:44<58:03,  1.63it/s]
                                                      
{'loss': 0.1598, 'grad_norm': 0.4925321936607361, 'learning_rate': 4.59088437699336e-06, 'epoch': 1.72}

 57%|█████▋    | 7637/13317 [1:18:44<58:03,  1.63it/s]
 57%|█████▋    | 7638/13317 [1:18:44<58:03,  1.63it/s]
                                                      
{'loss': 0.1753, 'grad_norm': 0.46240338683128357, 'learning_rate': 4.589578150565623e-06, 'epoch': 1.72}

 57%|█████▋    | 7638/13317 [1:18:44<58:03,  1.63it/s]
 57%|█████▋    | 7639/13317 [1:18:45<58:03,  1.63it/s]
                                                      
{'loss': 0.1824, 'grad_norm': 0.43542563915252686, 'learning_rate': 4.588271952338212e-06, 'epoch': 1.72}

 57%|█████▋    | 7639/13317 [1:18:45<58:03,  1.63it/s]
 57%|█████▋    | 7640/13317 [1:18:46<58:04,  1.63it/s]
                                                      
{'loss': 0.1789, 'grad_norm': 0.45190948247909546, 'learning_rate': 4.586965782400877e-06, 'epoch': 1.72}

 57%|█████▋    | 7640/13317 [1:18:46<58:04,  1.63it/s]
 57%|█████▋    | 7641/13317 [1:18:46<58:04,  1.63it/s]
                                                      
{'loss': 0.2138, 'grad_norm': 0.5975815057754517, 'learning_rate': 4.585659640843366e-06, 'epoch': 1.72}

 57%|█████▋    | 7641/13317 [1:18:46<58:04,  1.63it/s]
 57%|█████▋    | 7642/13317 [1:18:47<58:00,  1.63it/s]
                                                      
{'loss': 0.1558, 'grad_norm': 0.47067660093307495, 'learning_rate': 4.5843535277554265e-06, 'epoch': 1.72}

 57%|█████▋    | 7642/13317 [1:18:47<58:00,  1.63it/s]
 57%|█████▋    | 7643/13317 [1:18:47<57:59,  1.63it/s]
                                                      
{'loss': 0.2417, 'grad_norm': 0.592093825340271, 'learning_rate': 4.583047443226799e-06, 'epoch': 1.72}

 57%|█████▋    | 7643/13317 [1:18:48<57:59,  1.63it/s]
 57%|█████▋    | 7644/13317 [1:18:48<57:58,  1.63it/s]
                                                      
{'loss': 0.2326, 'grad_norm': 0.635246217250824, 'learning_rate': 4.581741387347227e-06, 'epoch': 1.72}

 57%|█████▋    | 7644/13317 [1:18:48<57:58,  1.63it/s]
 57%|█████▋    | 7645/13317 [1:18:49<57:57,  1.63it/s]
                                                      
{'loss': 0.1971, 'grad_norm': 0.4923061728477478, 'learning_rate': 4.580435360206452e-06, 'epoch': 1.72}

 57%|█████▋    | 7645/13317 [1:18:49<57:57,  1.63it/s]
 57%|█████▋    | 7646/13317 [1:18:49<57:58,  1.63it/s]
                                                      
{'loss': 0.1504, 'grad_norm': 0.4431741237640381, 'learning_rate': 4.579129361894208e-06, 'epoch': 1.72}

 57%|█████▋    | 7646/13317 [1:18:49<57:58,  1.63it/s]
 57%|█████▋    | 7647/13317 [1:18:50<57:57,  1.63it/s]
                                                      
{'loss': 0.2072, 'grad_norm': 0.5789255499839783, 'learning_rate': 4.577823392500234e-06, 'epoch': 1.72}

 57%|█████▋    | 7647/13317 [1:18:50<57:57,  1.63it/s]
 57%|█████▋    | 7648/13317 [1:18:50<57:57,  1.63it/s]
                                                      
{'loss': 0.182, 'grad_norm': 0.5183234214782715, 'learning_rate': 4.576517452114262e-06, 'epoch': 1.72}

 57%|█████▋    | 7648/13317 [1:18:51<57:57,  1.63it/s]
 57%|█████▋    | 7649/13317 [1:18:51<57:55,  1.63it/s]
                                                      
{'loss': 0.1898, 'grad_norm': 0.5074979066848755, 'learning_rate': 4.575211540826026e-06, 'epoch': 1.72}

 57%|█████▋    | 7649/13317 [1:18:51<57:55,  1.63it/s]
 57%|█████▋    | 7650/13317 [1:18:52<57:56,  1.63it/s]
                                                      
{'loss': 0.1737, 'grad_norm': 0.48497918248176575, 'learning_rate': 4.573905658725255e-06, 'epoch': 1.72}

 57%|█████▋    | 7650/13317 [1:18:52<57:56,  1.63it/s]
 57%|█████▋    | 7651/13317 [1:18:52<57:54,  1.63it/s]
                                                      
{'loss': 0.2242, 'grad_norm': 0.5450247526168823, 'learning_rate': 4.572599805901676e-06, 'epoch': 1.72}

 57%|█████▋    | 7651/13317 [1:18:52<57:54,  1.63it/s]
 57%|█████▋    | 7652/13317 [1:18:53<57:53,  1.63it/s]
                                                      
{'loss': 0.1893, 'grad_norm': 0.543199360370636, 'learning_rate': 4.571293982445018e-06, 'epoch': 1.72}

 57%|█████▋    | 7652/13317 [1:18:53<57:53,  1.63it/s]
 57%|█████▋    | 7653/13317 [1:18:54<57:56,  1.63it/s]
                                                      
{'loss': 0.1678, 'grad_norm': 0.5124589800834656, 'learning_rate': 4.569988188445e-06, 'epoch': 1.72}

 57%|█████▋    | 7653/13317 [1:18:54<57:56,  1.63it/s]
 57%|█████▋    | 7654/13317 [1:18:54<57:55,  1.63it/s]
                                                      
{'loss': 0.1466, 'grad_norm': 0.4022868275642395, 'learning_rate': 4.5686824239913465e-06, 'epoch': 1.72}

 57%|█████▋    | 7654/13317 [1:18:54<57:55,  1.63it/s]
 57%|█████▋    | 7655/13317 [1:18:55<57:55,  1.63it/s]
                                                      
{'loss': 0.2496, 'grad_norm': 0.6323570013046265, 'learning_rate': 4.56737668917378e-06, 'epoch': 1.72}

 57%|█████▋    | 7655/13317 [1:18:55<57:55,  1.63it/s]
 57%|█████▋    | 7656/13317 [1:18:55<57:55,  1.63it/s]
                                                      
{'loss': 0.1842, 'grad_norm': 0.4593401849269867, 'learning_rate': 4.566070984082013e-06, 'epoch': 1.72}

 57%|█████▋    | 7656/13317 [1:18:56<57:55,  1.63it/s]
 57%|█████▋    | 7657/13317 [1:18:56<57:53,  1.63it/s]
                                                      
{'loss': 0.1725, 'grad_norm': 0.505265474319458, 'learning_rate': 4.5647653088057645e-06, 'epoch': 1.72}

 57%|█████▋    | 7657/13317 [1:18:56<57:53,  1.63it/s]
 58%|█████▊    | 7658/13317 [1:18:57<57:52,  1.63it/s]
                                                      
{'loss': 0.1827, 'grad_norm': 0.5147578716278076, 'learning_rate': 4.563459663434748e-06, 'epoch': 1.73}

 58%|█████▊    | 7658/13317 [1:18:57<57:52,  1.63it/s]
 58%|█████▊    | 7659/13317 [1:18:57<57:53,  1.63it/s]
                                                      
{'loss': 0.1794, 'grad_norm': 0.521977961063385, 'learning_rate': 4.562154048058676e-06, 'epoch': 1.73}

 58%|█████▊    | 7659/13317 [1:18:57<57:53,  1.63it/s]
 58%|█████▊    | 7660/13317 [1:18:58<57:51,  1.63it/s]
                                                      
{'loss': 0.1688, 'grad_norm': 0.46413078904151917, 'learning_rate': 4.560848462767254e-06, 'epoch': 1.73}

 58%|█████▊    | 7660/13317 [1:18:58<57:51,  1.63it/s]
 58%|█████▊    | 7661/13317 [1:18:58<57:49,  1.63it/s]
                                                      
{'loss': 0.1735, 'grad_norm': 0.4557688236236572, 'learning_rate': 4.559542907650194e-06, 'epoch': 1.73}

 58%|█████▊    | 7661/13317 [1:18:59<57:49,  1.63it/s]
 58%|█████▊    | 7662/13317 [1:18:59<57:49,  1.63it/s]
                                                      
{'loss': 0.177, 'grad_norm': 0.46723252534866333, 'learning_rate': 4.558237382797201e-06, 'epoch': 1.73}

 58%|█████▊    | 7662/13317 [1:18:59<57:49,  1.63it/s]
 58%|█████▊    | 7663/13317 [1:19:00<57:47,  1.63it/s]
                                                      
{'loss': 0.1845, 'grad_norm': 0.48944970965385437, 'learning_rate': 4.556931888297976e-06, 'epoch': 1.73}

 58%|█████▊    | 7663/13317 [1:19:00<57:47,  1.63it/s]
 58%|█████▊    | 7664/13317 [1:19:00<57:48,  1.63it/s]
                                                      
{'loss': 0.2149, 'grad_norm': 0.5661089420318604, 'learning_rate': 4.555626424242221e-06, 'epoch': 1.73}

 58%|█████▊    | 7664/13317 [1:19:00<57:48,  1.63it/s]
 58%|█████▊    | 7665/13317 [1:19:01<57:47,  1.63it/s]
                                                      
{'loss': 0.2027, 'grad_norm': 0.6022225618362427, 'learning_rate': 4.554320990719638e-06, 'epoch': 1.73}

 58%|█████▊    | 7665/13317 [1:19:01<57:47,  1.63it/s]
 58%|█████▊    | 7666/13317 [1:19:02<57:46,  1.63it/s]
                                                      
{'loss': 0.1555, 'grad_norm': 0.4616997539997101, 'learning_rate': 4.5530155878199205e-06, 'epoch': 1.73}

 58%|█████▊    | 7666/13317 [1:19:02<57:46,  1.63it/s]
 58%|█████▊    | 7667/13317 [1:19:02<57:44,  1.63it/s]
                                                      
{'loss': 0.1909, 'grad_norm': 0.4861512780189514, 'learning_rate': 4.551710215632763e-06, 'epoch': 1.73}

 58%|█████▊    | 7667/13317 [1:19:02<57:44,  1.63it/s]
 58%|█████▊    | 7668/13317 [1:19:03<57:46,  1.63it/s]
                                                      
{'loss': 0.1615, 'grad_norm': 0.4444162845611572, 'learning_rate': 4.5504048742478615e-06, 'epoch': 1.73}

 58%|█████▊    | 7668/13317 [1:19:03<57:46,  1.63it/s]
 58%|█████▊    | 7669/13317 [1:19:03<57:45,  1.63it/s]
                                                      
{'loss': 0.16, 'grad_norm': 0.45373913645744324, 'learning_rate': 4.549099563754907e-06, 'epoch': 1.73}

 58%|█████▊    | 7669/13317 [1:19:04<57:45,  1.63it/s]
 58%|█████▊    | 7670/13317 [1:19:04<57:44,  1.63it/s]
                                                      
{'loss': 0.1641, 'grad_norm': 0.4308755099773407, 'learning_rate': 4.547794284243585e-06, 'epoch': 1.73}

 58%|█████▊    | 7670/13317 [1:19:04<57:44,  1.63it/s]
 58%|█████▊    | 7671/13317 [1:19:05<57:43,  1.63it/s]
                                                      
{'loss': 0.198, 'grad_norm': 0.8607664704322815, 'learning_rate': 4.546489035803583e-06, 'epoch': 1.73}

 58%|█████▊    | 7671/13317 [1:19:05<57:43,  1.63it/s]
 58%|█████▊    | 7672/13317 [1:19:05<57:44,  1.63it/s]
                                                      
{'loss': 0.1767, 'grad_norm': 0.5109912157058716, 'learning_rate': 4.54518381852459e-06, 'epoch': 1.73}

 58%|█████▊    | 7672/13317 [1:19:05<57:44,  1.63it/s]
 58%|█████▊    | 7673/13317 [1:19:06<57:42,  1.63it/s]
                                                      
{'loss': 0.2025, 'grad_norm': 0.6479805111885071, 'learning_rate': 4.54387863249628e-06, 'epoch': 1.73}

 58%|█████▊    | 7673/13317 [1:19:06<57:42,  1.63it/s]
 58%|█████▊    | 7674/13317 [1:19:06<57:42,  1.63it/s]
                                                      
{'loss': 0.1832, 'grad_norm': 0.4892086088657379, 'learning_rate': 4.542573477808339e-06, 'epoch': 1.73}

 58%|█████▊    | 7674/13317 [1:19:07<57:42,  1.63it/s]
 58%|█████▊    | 7675/13317 [1:19:07<57:41,  1.63it/s]
                                                      
{'loss': 0.1979, 'grad_norm': 0.5459135174751282, 'learning_rate': 4.541268354550444e-06, 'epoch': 1.73}

 58%|█████▊    | 7675/13317 [1:19:07<57:41,  1.63it/s]
 58%|█████▊    | 7676/13317 [1:19:08<57:40,  1.63it/s]
                                                      
{'loss': 0.1928, 'grad_norm': 0.5316866040229797, 'learning_rate': 4.53996326281227e-06, 'epoch': 1.73}

 58%|█████▊    | 7676/13317 [1:19:08<57:40,  1.63it/s]
 58%|█████▊    | 7677/13317 [1:19:08<57:39,  1.63it/s]
                                                      
{'loss': 0.2556, 'grad_norm': 0.6130155920982361, 'learning_rate': 4.53865820268349e-06, 'epoch': 1.73}

 58%|█████▊    | 7677/13317 [1:19:08<57:39,  1.63it/s]
 58%|█████▊    | 7678/13317 [1:19:09<57:39,  1.63it/s]
                                                      
{'loss': 0.2495, 'grad_norm': 0.5863503813743591, 'learning_rate': 4.537353174253777e-06, 'epoch': 1.73}

 58%|█████▊    | 7678/13317 [1:19:09<57:39,  1.63it/s]
 58%|█████▊    | 7679/13317 [1:19:10<57:39,  1.63it/s]
                                                      
{'loss': 0.2016, 'grad_norm': 0.56179279088974, 'learning_rate': 4.5360481776128e-06, 'epoch': 1.73}

 58%|█████▊    | 7679/13317 [1:19:10<57:39,  1.63it/s]
 58%|█████▊    | 7680/13317 [1:19:10<57:39,  1.63it/s]
                                                      
{'loss': 0.2001, 'grad_norm': 0.5642024874687195, 'learning_rate': 4.534743212850224e-06, 'epoch': 1.73}

 58%|█████▊    | 7680/13317 [1:19:10<57:39,  1.63it/s]
 58%|█████▊    | 7681/13317 [1:19:11<57:36,  1.63it/s]
                                                      
{'loss': 0.1807, 'grad_norm': 0.47169286012649536, 'learning_rate': 4.533438280055717e-06, 'epoch': 1.73}

 58%|█████▊    | 7681/13317 [1:19:11<57:36,  1.63it/s]
 58%|█████▊    | 7682/13317 [1:19:11<57:37,  1.63it/s]
                                                      
{'loss': 0.2257, 'grad_norm': 0.57949298620224, 'learning_rate': 4.53213337931894e-06, 'epoch': 1.73}

 58%|█████▊    | 7682/13317 [1:19:11<57:37,  1.63it/s]
 58%|█████▊    | 7683/13317 [1:19:12<57:36,  1.63it/s]
                                                      
{'loss': 0.2278, 'grad_norm': 0.5346466302871704, 'learning_rate': 4.530828510729554e-06, 'epoch': 1.73}

 58%|█████▊    | 7683/13317 [1:19:12<57:36,  1.63it/s]
 58%|█████▊    | 7684/13317 [1:19:13<57:36,  1.63it/s]
                                                      
{'loss': 0.176, 'grad_norm': 0.46503588557243347, 'learning_rate': 4.529523674377216e-06, 'epoch': 1.73}

 58%|█████▊    | 7684/13317 [1:19:13<57:36,  1.63it/s]
 58%|█████▊    | 7685/13317 [1:19:13<57:33,  1.63it/s]
                                                      
{'loss': 0.1509, 'grad_norm': 0.40517717599868774, 'learning_rate': 4.5282188703515815e-06, 'epoch': 1.73}

 58%|█████▊    | 7685/13317 [1:19:13<57:33,  1.63it/s]
 58%|█████▊    | 7686/13317 [1:19:14<57:34,  1.63it/s]
                                                      
{'loss': 0.2032, 'grad_norm': 0.5810286402702332, 'learning_rate': 4.52691409874231e-06, 'epoch': 1.73}

 58%|█████▊    | 7686/13317 [1:19:14<57:34,  1.63it/s]
 58%|█████▊    | 7687/13317 [1:19:14<57:33,  1.63it/s]
                                                      
{'loss': 0.2045, 'grad_norm': 0.5135610699653625, 'learning_rate': 4.525609359639045e-06, 'epoch': 1.73}

 58%|█████▊    | 7687/13317 [1:19:15<57:33,  1.63it/s]
 58%|█████▊    | 7688/13317 [1:19:15<57:34,  1.63it/s]
                                                      
{'loss': 0.1617, 'grad_norm': 0.44319960474967957, 'learning_rate': 4.524304653131442e-06, 'epoch': 1.73}

 58%|█████▊    | 7688/13317 [1:19:15<57:34,  1.63it/s]
 58%|█████▊    | 7689/13317 [1:19:16<57:32,  1.63it/s]
                                                      
{'loss': 0.1588, 'grad_norm': 0.45862114429473877, 'learning_rate': 4.5229999793091465e-06, 'epoch': 1.73}

 58%|█████▊    | 7689/13317 [1:19:16<57:32,  1.63it/s]
 58%|█████▊    | 7690/13317 [1:19:16<57:32,  1.63it/s]
                                                      
{'loss': 0.2496, 'grad_norm': 0.6511412262916565, 'learning_rate': 4.521695338261802e-06, 'epoch': 1.73}

 58%|█████▊    | 7690/13317 [1:19:16<57:32,  1.63it/s]
 58%|█████▊    | 7691/13317 [1:19:17<57:29,  1.63it/s]
                                                      
{'loss': 0.1792, 'grad_norm': 0.4691077172756195, 'learning_rate': 4.520390730079052e-06, 'epoch': 1.73}

 58%|█████▊    | 7691/13317 [1:19:17<57:29,  1.63it/s]
 58%|█████▊    | 7692/13317 [1:19:17<57:28,  1.63it/s]
                                                      
{'loss': 0.168, 'grad_norm': 0.5112218856811523, 'learning_rate': 4.519086154850536e-06, 'epoch': 1.73}

 58%|█████▊    | 7692/13317 [1:19:18<57:28,  1.63it/s]
 58%|█████▊    | 7693/13317 [1:19:18<57:29,  1.63it/s]
                                                      
{'loss': 0.1962, 'grad_norm': 0.4561102092266083, 'learning_rate': 4.517781612665896e-06, 'epoch': 1.73}

 58%|█████▊    | 7693/13317 [1:19:18<57:29,  1.63it/s]
 58%|█████▊    | 7694/13317 [1:19:19<57:28,  1.63it/s]
                                                      
{'loss': 0.2252, 'grad_norm': 0.5534678101539612, 'learning_rate': 4.516477103614763e-06, 'epoch': 1.73}

 58%|█████▊    | 7694/13317 [1:19:19<57:28,  1.63it/s]
 58%|█████▊    | 7695/13317 [1:19:19<57:28,  1.63it/s]
                                                      
{'loss': 0.1374, 'grad_norm': 0.4293347895145416, 'learning_rate': 4.515172627786773e-06, 'epoch': 1.73}

 58%|█████▊    | 7695/13317 [1:19:19<57:28,  1.63it/s]
 58%|█████▊    | 7696/13317 [1:19:20<57:32,  1.63it/s]
                                                      
{'loss': 0.1575, 'grad_norm': 0.4587951898574829, 'learning_rate': 4.513868185271558e-06, 'epoch': 1.73}

 58%|█████▊    | 7696/13317 [1:19:20<57:32,  1.63it/s]
 58%|█████▊    | 7697/13317 [1:19:21<57:30,  1.63it/s]
                                                      
{'loss': 0.1827, 'grad_norm': 0.44381019473075867, 'learning_rate': 4.512563776158745e-06, 'epoch': 1.73}

 58%|█████▊    | 7697/13317 [1:19:21<57:30,  1.63it/s]
 58%|█████▊    | 7698/13317 [1:19:21<57:30,  1.63it/s]
                                                      
{'loss': 0.227, 'grad_norm': 0.552191972732544, 'learning_rate': 4.51125940053796e-06, 'epoch': 1.73}

 58%|█████▊    | 7698/13317 [1:19:21<57:30,  1.63it/s]
 58%|█████▊    | 7699/13317 [1:19:22<57:28,  1.63it/s]
                                                      
{'loss': 0.2435, 'grad_norm': 0.6081753969192505, 'learning_rate': 4.509955058498833e-06, 'epoch': 1.73}

 58%|█████▊    | 7699/13317 [1:19:22<57:28,  1.63it/s]
 58%|█████▊    | 7700/13317 [1:19:22<57:27,  1.63it/s]
                                                      
{'loss': 0.2, 'grad_norm': 0.5497374534606934, 'learning_rate': 4.508650750130978e-06, 'epoch': 1.73}

 58%|█████▊    | 7700/13317 [1:19:23<57:27,  1.63it/s]
 58%|█████▊    | 7701/13317 [1:19:23<57:25,  1.63it/s]
                                                      
{'loss': 0.159, 'grad_norm': 0.44966456294059753, 'learning_rate': 4.507346475524021e-06, 'epoch': 1.73}

 58%|█████▊    | 7701/13317 [1:19:23<57:25,  1.63it/s]
 58%|█████▊    | 7702/13317 [1:19:24<57:25,  1.63it/s]
                                                      
{'loss': 0.1784, 'grad_norm': 0.48988303542137146, 'learning_rate': 4.506042234767577e-06, 'epoch': 1.74}

 58%|█████▊    | 7702/13317 [1:19:24<57:25,  1.63it/s]
 58%|█████▊    | 7703/13317 [1:19:24<57:25,  1.63it/s]
                                                      
{'loss': 0.1921, 'grad_norm': 0.5061683058738708, 'learning_rate': 4.504738027951263e-06, 'epoch': 1.74}

 58%|█████▊    | 7703/13317 [1:19:24<57:25,  1.63it/s]
 58%|█████▊    | 7704/13317 [1:19:25<57:23,  1.63it/s]
                                                      
{'loss': 0.1782, 'grad_norm': 0.45027053356170654, 'learning_rate': 4.503433855164689e-06, 'epoch': 1.74}

 58%|█████▊    | 7704/13317 [1:19:25<57:23,  1.63it/s]
 58%|█████▊    | 7705/13317 [1:19:25<57:23,  1.63it/s]
                                                      
{'loss': 0.2094, 'grad_norm': 0.5888687968254089, 'learning_rate': 4.502129716497466e-06, 'epoch': 1.74}

 58%|█████▊    | 7705/13317 [1:19:26<57:23,  1.63it/s]
 58%|█████▊    | 7706/13317 [1:19:26<57:22,  1.63it/s]
                                                      
{'loss': 0.2015, 'grad_norm': 0.5097532272338867, 'learning_rate': 4.500825612039206e-06, 'epoch': 1.74}

 58%|█████▊    | 7706/13317 [1:19:26<57:22,  1.63it/s]
 58%|█████▊    | 7707/13317 [1:19:27<57:21,  1.63it/s]
                                                      
{'loss': 0.1921, 'grad_norm': 0.4887372851371765, 'learning_rate': 4.499521541879508e-06, 'epoch': 1.74}

 58%|█████▊    | 7707/13317 [1:19:27<57:21,  1.63it/s]
 58%|█████▊    | 7708/13317 [1:19:27<57:20,  1.63it/s]
                                                      
{'loss': 0.1932, 'grad_norm': 0.541205108165741, 'learning_rate': 4.498217506107982e-06, 'epoch': 1.74}

 58%|█████▊    | 7708/13317 [1:19:27<57:20,  1.63it/s]
 58%|█████▊    | 7709/13317 [1:19:28<57:19,  1.63it/s]
                                                      
{'loss': 0.1829, 'grad_norm': 0.48372793197631836, 'learning_rate': 4.496913504814226e-06, 'epoch': 1.74}

 58%|█████▊    | 7709/13317 [1:19:28<57:19,  1.63it/s]
 58%|█████▊    | 7710/13317 [1:19:29<57:20,  1.63it/s]
                                                      
{'loss': 0.2299, 'grad_norm': 0.5303306579589844, 'learning_rate': 4.4956095380878385e-06, 'epoch': 1.74}

 58%|█████▊    | 7710/13317 [1:19:29<57:20,  1.63it/s]
 58%|█████▊    | 7711/13317 [1:19:29<57:20,  1.63it/s]
                                                      
{'loss': 0.1647, 'grad_norm': 0.4949881434440613, 'learning_rate': 4.494305606018416e-06, 'epoch': 1.74}

 58%|█████▊    | 7711/13317 [1:19:29<57:20,  1.63it/s]
 58%|█████▊    | 7712/13317 [1:19:30<57:25,  1.63it/s]
                                                      
{'loss': 0.1964, 'grad_norm': 0.6295156478881836, 'learning_rate': 4.493001708695551e-06, 'epoch': 1.74}

 58%|█████▊    | 7712/13317 [1:19:30<57:25,  1.63it/s]
 58%|█████▊    | 7713/13317 [1:19:30<57:23,  1.63it/s]
                                                      
{'loss': 0.2169, 'grad_norm': 0.6498062610626221, 'learning_rate': 4.491697846208841e-06, 'epoch': 1.74}

 58%|█████▊    | 7713/13317 [1:19:31<57:23,  1.63it/s]
 58%|█████▊    | 7714/13317 [1:19:31<57:21,  1.63it/s]
                                                      
{'loss': 0.1735, 'grad_norm': 0.4477301239967346, 'learning_rate': 4.490394018647868e-06, 'epoch': 1.74}

 58%|█████▊    | 7714/13317 [1:19:31<57:21,  1.63it/s]
 58%|█████▊    | 7715/13317 [1:19:32<57:19,  1.63it/s]
                                                      
{'loss': 0.2278, 'grad_norm': 0.5581810474395752, 'learning_rate': 4.4890902261022215e-06, 'epoch': 1.74}

 58%|█████▊    | 7715/13317 [1:19:32<57:19,  1.63it/s]
 58%|█████▊    | 7716/13317 [1:19:32<57:17,  1.63it/s]
                                                      
{'loss': 0.2784, 'grad_norm': 0.6265382170677185, 'learning_rate': 4.487786468661488e-06, 'epoch': 1.74}

 58%|█████▊    | 7716/13317 [1:19:32<57:17,  1.63it/s]
 58%|█████▊    | 7717/13317 [1:19:33<57:15,  1.63it/s]
                                                      
{'loss': 0.2006, 'grad_norm': 0.5510770678520203, 'learning_rate': 4.486482746415247e-06, 'epoch': 1.74}

 58%|█████▊    | 7717/13317 [1:19:33<57:15,  1.63it/s]
 58%|█████▊    | 7718/13317 [1:19:33<57:16,  1.63it/s]
                                                      
{'loss': 0.2219, 'grad_norm': 0.6165611147880554, 'learning_rate': 4.485179059453077e-06, 'epoch': 1.74}

 58%|█████▊    | 7718/13317 [1:19:34<57:16,  1.63it/s]
 58%|█████▊    | 7719/13317 [1:19:34<57:14,  1.63it/s]
                                                      
{'loss': 0.2295, 'grad_norm': 0.6469324231147766, 'learning_rate': 4.483875407864557e-06, 'epoch': 1.74}

 58%|█████▊    | 7719/13317 [1:19:34<57:14,  1.63it/s]
 58%|█████▊    | 7720/13317 [1:19:35<57:15,  1.63it/s]
                                                      
{'loss': 0.2097, 'grad_norm': 0.5212697982788086, 'learning_rate': 4.482571791739262e-06, 'epoch': 1.74}

 58%|█████▊    | 7720/13317 [1:19:35<57:15,  1.63it/s]
 58%|█████▊    | 7721/13317 [1:19:35<57:13,  1.63it/s]
                                                      
{'loss': 0.2485, 'grad_norm': 0.6091662049293518, 'learning_rate': 4.481268211166764e-06, 'epoch': 1.74}

 58%|█████▊    | 7721/13317 [1:19:35<57:13,  1.63it/s]
 58%|█████▊    | 7722/13317 [1:19:36<57:14,  1.63it/s]
                                                      
{'loss': 0.2218, 'grad_norm': 0.5514115691184998, 'learning_rate': 4.479964666236631e-06, 'epoch': 1.74}

 58%|█████▊    | 7722/13317 [1:19:36<57:14,  1.63it/s]
 58%|█████▊    | 7723/13317 [1:19:37<57:13,  1.63it/s]
                                                      
{'loss': 0.2222, 'grad_norm': 0.7011690735816956, 'learning_rate': 4.4786611570384346e-06, 'epoch': 1.74}

 58%|█████▊    | 7723/13317 [1:19:37<57:13,  1.63it/s]
 58%|█████▊    | 7724/13317 [1:19:37<57:12,  1.63it/s]
                                                      
{'loss': 0.2697, 'grad_norm': 0.666415274143219, 'learning_rate': 4.477357683661734e-06, 'epoch': 1.74}

 58%|█████▊    | 7724/13317 [1:19:37<57:12,  1.63it/s]
 58%|█████▊    | 7725/13317 [1:19:38<57:10,  1.63it/s]
                                                      
{'loss': 0.1821, 'grad_norm': 0.44610536098480225, 'learning_rate': 4.476054246196093e-06, 'epoch': 1.74}

 58%|█████▊    | 7725/13317 [1:19:38<57:10,  1.63it/s]
 58%|█████▊    | 7726/13317 [1:19:38<57:10,  1.63it/s]
                                                      
{'loss': 0.1608, 'grad_norm': 0.43565067648887634, 'learning_rate': 4.4747508447310774e-06, 'epoch': 1.74}

 58%|█████▊    | 7726/13317 [1:19:38<57:10,  1.63it/s]
 58%|█████▊    | 7727/13317 [1:19:39<57:10,  1.63it/s]
                                                      
{'loss': 0.2411, 'grad_norm': 0.6590805649757385, 'learning_rate': 4.473447479356236e-06, 'epoch': 1.74}

 58%|█████▊    | 7727/13317 [1:19:39<57:10,  1.63it/s]
 58%|█████▊    | 7728/13317 [1:19:40<57:09,  1.63it/s]
                                                      
{'loss': 0.2328, 'grad_norm': 0.5694292783737183, 'learning_rate': 4.47214415016113e-06, 'epoch': 1.74}

 58%|█████▊    | 7728/13317 [1:19:40<57:09,  1.63it/s]
 58%|█████▊    | 7729/13317 [1:19:40<57:07,  1.63it/s]
                                                      
{'loss': 0.157, 'grad_norm': 0.4141363501548767, 'learning_rate': 4.470840857235309e-06, 'epoch': 1.74}

 58%|█████▊    | 7729/13317 [1:19:40<57:07,  1.63it/s]
 58%|█████▊    | 7730/13317 [1:19:41<57:06,  1.63it/s]
                                                      
{'loss': 0.2166, 'grad_norm': 0.5911993980407715, 'learning_rate': 4.469537600668325e-06, 'epoch': 1.74}

 58%|█████▊    | 7730/13317 [1:19:41<57:06,  1.63it/s]
 58%|█████▊    | 7731/13317 [1:19:41<57:06,  1.63it/s]
                                                      
{'loss': 0.1928, 'grad_norm': 0.52330482006073, 'learning_rate': 4.468234380549724e-06, 'epoch': 1.74}

 58%|█████▊    | 7731/13317 [1:19:42<57:06,  1.63it/s]
 58%|█████▊    | 7732/13317 [1:19:42<57:06,  1.63it/s]
                                                      
{'loss': 0.1914, 'grad_norm': 0.49128633737564087, 'learning_rate': 4.46693119696905e-06, 'epoch': 1.74}

 58%|█████▊    | 7732/13317 [1:19:42<57:06,  1.63it/s]
 58%|█████▊    | 7733/13317 [1:19:43<57:05,  1.63it/s]
                                                      
{'loss': 0.1996, 'grad_norm': 0.49739786982536316, 'learning_rate': 4.4656280500158505e-06, 'epoch': 1.74}

 58%|█████▊    | 7733/13317 [1:19:43<57:05,  1.63it/s]
 58%|█████▊    | 7734/13317 [1:19:43<57:03,  1.63it/s]
                                                      
{'loss': 0.2048, 'grad_norm': 0.49206337332725525, 'learning_rate': 4.464324939779659e-06, 'epoch': 1.74}

 58%|█████▊    | 7734/13317 [1:19:43<57:03,  1.63it/s]
 58%|█████▊    | 7735/13317 [1:19:44<57:00,  1.63it/s]
                                                      
{'loss': 0.1773, 'grad_norm': 0.4544154703617096, 'learning_rate': 4.463021866350017e-06, 'epoch': 1.74}

 58%|█████▊    | 7735/13317 [1:19:44<57:00,  1.63it/s]
 58%|█████▊    | 7736/13317 [1:19:44<57:01,  1.63it/s]
                                                      
{'loss': 0.1851, 'grad_norm': 0.6107618808746338, 'learning_rate': 4.461718829816458e-06, 'epoch': 1.74}

 58%|█████▊    | 7736/13317 [1:19:45<57:01,  1.63it/s]
 58%|█████▊    | 7737/13317 [1:19:45<57:01,  1.63it/s]
                                                      
{'loss': 0.1694, 'grad_norm': 0.4959004819393158, 'learning_rate': 4.460415830268515e-06, 'epoch': 1.74}

 58%|█████▊    | 7737/13317 [1:19:45<57:01,  1.63it/s]
 58%|█████▊    | 7738/13317 [1:19:46<57:02,  1.63it/s]
                                                      
{'loss': 0.2152, 'grad_norm': 0.516842782497406, 'learning_rate': 4.4591128677957176e-06, 'epoch': 1.74}

 58%|█████▊    | 7738/13317 [1:19:46<57:02,  1.63it/s]
 58%|█████▊    | 7739/13317 [1:19:46<57:02,  1.63it/s]
                                                      
{'loss': 0.1861, 'grad_norm': 0.5756517648696899, 'learning_rate': 4.457809942487593e-06, 'epoch': 1.74}

 58%|█████▊    | 7739/13317 [1:19:46<57:02,  1.63it/s]
 58%|█████▊    | 7740/13317 [1:19:47<57:02,  1.63it/s]
                                                      
{'loss': 0.1883, 'grad_norm': 0.47235843539237976, 'learning_rate': 4.456507054433668e-06, 'epoch': 1.74}

 58%|█████▊    | 7740/13317 [1:19:47<57:02,  1.63it/s]
 58%|█████▊    | 7741/13317 [1:19:48<57:00,  1.63it/s]
                                                      
{'loss': 0.179, 'grad_norm': 0.4462679624557495, 'learning_rate': 4.4552042037234596e-06, 'epoch': 1.74}

 58%|█████▊    | 7741/13317 [1:19:48<57:00,  1.63it/s]
 58%|█████▊    | 7742/13317 [1:19:48<57:00,  1.63it/s]
                                                      
{'loss': 0.1755, 'grad_norm': 0.451756089925766, 'learning_rate': 4.453901390446493e-06, 'epoch': 1.74}

 58%|█████▊    | 7742/13317 [1:19:48<57:00,  1.63it/s]
 58%|█████▊    | 7743/13317 [1:19:49<56:59,  1.63it/s]
                                                      
{'loss': 0.1737, 'grad_norm': 0.46033886075019836, 'learning_rate': 4.452598614692283e-06, 'epoch': 1.74}

 58%|█████▊    | 7743/13317 [1:19:49<56:59,  1.63it/s]
 58%|█████▊    | 7744/13317 [1:19:49<56:58,  1.63it/s]
                                                      
{'loss': 0.1808, 'grad_norm': 0.5296460390090942, 'learning_rate': 4.4512958765503435e-06, 'epoch': 1.74}

 58%|█████▊    | 7744/13317 [1:19:50<56:58,  1.63it/s]
 58%|█████▊    | 7745/13317 [1:19:50<57:00,  1.63it/s]
                                                      
{'loss': 0.1638, 'grad_norm': 0.47447434067726135, 'learning_rate': 4.4499931761101875e-06, 'epoch': 1.74}

 58%|█████▊    | 7745/13317 [1:19:50<57:00,  1.63it/s]
 58%|█████▊    | 7746/13317 [1:19:51<56:59,  1.63it/s]
                                                      
{'loss': 0.199, 'grad_norm': 0.5433229804039001, 'learning_rate': 4.448690513461323e-06, 'epoch': 1.74}

 58%|█████▊    | 7746/13317 [1:19:51<56:59,  1.63it/s]
 58%|█████▊    | 7747/13317 [1:19:51<56:58,  1.63it/s]
                                                      
{'loss': 0.2115, 'grad_norm': 0.5255956649780273, 'learning_rate': 4.4473878886932584e-06, 'epoch': 1.75}

 58%|█████▊    | 7747/13317 [1:19:51<56:58,  1.63it/s]
 58%|█████▊    | 7748/13317 [1:19:52<56:57,  1.63it/s]
                                                      
{'loss': 0.2461, 'grad_norm': 0.6022489070892334, 'learning_rate': 4.446085301895497e-06, 'epoch': 1.75}

 58%|█████▊    | 7748/13317 [1:19:52<56:57,  1.63it/s]
 58%|█████▊    | 7749/13317 [1:19:52<56:55,  1.63it/s]
                                                      
{'loss': 0.1793, 'grad_norm': 0.4641408920288086, 'learning_rate': 4.44478275315754e-06, 'epoch': 1.75}

 58%|█████▊    | 7749/13317 [1:19:53<56:55,  1.63it/s]
 58%|█████▊    | 7750/13317 [1:19:53<56:53,  1.63it/s]
                                                      
{'loss': 0.1638, 'grad_norm': 0.4956471920013428, 'learning_rate': 4.4434802425688885e-06, 'epoch': 1.75}

 58%|█████▊    | 7750/13317 [1:19:53<56:53,  1.63it/s]
 58%|█████▊    | 7751/13317 [1:19:54<56:52,  1.63it/s]
                                                      
{'loss': 0.1834, 'grad_norm': 0.5058021545410156, 'learning_rate': 4.442177770219034e-06, 'epoch': 1.75}

 58%|█████▊    | 7751/13317 [1:19:54<56:52,  1.63it/s]
 58%|█████▊    | 7752/13317 [1:19:54<56:55,  1.63it/s]
                                                      
{'loss': 0.185, 'grad_norm': 0.5114395022392273, 'learning_rate': 4.4408753361974735e-06, 'epoch': 1.75}

 58%|█████▊    | 7752/13317 [1:19:54<56:55,  1.63it/s]
 58%|█████▊    | 7753/13317 [1:19:55<56:55,  1.63it/s]
                                                      
{'loss': 0.1871, 'grad_norm': 0.5347644686698914, 'learning_rate': 4.4395729405936984e-06, 'epoch': 1.75}

 58%|█████▊    | 7753/13317 [1:19:55<56:55,  1.63it/s]
 58%|█████▊    | 7754/13317 [1:19:56<56:55,  1.63it/s]
                                                      
{'loss': 0.193, 'grad_norm': 0.573581337928772, 'learning_rate': 4.4382705834971965e-06, 'epoch': 1.75}

 58%|█████▊    | 7754/13317 [1:19:56<56:55,  1.63it/s]
 58%|█████▊    | 7755/13317 [1:19:56<56:53,  1.63it/s]
                                                      
{'loss': 0.2864, 'grad_norm': 0.539984941482544, 'learning_rate': 4.4369682649974526e-06, 'epoch': 1.75}

 58%|█████▊    | 7755/13317 [1:19:56<56:53,  1.63it/s]
 58%|█████▊    | 7756/13317 [1:19:57<56:55,  1.63it/s]
                                                      
{'loss': 0.1427, 'grad_norm': 0.4453819990158081, 'learning_rate': 4.4356659851839505e-06, 'epoch': 1.75}

 58%|█████▊    | 7756/13317 [1:19:57<56:55,  1.63it/s]
 58%|█████▊    | 7757/13317 [1:19:57<56:52,  1.63it/s]
                                                      
{'loss': 0.231, 'grad_norm': 0.5753533244132996, 'learning_rate': 4.434363744146171e-06, 'epoch': 1.75}

 58%|█████▊    | 7757/13317 [1:19:58<56:52,  1.63it/s]
 58%|█████▊    | 7758/13317 [1:19:58<56:50,  1.63it/s]
                                                      
{'loss': 0.2001, 'grad_norm': 0.549559473991394, 'learning_rate': 4.433061541973591e-06, 'epoch': 1.75}

 58%|█████▊    | 7758/13317 [1:19:58<56:50,  1.63it/s]
 58%|█████▊    | 7759/13317 [1:19:59<56:48,  1.63it/s]
                                                      
{'loss': 0.2425, 'grad_norm': 0.6673786640167236, 'learning_rate': 4.431759378755684e-06, 'epoch': 1.75}

 58%|█████▊    | 7759/13317 [1:19:59<56:48,  1.63it/s]
 58%|█████▊    | 7760/13317 [1:19:59<56:47,  1.63it/s]
                                                      
{'loss': 0.144, 'grad_norm': 0.422685831785202, 'learning_rate': 4.430457254581928e-06, 'epoch': 1.75}

 58%|█████▊    | 7760/13317 [1:19:59<56:47,  1.63it/s]
 58%|█████▊    | 7761/13317 [1:20:00<56:46,  1.63it/s]
                                                      
{'loss': 0.1876, 'grad_norm': 0.5429623126983643, 'learning_rate': 4.429155169541785e-06, 'epoch': 1.75}

 58%|█████▊    | 7761/13317 [1:20:00<56:46,  1.63it/s]
 58%|█████▊    | 7762/13317 [1:20:00<56:47,  1.63it/s]
                                                      
{'loss': 0.1835, 'grad_norm': 0.5603286623954773, 'learning_rate': 4.427853123724729e-06, 'epoch': 1.75}

 58%|█████▊    | 7762/13317 [1:20:01<56:47,  1.63it/s]
 58%|█████▊    | 7763/13317 [1:20:01<56:47,  1.63it/s]
                                                      
{'loss': 0.2265, 'grad_norm': 0.5925089120864868, 'learning_rate': 4.42655111722022e-06, 'epoch': 1.75}

 58%|█████▊    | 7763/13317 [1:20:01<56:47,  1.63it/s]
 58%|█████▊    | 7764/13317 [1:20:02<56:45,  1.63it/s]
                                                      
{'loss': 0.1519, 'grad_norm': 0.49836409091949463, 'learning_rate': 4.425249150117722e-06, 'epoch': 1.75}

 58%|█████▊    | 7764/13317 [1:20:02<56:45,  1.63it/s]
 58%|█████▊    | 7765/13317 [1:20:02<56:46,  1.63it/s]
                                                      
{'loss': 0.1835, 'grad_norm': 0.49658098816871643, 'learning_rate': 4.423947222506692e-06, 'epoch': 1.75}

 58%|█████▊    | 7765/13317 [1:20:02<56:46,  1.63it/s]
 58%|█████▊    | 7766/13317 [1:20:03<56:46,  1.63it/s]
                                                      
{'loss': 0.1852, 'grad_norm': 0.5013461112976074, 'learning_rate': 4.422645334476587e-06, 'epoch': 1.75}

 58%|█████▊    | 7766/13317 [1:20:03<56:46,  1.63it/s]
 58%|█████▊    | 7767/13317 [1:20:04<56:44,  1.63it/s]
                                                      
{'loss': 0.1842, 'grad_norm': 0.572212815284729, 'learning_rate': 4.4213434861168625e-06, 'epoch': 1.75}

 58%|█████▊    | 7767/13317 [1:20:04<56:44,  1.63it/s]
 58%|█████▊    | 7768/13317 [1:20:04<56:44,  1.63it/s]
                                                      
{'loss': 0.2089, 'grad_norm': 0.5415985584259033, 'learning_rate': 4.420041677516965e-06, 'epoch': 1.75}

 58%|█████▊    | 7768/13317 [1:20:04<56:44,  1.63it/s]
 58%|█████▊    | 7769/13317 [1:20:05<56:44,  1.63it/s]
                                                      
{'loss': 0.2004, 'grad_norm': 0.47305765748023987, 'learning_rate': 4.418739908766346e-06, 'epoch': 1.75}

 58%|█████▊    | 7769/13317 [1:20:05<56:44,  1.63it/s]
 58%|█████▊    | 7770/13317 [1:20:05<56:44,  1.63it/s]
                                                      
{'loss': 0.247, 'grad_norm': 0.6548358201980591, 'learning_rate': 4.417438179954451e-06, 'epoch': 1.75}

 58%|█████▊    | 7770/13317 [1:20:05<56:44,  1.63it/s]
 58%|█████▊    | 7771/13317 [1:20:06<56:42,  1.63it/s]
                                                      
{'loss': 0.1459, 'grad_norm': 0.5253310799598694, 'learning_rate': 4.41613649117072e-06, 'epoch': 1.75}

 58%|█████▊    | 7771/13317 [1:20:06<56:42,  1.63it/s]
 58%|█████▊    | 7772/13317 [1:20:07<56:41,  1.63it/s]
                                                      
{'loss': 0.1906, 'grad_norm': 0.5154284238815308, 'learning_rate': 4.414834842504593e-06, 'epoch': 1.75}

 58%|█████▊    | 7772/13317 [1:20:07<56:41,  1.63it/s]
 58%|█████▊    | 7773/13317 [1:20:07<56:40,  1.63it/s]
                                                      
{'loss': 0.2167, 'grad_norm': 0.5098673701286316, 'learning_rate': 4.413533234045509e-06, 'epoch': 1.75}

 58%|█████▊    | 7773/13317 [1:20:07<56:40,  1.63it/s]
 58%|█████▊    | 7774/13317 [1:20:08<56:40,  1.63it/s]
                                                      
{'loss': 0.2224, 'grad_norm': 0.5433037877082825, 'learning_rate': 4.412231665882902e-06, 'epoch': 1.75}

 58%|█████▊    | 7774/13317 [1:20:08<56:40,  1.63it/s]
 58%|█████▊    | 7775/13317 [1:20:08<56:38,  1.63it/s]
                                                      
{'loss': 0.157, 'grad_norm': 0.459534227848053, 'learning_rate': 4.410930138106203e-06, 'epoch': 1.75}

 58%|█████▊    | 7775/13317 [1:20:09<56:38,  1.63it/s]
 58%|█████▊    | 7776/13317 [1:20:09<56:38,  1.63it/s]
                                                      
{'loss': 0.1743, 'grad_norm': 0.46267789602279663, 'learning_rate': 4.40962865080484e-06, 'epoch': 1.75}

 58%|█████▊    | 7776/13317 [1:20:09<56:38,  1.63it/s]
 58%|█████▊    | 7777/13317 [1:20:10<56:38,  1.63it/s]
                                                      
{'loss': 0.3121, 'grad_norm': 0.5518910884857178, 'learning_rate': 4.408327204068241e-06, 'epoch': 1.75}

 58%|█████▊    | 7777/13317 [1:20:10<56:38,  1.63it/s]
 58%|█████▊    | 7778/13317 [1:20:10<56:37,  1.63it/s]
                                                      
{'loss': 0.1914, 'grad_norm': 0.5416542291641235, 'learning_rate': 4.4070257979858275e-06, 'epoch': 1.75}

 58%|█████▊    | 7778/13317 [1:20:10<56:37,  1.63it/s]
 58%|█████▊    | 7779/13317 [1:20:11<56:37,  1.63it/s]
                                                      
{'loss': 0.2627, 'grad_norm': 0.6462382674217224, 'learning_rate': 4.405724432647018e-06, 'epoch': 1.75}

 58%|█████▊    | 7779/13317 [1:20:11<56:37,  1.63it/s]
 58%|█████▊    | 7780/13317 [1:20:11<56:36,  1.63it/s]
                                                      
{'loss': 0.2594, 'grad_norm': 0.5600117444992065, 'learning_rate': 4.4044231081412335e-06, 'epoch': 1.75}

 58%|█████▊    | 7780/13317 [1:20:12<56:36,  1.63it/s]
 58%|█████▊    | 7781/13317 [1:20:12<56:35,  1.63it/s]
                                                      
{'loss': 0.1539, 'grad_norm': 0.4182533621788025, 'learning_rate': 4.403121824557889e-06, 'epoch': 1.75}

 58%|█████▊    | 7781/13317 [1:20:12<56:35,  1.63it/s]
 58%|█████▊    | 7782/13317 [1:20:13<56:35,  1.63it/s]
                                                      
{'loss': 0.2499, 'grad_norm': 0.6678471565246582, 'learning_rate': 4.4018205819863935e-06, 'epoch': 1.75}

 58%|█████▊    | 7782/13317 [1:20:13<56:35,  1.63it/s]
 58%|█████▊    | 7783/13317 [1:20:13<56:37,  1.63it/s]
                                                      
{'loss': 0.2533, 'grad_norm': 0.622355043888092, 'learning_rate': 4.400519380516158e-06, 'epoch': 1.75}

 58%|█████▊    | 7783/13317 [1:20:13<56:37,  1.63it/s]
 58%|█████▊    | 7784/13317 [1:20:14<56:36,  1.63it/s]
                                                      
{'loss': 0.1721, 'grad_norm': 0.5804815888404846, 'learning_rate': 4.399218220236589e-06, 'epoch': 1.75}

 58%|█████▊    | 7784/13317 [1:20:14<56:36,  1.63it/s]
 58%|█████▊    | 7785/13317 [1:20:15<56:34,  1.63it/s]
                                                      
{'loss': 0.1545, 'grad_norm': 0.4344940483570099, 'learning_rate': 4.3979171012370896e-06, 'epoch': 1.75}

 58%|█████▊    | 7785/13317 [1:20:15<56:34,  1.63it/s]
 58%|█████▊    | 7786/13317 [1:20:15<56:33,  1.63it/s]
                                                      
{'loss': 0.164, 'grad_norm': 0.47852474451065063, 'learning_rate': 4.396616023607057e-06, 'epoch': 1.75}

 58%|█████▊    | 7786/13317 [1:20:15<56:33,  1.63it/s]
 58%|█████▊    | 7787/13317 [1:20:16<56:31,  1.63it/s]
                                                      
{'loss': 0.2359, 'grad_norm': 0.615575909614563, 'learning_rate': 4.395314987435898e-06, 'epoch': 1.75}

 58%|█████▊    | 7787/13317 [1:20:16<56:31,  1.63it/s]
 58%|█████▊    | 7788/13317 [1:20:16<56:32,  1.63it/s]
                                                      
{'loss': 0.2062, 'grad_norm': 0.5329055786132812, 'learning_rate': 4.394013992812997e-06, 'epoch': 1.75}

 58%|█████▊    | 7788/13317 [1:20:17<56:32,  1.63it/s]
 58%|█████▊    | 7789/13317 [1:20:17<56:33,  1.63it/s]
                                                      
{'loss': 0.1632, 'grad_norm': 0.4782421290874481, 'learning_rate': 4.392713039827753e-06, 'epoch': 1.75}

 58%|█████▊    | 7789/13317 [1:20:17<56:33,  1.63it/s]
 58%|█████▊    | 7790/13317 [1:20:18<56:32,  1.63it/s]
                                                      
{'loss': 0.1862, 'grad_norm': 0.5528939366340637, 'learning_rate': 4.3914121285695525e-06, 'epoch': 1.75}

 58%|█████▊    | 7790/13317 [1:20:18<56:32,  1.63it/s]
 59%|█████▊    | 7791/13317 [1:20:18<56:30,  1.63it/s]
                                                      
{'loss': 0.1887, 'grad_norm': 0.6117071509361267, 'learning_rate': 4.390111259127784e-06, 'epoch': 1.76}

 59%|█████▊    | 7791/13317 [1:20:18<56:30,  1.63it/s]
 59%|█████▊    | 7792/13317 [1:20:19<56:30,  1.63it/s]
                                                      
{'loss': 0.2104, 'grad_norm': 0.5018525123596191, 'learning_rate': 4.388810431591829e-06, 'epoch': 1.76}

 59%|█████▊    | 7792/13317 [1:20:19<56:30,  1.63it/s]
 59%|█████▊    | 7793/13317 [1:20:19<56:26,  1.63it/s]
                                                      
{'loss': 0.22, 'grad_norm': 0.590914785861969, 'learning_rate': 4.387509646051067e-06, 'epoch': 1.76}

 59%|█████▊    | 7793/13317 [1:20:20<56:26,  1.63it/s]
 59%|█████▊    | 7794/13317 [1:20:20<56:27,  1.63it/s]
                                                      
{'loss': 0.204, 'grad_norm': 0.5502113103866577, 'learning_rate': 4.3862089025948805e-06, 'epoch': 1.76}

 59%|█████▊    | 7794/13317 [1:20:20<56:27,  1.63it/s]
 59%|█████▊    | 7795/13317 [1:20:21<56:26,  1.63it/s]
                                                      
{'loss': 0.234, 'grad_norm': 0.5521612763404846, 'learning_rate': 4.384908201312639e-06, 'epoch': 1.76}

 59%|█████▊    | 7795/13317 [1:20:21<56:26,  1.63it/s]
 59%|█████▊    | 7796/13317 [1:20:21<56:25,  1.63it/s]
                                                      
{'loss': 0.1789, 'grad_norm': 0.46398743987083435, 'learning_rate': 4.3836075422937174e-06, 'epoch': 1.76}

 59%|█████▊    | 7796/13317 [1:20:21<56:25,  1.63it/s]
 59%|█████▊    | 7797/13317 [1:20:22<56:25,  1.63it/s]
                                                      
{'loss': 0.1853, 'grad_norm': 0.5393356084823608, 'learning_rate': 4.3823069256274845e-06, 'epoch': 1.76}

 59%|█████▊    | 7797/13317 [1:20:22<56:25,  1.63it/s]
 59%|█████▊    | 7798/13317 [1:20:23<56:26,  1.63it/s]
                                                      
{'loss': 0.193, 'grad_norm': 0.5064312815666199, 'learning_rate': 4.381006351403306e-06, 'epoch': 1.76}

 59%|█████▊    | 7798/13317 [1:20:23<56:26,  1.63it/s]
 59%|█████▊    | 7799/13317 [1:20:23<56:25,  1.63it/s]
                                                      
{'loss': 0.1581, 'grad_norm': 0.5717535614967346, 'learning_rate': 4.3797058197105435e-06, 'epoch': 1.76}

 59%|█████▊    | 7799/13317 [1:20:23<56:25,  1.63it/s]
 59%|█████▊    | 7800/13317 [1:20:24<56:23,  1.63it/s]
                                                      
{'loss': 0.2067, 'grad_norm': 0.5597580075263977, 'learning_rate': 4.37840533063856e-06, 'epoch': 1.76}

 59%|█████▊    | 7800/13317 [1:20:24<56:23,  1.63it/s]
 59%|█████▊    | 7801/13317 [1:20:24<56:23,  1.63it/s]
                                                      
{'loss': 0.1574, 'grad_norm': 0.4554064869880676, 'learning_rate': 4.377104884276713e-06, 'epoch': 1.76}

 59%|█████▊    | 7801/13317 [1:20:25<56:23,  1.63it/s]
 59%|█████▊    | 7802/13317 [1:20:25<56:24,  1.63it/s]
                                                      
{'loss': 0.1887, 'grad_norm': 0.5296183228492737, 'learning_rate': 4.375804480714355e-06, 'epoch': 1.76}

 59%|█████▊    | 7802/13317 [1:20:25<56:24,  1.63it/s]
 59%|█████▊    | 7803/13317 [1:20:26<56:22,  1.63it/s]
                                                      
{'loss': 0.2262, 'grad_norm': 0.6996018290519714, 'learning_rate': 4.374504120040839e-06, 'epoch': 1.76}

 59%|█████▊    | 7803/13317 [1:20:26<56:22,  1.63it/s]
 59%|█████▊    | 7804/13317 [1:20:26<56:20,  1.63it/s]
                                                      
{'loss': 0.2107, 'grad_norm': 0.5678831338882446, 'learning_rate': 4.373203802345512e-06, 'epoch': 1.76}

 59%|█████▊    | 7804/13317 [1:20:26<56:20,  1.63it/s]
 59%|█████▊    | 7805/13317 [1:20:27<56:19,  1.63it/s]
                                                      
{'loss': 0.1875, 'grad_norm': 0.5380297899246216, 'learning_rate': 4.371903527717721e-06, 'epoch': 1.76}

 59%|█████▊    | 7805/13317 [1:20:27<56:19,  1.63it/s]
 59%|█████▊    | 7806/13317 [1:20:27<56:18,  1.63it/s]
                                                      
{'loss': 0.1945, 'grad_norm': 0.533481240272522, 'learning_rate': 4.370603296246806e-06, 'epoch': 1.76}

 59%|█████▊    | 7806/13317 [1:20:28<56:18,  1.63it/s]
 59%|█████▊    | 7807/13317 [1:20:28<56:17,  1.63it/s]
                                                      
{'loss': 0.1809, 'grad_norm': 0.5605323314666748, 'learning_rate': 4.36930310802211e-06, 'epoch': 1.76}

 59%|█████▊    | 7807/13317 [1:20:28<56:17,  1.63it/s]
 59%|█████▊    | 7808/13317 [1:20:29<56:18,  1.63it/s]
                                                      
{'loss': 0.1936, 'grad_norm': 0.49652355909347534, 'learning_rate': 4.36800296313297e-06, 'epoch': 1.76}

 59%|█████▊    | 7808/13317 [1:20:29<56:18,  1.63it/s]
 59%|█████▊    | 7809/13317 [1:20:29<56:18,  1.63it/s]
                                                      
{'loss': 0.2454, 'grad_norm': 0.6008135080337524, 'learning_rate': 4.366702861668717e-06, 'epoch': 1.76}

 59%|█████▊    | 7809/13317 [1:20:29<56:18,  1.63it/s]
 59%|█████▊    | 7810/13317 [1:20:30<56:20,  1.63it/s]
                                                      
{'loss': 0.1694, 'grad_norm': 0.5075818300247192, 'learning_rate': 4.365402803718681e-06, 'epoch': 1.76}

 59%|█████▊    | 7810/13317 [1:20:30<56:20,  1.63it/s]
 59%|█████▊    | 7811/13317 [1:20:31<56:18,  1.63it/s]
                                                      
{'loss': 0.1684, 'grad_norm': 0.4602036774158478, 'learning_rate': 4.364102789372194e-06, 'epoch': 1.76}

 59%|█████▊    | 7811/13317 [1:20:31<56:18,  1.63it/s]
 59%|█████▊    | 7812/13317 [1:20:31<56:20,  1.63it/s]
                                                      
{'loss': 0.1885, 'grad_norm': 0.5773910880088806, 'learning_rate': 4.362802818718578e-06, 'epoch': 1.76}

 59%|█████▊    | 7812/13317 [1:20:31<56:20,  1.63it/s]
 59%|█████▊    | 7813/13317 [1:20:32<56:20,  1.63it/s]
                                                      
{'loss': 0.2767, 'grad_norm': 0.6814696192741394, 'learning_rate': 4.361502891847153e-06, 'epoch': 1.76}

 59%|█████▊    | 7813/13317 [1:20:32<56:20,  1.63it/s]
 59%|█████▊    | 7814/13317 [1:20:32<56:20,  1.63it/s]
                                                      
{'loss': 0.2167, 'grad_norm': 0.5000516176223755, 'learning_rate': 4.360203008847244e-06, 'epoch': 1.76}

 59%|█████▊    | 7814/13317 [1:20:32<56:20,  1.63it/s]
 59%|█████▊    | 7815/13317 [1:20:33<56:18,  1.63it/s]
                                                      
{'loss': 0.1685, 'grad_norm': 0.4994903802871704, 'learning_rate': 4.358903169808158e-06, 'epoch': 1.76}

 59%|█████▊    | 7815/13317 [1:20:33<56:18,  1.63it/s]
 59%|█████▊    | 7816/13317 [1:20:34<56:16,  1.63it/s]
                                                      
{'loss': 0.1908, 'grad_norm': 0.49843817949295044, 'learning_rate': 4.357603374819215e-06, 'epoch': 1.76}

 59%|█████▊    | 7816/13317 [1:20:34<56:16,  1.63it/s]
 59%|█████▊    | 7817/13317 [1:20:34<56:16,  1.63it/s]
                                                      
{'loss': 0.2454, 'grad_norm': 0.5873929858207703, 'learning_rate': 4.356303623969721e-06, 'epoch': 1.76}

 59%|█████▊    | 7817/13317 [1:20:34<56:16,  1.63it/s]
 59%|█████▊    | 7818/13317 [1:20:35<56:14,  1.63it/s]
                                                      
{'loss': 0.1885, 'grad_norm': 0.5403101444244385, 'learning_rate': 4.355003917348985e-06, 'epoch': 1.76}

 59%|█████▊    | 7818/13317 [1:20:35<56:14,  1.63it/s]
 59%|█████▊    | 7819/13317 [1:20:35<56:16,  1.63it/s]
                                                      
{'loss': 0.2146, 'grad_norm': 0.5385664701461792, 'learning_rate': 4.353704255046308e-06, 'epoch': 1.76}

 59%|█████▊    | 7819/13317 [1:20:36<56:16,  1.63it/s]
 59%|█████▊    | 7820/13317 [1:20:36<56:17,  1.63it/s]
                                                      
{'loss': 0.2092, 'grad_norm': 0.5370784401893616, 'learning_rate': 4.352404637150991e-06, 'epoch': 1.76}

 59%|█████▊    | 7820/13317 [1:20:36<56:17,  1.63it/s]
 59%|█████▊    | 7821/13317 [1:20:37<56:16,  1.63it/s]
                                                      
{'loss': 0.2059, 'grad_norm': 0.5595847964286804, 'learning_rate': 4.351105063752334e-06, 'epoch': 1.76}

 59%|█████▊    | 7821/13317 [1:20:37<56:16,  1.63it/s]
 59%|█████▊    | 7822/13317 [1:20:37<56:13,  1.63it/s]
                                                      
{'loss': 0.2128, 'grad_norm': 0.5160836577415466, 'learning_rate': 4.349805534939628e-06, 'epoch': 1.76}

 59%|█████▊    | 7822/13317 [1:20:37<56:13,  1.63it/s]
 59%|█████▊    | 7823/13317 [1:20:38<56:11,  1.63it/s]
                                                      
{'loss': 0.2207, 'grad_norm': 0.5672423839569092, 'learning_rate': 4.348506050802167e-06, 'epoch': 1.76}

 59%|█████▊    | 7823/13317 [1:20:38<56:11,  1.63it/s]
 59%|█████▉    | 7824/13317 [1:20:38<56:11,  1.63it/s]
                                                      
{'loss': 0.2182, 'grad_norm': 0.5768345594406128, 'learning_rate': 4.347206611429239e-06, 'epoch': 1.76}

 59%|█████▉    | 7824/13317 [1:20:39<56:11,  1.63it/s]
 59%|█████▉    | 7825/13317 [1:20:39<56:12,  1.63it/s]
                                                      
{'loss': 0.2345, 'grad_norm': 0.5349088907241821, 'learning_rate': 4.3459072169101285e-06, 'epoch': 1.76}

 59%|█████▉    | 7825/13317 [1:20:39<56:12,  1.63it/s]
 59%|█████▉    | 7826/13317 [1:20:40<56:13,  1.63it/s]
                                                      
{'loss': 0.1593, 'grad_norm': 0.46121397614479065, 'learning_rate': 4.344607867334116e-06, 'epoch': 1.76}

 59%|█████▉    | 7826/13317 [1:20:40<56:13,  1.63it/s]
 59%|█████▉    | 7827/13317 [1:20:40<56:18,  1.63it/s]
                                                      
{'loss': 0.1858, 'grad_norm': 0.4496390223503113, 'learning_rate': 4.343308562790484e-06, 'epoch': 1.76}

 59%|█████▉    | 7827/13317 [1:20:40<56:18,  1.63it/s]
 59%|█████▉    | 7828/13317 [1:20:41<56:13,  1.63it/s]
                                                      
{'loss': 0.2003, 'grad_norm': 0.5348097681999207, 'learning_rate': 4.342009303368508e-06, 'epoch': 1.76}

 59%|█████▉    | 7828/13317 [1:20:41<56:13,  1.63it/s]
 59%|█████▉    | 7829/13317 [1:20:42<56:10,  1.63it/s]
                                                      
{'loss': 0.1751, 'grad_norm': 0.5101079344749451, 'learning_rate': 4.340710089157459e-06, 'epoch': 1.76}

 59%|█████▉    | 7829/13317 [1:20:42<56:10,  1.63it/s]
 59%|█████▉    | 7830/13317 [1:20:42<56:07,  1.63it/s]
                                                      
{'loss': 0.1465, 'grad_norm': 0.474337637424469, 'learning_rate': 4.339410920246606e-06, 'epoch': 1.76}

 59%|█████▉    | 7830/13317 [1:20:42<56:07,  1.63it/s]
 59%|█████▉    | 7831/13317 [1:20:43<56:06,  1.63it/s]
                                                      
{'loss': 0.2061, 'grad_norm': 0.49553653597831726, 'learning_rate': 4.3381117967252195e-06, 'epoch': 1.76}

 59%|█████▉    | 7831/13317 [1:20:43<56:06,  1.63it/s]
 59%|█████▉    | 7832/13317 [1:20:43<56:10,  1.63it/s]
                                                      
{'loss': 0.1788, 'grad_norm': 0.4993652105331421, 'learning_rate': 4.336812718682559e-06, 'epoch': 1.76}

 59%|█████▉    | 7832/13317 [1:20:44<56:10,  1.63it/s]
 59%|█████▉    | 7833/13317 [1:20:44<56:09,  1.63it/s]
                                                      
{'loss': 0.244, 'grad_norm': 0.5563398599624634, 'learning_rate': 4.335513686207886e-06, 'epoch': 1.76}

 59%|█████▉    | 7833/13317 [1:20:44<56:09,  1.63it/s]
 59%|█████▉    | 7834/13317 [1:20:45<56:08,  1.63it/s]
                                                      
{'loss': 0.2199, 'grad_norm': 0.5552790760993958, 'learning_rate': 4.33421469939046e-06, 'epoch': 1.76}

 59%|█████▉    | 7834/13317 [1:20:45<56:08,  1.63it/s]
 59%|█████▉    | 7835/13317 [1:20:45<56:08,  1.63it/s]
                                                      
{'loss': 0.2139, 'grad_norm': 0.5677443742752075, 'learning_rate': 4.332915758319534e-06, 'epoch': 1.77}

 59%|█████▉    | 7835/13317 [1:20:45<56:08,  1.63it/s]
 59%|█████▉    | 7836/13317 [1:20:46<56:06,  1.63it/s]
                                                      
{'loss': 0.2106, 'grad_norm': 0.536497175693512, 'learning_rate': 4.331616863084358e-06, 'epoch': 1.77}

 59%|█████▉    | 7836/13317 [1:20:46<56:06,  1.63it/s]
 59%|█████▉    | 7837/13317 [1:20:46<56:09,  1.63it/s]
                                                      
{'loss': 0.1883, 'grad_norm': 0.5187280774116516, 'learning_rate': 4.3303180137741795e-06, 'epoch': 1.77}

 59%|█████▉    | 7837/13317 [1:20:47<56:09,  1.63it/s]
 59%|█████▉    | 7838/13317 [1:20:47<56:07,  1.63it/s]
                                                      
{'loss': 0.2464, 'grad_norm': 0.5989038944244385, 'learning_rate': 4.329019210478246e-06, 'epoch': 1.77}

 59%|█████▉    | 7838/13317 [1:20:47<56:07,  1.63it/s]
 59%|█████▉    | 7839/13317 [1:20:48<56:03,  1.63it/s]
                                                      
{'loss': 0.223, 'grad_norm': 0.5728517174720764, 'learning_rate': 4.327720453285796e-06, 'epoch': 1.77}

 59%|█████▉    | 7839/13317 [1:20:48<56:03,  1.63it/s]
 59%|█████▉    | 7840/13317 [1:20:48<56:05,  1.63it/s]
                                                      
{'loss': 0.22, 'grad_norm': 0.6874042749404907, 'learning_rate': 4.3264217422860685e-06, 'epoch': 1.77}

 59%|█████▉    | 7840/13317 [1:20:48<56:05,  1.63it/s]
 59%|█████▉    | 7841/13317 [1:20:49<56:02,  1.63it/s]
                                                      
{'loss': 0.1545, 'grad_norm': 0.4401213824748993, 'learning_rate': 4.3251230775683e-06, 'epoch': 1.77}

 59%|█████▉    | 7841/13317 [1:20:49<56:02,  1.63it/s]
 59%|█████▉    | 7842/13317 [1:20:50<56:06,  1.63it/s]
                                                      
{'loss': 0.1962, 'grad_norm': 0.5379769802093506, 'learning_rate': 4.323824459221723e-06, 'epoch': 1.77}

 59%|█████▉    | 7842/13317 [1:20:50<56:06,  1.63it/s]
 59%|█████▉    | 7843/13317 [1:20:50<56:03,  1.63it/s]
                                                      
{'loss': 0.201, 'grad_norm': 0.5441294312477112, 'learning_rate': 4.322525887335563e-06, 'epoch': 1.77}

 59%|█████▉    | 7843/13317 [1:20:50<56:03,  1.63it/s]
 59%|█████▉    | 7844/13317 [1:20:51<56:04,  1.63it/s]
                                                      
{'loss': 0.2005, 'grad_norm': 0.5094345808029175, 'learning_rate': 4.321227361999049e-06, 'epoch': 1.77}

 59%|█████▉    | 7844/13317 [1:20:51<56:04,  1.63it/s]
 59%|█████▉    | 7845/13317 [1:20:51<56:03,  1.63it/s]
                                                      
{'loss': 0.2217, 'grad_norm': 0.5449227094650269, 'learning_rate': 4.319928883301403e-06, 'epoch': 1.77}

 59%|█████▉    | 7845/13317 [1:20:52<56:03,  1.63it/s]
 59%|█████▉    | 7846/13317 [1:20:52<56:01,  1.63it/s]
                                                      
{'loss': 0.2185, 'grad_norm': 0.6501213908195496, 'learning_rate': 4.318630451331841e-06, 'epoch': 1.77}

 59%|█████▉    | 7846/13317 [1:20:52<56:01,  1.63it/s]
 59%|█████▉    | 7847/13317 [1:20:53<56:05,  1.63it/s]
                                                      
{'loss': 0.1739, 'grad_norm': 0.44365882873535156, 'learning_rate': 4.317332066179582e-06, 'epoch': 1.77}

 59%|█████▉    | 7847/13317 [1:20:53<56:05,  1.63it/s]
 59%|█████▉    | 7848/13317 [1:20:53<56:00,  1.63it/s]
                                                      
{'loss': 0.1916, 'grad_norm': 0.5006265640258789, 'learning_rate': 4.31603372793384e-06, 'epoch': 1.77}

 59%|█████▉    | 7848/13317 [1:20:53<56:00,  1.63it/s]
 59%|█████▉    | 7849/13317 [1:20:54<55:59,  1.63it/s]
                                                      
{'loss': 0.207, 'grad_norm': 0.5072212219238281, 'learning_rate': 4.314735436683822e-06, 'epoch': 1.77}

 59%|█████▉    | 7849/13317 [1:20:54<55:59,  1.63it/s]
 59%|█████▉    | 7850/13317 [1:20:54<56:00,  1.63it/s]
                                                      
{'loss': 0.1576, 'grad_norm': 0.4449755847454071, 'learning_rate': 4.313437192518734e-06, 'epoch': 1.77}

 59%|█████▉    | 7850/13317 [1:20:55<56:00,  1.63it/s]
 59%|█████▉    | 7851/13317 [1:20:55<55:56,  1.63it/s]
                                                      
{'loss': 0.2749, 'grad_norm': 0.70830237865448, 'learning_rate': 4.312138995527781e-06, 'epoch': 1.77}

 59%|█████▉    | 7851/13317 [1:20:55<55:56,  1.63it/s]
 59%|█████▉    | 7852/13317 [1:20:56<56:16,  1.62it/s]
                                                      
{'loss': 0.1688, 'grad_norm': 0.4626966714859009, 'learning_rate': 4.310840845800165e-06, 'epoch': 1.77}

 59%|█████▉    | 7852/13317 [1:20:56<56:16,  1.62it/s]
 59%|█████▉    | 7853/13317 [1:20:56<56:09,  1.62it/s]
                                                      
{'loss': 0.1966, 'grad_norm': 0.4431872069835663, 'learning_rate': 4.3095427434250756e-06, 'epoch': 1.77}

 59%|█████▉    | 7853/13317 [1:20:56<56:09,  1.62it/s]
 59%|█████▉    | 7854/13317 [1:20:57<56:05,  1.62it/s]
                                                      
{'loss': 0.1845, 'grad_norm': 0.44076839089393616, 'learning_rate': 4.308244688491712e-06, 'epoch': 1.77}

 59%|█████▉    | 7854/13317 [1:20:57<56:05,  1.62it/s]
 59%|█████▉    | 7855/13317 [1:20:58<56:07,  1.62it/s]
                                                      
{'loss': 0.1591, 'grad_norm': 0.45315271615982056, 'learning_rate': 4.306946681089264e-06, 'epoch': 1.77}

 59%|█████▉    | 7855/13317 [1:20:58<56:07,  1.62it/s]
 59%|█████▉    | 7856/13317 [1:20:58<56:01,  1.62it/s]
                                                      
{'loss': 0.3149, 'grad_norm': 0.6514309048652649, 'learning_rate': 4.3056487213069155e-06, 'epoch': 1.77}

 59%|█████▉    | 7856/13317 [1:20:58<56:01,  1.62it/s]
 59%|█████▉    | 7857/13317 [1:20:59<56:03,  1.62it/s]
                                                      
{'loss': 0.1738, 'grad_norm': 0.4294658899307251, 'learning_rate': 4.304350809233853e-06, 'epoch': 1.77}

 59%|█████▉    | 7857/13317 [1:20:59<56:03,  1.62it/s]
 59%|█████▉    | 7858/13317 [1:20:59<55:59,  1.63it/s]
                                                      
{'loss': 0.194, 'grad_norm': 0.4787995219230652, 'learning_rate': 4.303052944959254e-06, 'epoch': 1.77}

 59%|█████▉    | 7858/13317 [1:21:00<55:59,  1.63it/s]
 59%|█████▉    | 7859/13317 [1:21:00<55:55,  1.63it/s]
                                                      
{'loss': 0.1509, 'grad_norm': 0.45933231711387634, 'learning_rate': 4.301755128572299e-06, 'epoch': 1.77}

 59%|█████▉    | 7859/13317 [1:21:00<55:55,  1.63it/s]
 59%|█████▉    | 7860/13317 [1:21:01<55:56,  1.63it/s]
                                                      
{'loss': 0.1509, 'grad_norm': 0.4423893690109253, 'learning_rate': 4.300457360162158e-06, 'epoch': 1.77}

 59%|█████▉    | 7860/13317 [1:21:01<55:56,  1.63it/s]
 59%|█████▉    | 7861/13317 [1:21:01<55:52,  1.63it/s]
                                                      
{'loss': 0.2095, 'grad_norm': 0.5668229460716248, 'learning_rate': 4.2991596398180035e-06, 'epoch': 1.77}

 59%|█████▉    | 7861/13317 [1:21:01<55:52,  1.63it/s]
 59%|█████▉    | 7862/13317 [1:21:02<55:54,  1.63it/s]
                                                      
{'loss': 0.1853, 'grad_norm': 0.5120302438735962, 'learning_rate': 4.297861967629004e-06, 'epoch': 1.77}

 59%|█████▉    | 7862/13317 [1:21:02<55:54,  1.63it/s]
 59%|█████▉    | 7863/13317 [1:21:02<55:52,  1.63it/s]
                                                      
{'loss': 0.2294, 'grad_norm': 0.519342303276062, 'learning_rate': 4.296564343684321e-06, 'epoch': 1.77}

 59%|█████▉    | 7863/13317 [1:21:03<55:52,  1.63it/s]
 59%|█████▉    | 7864/13317 [1:21:03<55:49,  1.63it/s]
                                                      
{'loss': 0.2021, 'grad_norm': 0.5022819638252258, 'learning_rate': 4.295266768073115e-06, 'epoch': 1.77}

 59%|█████▉    | 7864/13317 [1:21:03<55:49,  1.63it/s]
 59%|█████▉    | 7865/13317 [1:21:04<55:53,  1.63it/s]
                                                      
{'loss': 0.1971, 'grad_norm': 0.5609865188598633, 'learning_rate': 4.293969240884545e-06, 'epoch': 1.77}

 59%|█████▉    | 7865/13317 [1:21:04<55:53,  1.63it/s]
 59%|█████▉    | 7866/13317 [1:21:04<55:49,  1.63it/s]
                                                      
{'loss': 0.1685, 'grad_norm': 0.4487365782260895, 'learning_rate': 4.292671762207762e-06, 'epoch': 1.77}

 59%|█████▉    | 7866/13317 [1:21:04<55:49,  1.63it/s]
 59%|█████▉    | 7867/13317 [1:21:05<55:51,  1.63it/s]
                                                      
{'loss': 0.189, 'grad_norm': 0.5105443000793457, 'learning_rate': 4.2913743321319175e-06, 'epoch': 1.77}

 59%|█████▉    | 7867/13317 [1:21:05<55:51,  1.63it/s]
 59%|█████▉    | 7868/13317 [1:21:06<55:47,  1.63it/s]
                                                      
{'loss': 0.2554, 'grad_norm': 0.6346419453620911, 'learning_rate': 4.290076950746162e-06, 'epoch': 1.77}

 59%|█████▉    | 7868/13317 [1:21:06<55:47,  1.63it/s]
 59%|█████▉    | 7869/13317 [1:21:06<55:46,  1.63it/s]
                                                      
{'loss': 0.1675, 'grad_norm': 0.492570161819458, 'learning_rate': 4.288779618139637e-06, 'epoch': 1.77}

 59%|█████▉    | 7869/13317 [1:21:06<55:46,  1.63it/s]
 59%|█████▉    | 7870/13317 [1:21:07<55:49,  1.63it/s]
                                                      
{'loss': 0.1706, 'grad_norm': 0.46529990434646606, 'learning_rate': 4.287482334401481e-06, 'epoch': 1.77}

 59%|█████▉    | 7870/13317 [1:21:07<55:49,  1.63it/s]
 59%|█████▉    | 7871/13317 [1:21:07<55:47,  1.63it/s]
                                                      
{'loss': 0.1778, 'grad_norm': 0.4931381642818451, 'learning_rate': 4.286185099620833e-06, 'epoch': 1.77}

 59%|█████▉    | 7871/13317 [1:21:08<55:47,  1.63it/s]
 59%|█████▉    | 7872/13317 [1:21:08<55:50,  1.63it/s]
                                                      
{'loss': 0.1817, 'grad_norm': 0.5466057062149048, 'learning_rate': 4.284887913886827e-06, 'epoch': 1.77}

 59%|█████▉    | 7872/13317 [1:21:08<55:50,  1.63it/s]
 59%|█████▉    | 7873/13317 [1:21:09<55:46,  1.63it/s]
                                                      
{'loss': 0.1799, 'grad_norm': 0.5283612012863159, 'learning_rate': 4.2835907772885924e-06, 'epoch': 1.77}

 59%|█████▉    | 7873/13317 [1:21:09<55:46,  1.63it/s]
 59%|█████▉    | 7874/13317 [1:21:09<55:42,  1.63it/s]
                                                      
{'loss': 0.2673, 'grad_norm': 0.8283920288085938, 'learning_rate': 4.282293689915258e-06, 'epoch': 1.77}

 59%|█████▉    | 7874/13317 [1:21:09<55:42,  1.63it/s]
 59%|█████▉    | 7875/13317 [1:21:10<55:43,  1.63it/s]
                                                      
{'loss': 0.1551, 'grad_norm': 0.4503425359725952, 'learning_rate': 4.280996651855946e-06, 'epoch': 1.77}

 59%|█████▉    | 7875/13317 [1:21:10<55:43,  1.63it/s]
 59%|█████▉    | 7876/13317 [1:21:10<55:41,  1.63it/s]
                                                      
{'loss': 0.1751, 'grad_norm': 0.5464027523994446, 'learning_rate': 4.279699663199777e-06, 'epoch': 1.77}

 59%|█████▉    | 7876/13317 [1:21:11<55:41,  1.63it/s]
 59%|█████▉    | 7877/13317 [1:21:11<55:42,  1.63it/s]
                                                      
{'loss': 0.1837, 'grad_norm': 0.5028178095817566, 'learning_rate': 4.278402724035868e-06, 'epoch': 1.77}

 59%|█████▉    | 7877/13317 [1:21:11<55:42,  1.63it/s]
 59%|█████▉    | 7878/13317 [1:21:12<55:41,  1.63it/s]
                                                      
{'loss': 0.1938, 'grad_norm': 0.4915176033973694, 'learning_rate': 4.27710583445333e-06, 'epoch': 1.77}

 59%|█████▉    | 7878/13317 [1:21:12<55:41,  1.63it/s]
 59%|█████▉    | 7879/13317 [1:21:12<55:38,  1.63it/s]
                                                      
{'loss': 0.2422, 'grad_norm': 0.5818156003952026, 'learning_rate': 4.2758089945412795e-06, 'epoch': 1.77}

 59%|█████▉    | 7879/13317 [1:21:12<55:38,  1.63it/s]
 59%|█████▉    | 7880/13317 [1:21:13<55:40,  1.63it/s]
                                                      
{'loss': 0.1671, 'grad_norm': 0.4562281370162964, 'learning_rate': 4.274512204388815e-06, 'epoch': 1.78}

 59%|█████▉    | 7880/13317 [1:21:13<55:40,  1.63it/s]
 59%|█████▉    | 7881/13317 [1:21:14<55:39,  1.63it/s]
                                                      
{'loss': 0.2348, 'grad_norm': 0.5783881545066833, 'learning_rate': 4.273215464085046e-06, 'epoch': 1.78}

 59%|█████▉    | 7881/13317 [1:21:14<55:39,  1.63it/s]
 59%|█████▉    | 7882/13317 [1:21:14<55:40,  1.63it/s]
                                                      
{'loss': 0.2004, 'grad_norm': 0.5559356212615967, 'learning_rate': 4.27191877371907e-06, 'epoch': 1.78}

 59%|█████▉    | 7882/13317 [1:21:14<55:40,  1.63it/s]
 59%|█████▉    | 7883/13317 [1:21:15<55:39,  1.63it/s]
                                                      
{'loss': 0.1943, 'grad_norm': 0.5895378589630127, 'learning_rate': 4.2706221333799815e-06, 'epoch': 1.78}

 59%|█████▉    | 7883/13317 [1:21:15<55:39,  1.63it/s]
 59%|█████▉    | 7884/13317 [1:21:15<55:37,  1.63it/s]
                                                      
{'loss': 0.1797, 'grad_norm': 0.46742382645606995, 'learning_rate': 4.269325543156876e-06, 'epoch': 1.78}

 59%|█████▉    | 7884/13317 [1:21:16<55:37,  1.63it/s]
 59%|█████▉    | 7885/13317 [1:21:16<55:36,  1.63it/s]
                                                      
{'loss': 0.1691, 'grad_norm': 0.4869192838668823, 'learning_rate': 4.268029003138841e-06, 'epoch': 1.78}

 59%|█████▉    | 7885/13317 [1:21:16<55:36,  1.63it/s]
 59%|█████▉    | 7886/13317 [1:21:17<55:33,  1.63it/s]
                                                      
{'loss': 0.1863, 'grad_norm': 0.5136642456054688, 'learning_rate': 4.266732513414968e-06, 'epoch': 1.78}

 59%|█████▉    | 7886/13317 [1:21:17<55:33,  1.63it/s]
 59%|█████▉    | 7887/13317 [1:21:17<55:34,  1.63it/s]
                                                      
{'loss': 0.2135, 'grad_norm': 0.4916728436946869, 'learning_rate': 4.2654360740743314e-06, 'epoch': 1.78}

 59%|█████▉    | 7887/13317 [1:21:17<55:34,  1.63it/s]
 59%|█████▉    | 7888/13317 [1:21:18<55:33,  1.63it/s]
                                                      
{'loss': 0.1837, 'grad_norm': 0.4902981221675873, 'learning_rate': 4.264139685206015e-06, 'epoch': 1.78}

 59%|█████▉    | 7888/13317 [1:21:18<55:33,  1.63it/s]
 59%|█████▉    | 7889/13317 [1:21:18<55:31,  1.63it/s]
                                                      
{'loss': 0.2581, 'grad_norm': 0.6731853485107422, 'learning_rate': 4.2628433468990945e-06, 'epoch': 1.78}

 59%|█████▉    | 7889/13317 [1:21:19<55:31,  1.63it/s]
 59%|█████▉    | 7890/13317 [1:21:19<55:31,  1.63it/s]
                                                      
{'loss': 0.184, 'grad_norm': 0.47273287177085876, 'learning_rate': 4.26154705924264e-06, 'epoch': 1.78}

 59%|█████▉    | 7890/13317 [1:21:19<55:31,  1.63it/s]
 59%|█████▉    | 7891/13317 [1:21:20<55:31,  1.63it/s]
                                                      
{'loss': 0.2013, 'grad_norm': 0.5396926999092102, 'learning_rate': 4.2602508223257225e-06, 'epoch': 1.78}

 59%|█████▉    | 7891/13317 [1:21:20<55:31,  1.63it/s]
 59%|█████▉    | 7892/13317 [1:21:20<55:31,  1.63it/s]
                                                      
{'loss': 0.1643, 'grad_norm': 0.4575212895870209, 'learning_rate': 4.258954636237406e-06, 'epoch': 1.78}

 59%|█████▉    | 7892/13317 [1:21:20<55:31,  1.63it/s]
 59%|█████▉    | 7893/13317 [1:21:21<55:29,  1.63it/s]
                                                      
{'loss': 0.1939, 'grad_norm': 0.5780037045478821, 'learning_rate': 4.257658501066751e-06, 'epoch': 1.78}

 59%|█████▉    | 7893/13317 [1:21:21<55:29,  1.63it/s]
 59%|█████▉    | 7894/13317 [1:21:22<55:28,  1.63it/s]
                                                      
{'loss': 0.164, 'grad_norm': 0.46210619807243347, 'learning_rate': 4.256362416902817e-06, 'epoch': 1.78}

 59%|█████▉    | 7894/13317 [1:21:22<55:28,  1.63it/s]
 59%|█████▉    | 7895/13317 [1:21:22<55:28,  1.63it/s]
                                                      
{'loss': 0.163, 'grad_norm': 0.409822940826416, 'learning_rate': 4.255066383834659e-06, 'epoch': 1.78}

 59%|█████▉    | 7895/13317 [1:21:22<55:28,  1.63it/s]
 59%|█████▉    | 7896/13317 [1:21:23<55:26,  1.63it/s]
                                                      
{'loss': 0.1946, 'grad_norm': 0.5064714550971985, 'learning_rate': 4.25377040195133e-06, 'epoch': 1.78}

 59%|█████▉    | 7896/13317 [1:21:23<55:26,  1.63it/s]
 59%|█████▉    | 7897/13317 [1:21:23<55:31,  1.63it/s]
                                                      
{'loss': 0.2733, 'grad_norm': 0.6666949987411499, 'learning_rate': 4.252474471341874e-06, 'epoch': 1.78}

 59%|█████▉    | 7897/13317 [1:21:23<55:31,  1.63it/s]
 59%|█████▉    | 7898/13317 [1:21:24<55:29,  1.63it/s]
                                                      
{'loss': 0.2472, 'grad_norm': 0.6117306351661682, 'learning_rate': 4.251178592095337e-06, 'epoch': 1.78}

 59%|█████▉    | 7898/13317 [1:21:24<55:29,  1.63it/s]
 59%|█████▉    | 7899/13317 [1:21:25<55:27,  1.63it/s]
                                                      
{'loss': 0.2186, 'grad_norm': 0.6237326264381409, 'learning_rate': 4.24988276430076e-06, 'epoch': 1.78}

 59%|█████▉    | 7899/13317 [1:21:25<55:27,  1.63it/s]
 59%|█████▉    | 7900/13317 [1:21:25<55:25,  1.63it/s]
                                                      
{'loss': 0.173, 'grad_norm': 0.48863738775253296, 'learning_rate': 4.248586988047177e-06, 'epoch': 1.78}

 59%|█████▉    | 7900/13317 [1:21:25<55:25,  1.63it/s]
 59%|█████▉    | 7901/13317 [1:21:26<55:24,  1.63it/s]
                                                      
{'loss': 0.1259, 'grad_norm': 0.3683217763900757, 'learning_rate': 4.247291263423626e-06, 'epoch': 1.78}

 59%|█████▉    | 7901/13317 [1:21:26<55:24,  1.63it/s]
 59%|█████▉    | 7902/13317 [1:21:26<55:40,  1.62it/s]
                                                      
{'loss': 0.1955, 'grad_norm': 0.5978696942329407, 'learning_rate': 4.245995590519134e-06, 'epoch': 1.78}

 59%|█████▉    | 7902/13317 [1:21:27<55:40,  1.62it/s]
 59%|█████▉    | 7903/13317 [1:21:27<55:35,  1.62it/s]
                                                      
{'loss': 0.1381, 'grad_norm': 0.37191861867904663, 'learning_rate': 4.24469996942273e-06, 'epoch': 1.78}

 59%|█████▉    | 7903/13317 [1:21:27<55:35,  1.62it/s]
 59%|█████▉    | 7904/13317 [1:21:28<55:31,  1.62it/s]
                                                      
{'loss': 0.1844, 'grad_norm': 0.4880569577217102, 'learning_rate': 4.243404400223435e-06, 'epoch': 1.78}

 59%|█████▉    | 7904/13317 [1:21:28<55:31,  1.62it/s]
 59%|█████▉    | 7905/13317 [1:21:28<55:24,  1.63it/s]
                                                      
{'loss': 0.2297, 'grad_norm': 0.5031540393829346, 'learning_rate': 4.242108883010267e-06, 'epoch': 1.78}

 59%|█████▉    | 7905/13317 [1:21:28<55:24,  1.63it/s]
 59%|█████▉    | 7906/13317 [1:21:29<55:23,  1.63it/s]
                                                      
{'loss': 0.2152, 'grad_norm': 0.5268025994300842, 'learning_rate': 4.240813417872246e-06, 'epoch': 1.78}

 59%|█████▉    | 7906/13317 [1:21:29<55:23,  1.63it/s]
 59%|█████▉    | 7907/13317 [1:21:30<56:49,  1.59it/s]
                                                      
{'loss': 0.1845, 'grad_norm': 0.4941597878932953, 'learning_rate': 4.2395180048983796e-06, 'epoch': 1.78}

 59%|█████▉    | 7907/13317 [1:21:30<56:49,  1.59it/s]
 59%|█████▉    | 7908/13317 [1:21:30<56:22,  1.60it/s]
                                                      
{'loss': 0.1774, 'grad_norm': 0.4770977199077606, 'learning_rate': 4.238222644177679e-06, 'epoch': 1.78}

 59%|█████▉    | 7908/13317 [1:21:30<56:22,  1.60it/s]
 59%|█████▉    | 7909/13317 [1:21:31<56:02,  1.61it/s]
                                                      
{'loss': 0.2629, 'grad_norm': 0.5449137687683105, 'learning_rate': 4.236927335799148e-06, 'epoch': 1.78}

 59%|█████▉    | 7909/13317 [1:21:31<56:02,  1.61it/s]
 59%|█████▉    | 7910/13317 [1:21:31<55:52,  1.61it/s]
                                                      
{'loss': 0.2043, 'grad_norm': 0.5663343667984009, 'learning_rate': 4.23563207985179e-06, 'epoch': 1.78}

 59%|█████▉    | 7910/13317 [1:21:32<55:52,  1.61it/s]
 59%|█████▉    | 7911/13317 [1:21:32<55:42,  1.62it/s]
                                                      
{'loss': 0.2053, 'grad_norm': 0.5000289678573608, 'learning_rate': 4.2343368764246005e-06, 'epoch': 1.78}

 59%|█████▉    | 7911/13317 [1:21:32<55:42,  1.62it/s]
 59%|█████▉    | 7912/13317 [1:21:33<55:39,  1.62it/s]
                                                      
{'loss': 0.1785, 'grad_norm': 0.4412769675254822, 'learning_rate': 4.233041725606573e-06, 'epoch': 1.78}

 59%|█████▉    | 7912/13317 [1:21:33<55:39,  1.62it/s]
 59%|█████▉    | 7913/13317 [1:21:33<55:33,  1.62it/s]
                                                      
{'loss': 0.1689, 'grad_norm': 0.4259336292743683, 'learning_rate': 4.231746627486703e-06, 'epoch': 1.78}

 59%|█████▉    | 7913/13317 [1:21:33<55:33,  1.62it/s]
 59%|█████▉    | 7914/13317 [1:21:34<55:27,  1.62it/s]
                                                      
{'loss': 0.2034, 'grad_norm': 0.5165207386016846, 'learning_rate': 4.2304515821539714e-06, 'epoch': 1.78}

 59%|█████▉    | 7914/13317 [1:21:34<55:27,  1.62it/s]
 59%|█████▉    | 7915/13317 [1:21:34<55:42,  1.62it/s]
                                                      
{'loss': 0.169, 'grad_norm': 0.44016584753990173, 'learning_rate': 4.229156589697364e-06, 'epoch': 1.78}

 59%|█████▉    | 7915/13317 [1:21:35<55:42,  1.62it/s]
 59%|█████▉    | 7916/13317 [1:21:35<55:31,  1.62it/s]
                                                      
{'loss': 0.2086, 'grad_norm': 0.5058843493461609, 'learning_rate': 4.227861650205863e-06, 'epoch': 1.78}

 59%|█████▉    | 7916/13317 [1:21:35<55:31,  1.62it/s]
 59%|█████▉    | 7917/13317 [1:21:36<55:31,  1.62it/s]
                                                      
{'loss': 0.2434, 'grad_norm': 0.611801028251648, 'learning_rate': 4.22656676376844e-06, 'epoch': 1.78}

 59%|█████▉    | 7917/13317 [1:21:36<55:31,  1.62it/s]
 59%|█████▉    | 7918/13317 [1:21:36<55:25,  1.62it/s]
                                                      
{'loss': 0.1877, 'grad_norm': 0.6011483669281006, 'learning_rate': 4.22527193047407e-06, 'epoch': 1.78}

 59%|█████▉    | 7918/13317 [1:21:36<55:25,  1.62it/s]
 59%|█████▉    | 7919/13317 [1:21:37<55:19,  1.63it/s]
                                                      
{'loss': 0.2171, 'grad_norm': 0.5397281050682068, 'learning_rate': 4.223977150411722e-06, 'epoch': 1.78}

 59%|█████▉    | 7919/13317 [1:21:37<55:19,  1.63it/s]
 59%|█████▉    | 7920/13317 [1:21:38<55:20,  1.63it/s]
                                                      
{'loss': 0.2058, 'grad_norm': 0.5135673880577087, 'learning_rate': 4.222682423670362e-06, 'epoch': 1.78}

 59%|█████▉    | 7920/13317 [1:21:38<55:20,  1.63it/s]
 59%|█████▉    | 7921/13317 [1:21:38<55:17,  1.63it/s]
                                                      
{'loss': 0.2361, 'grad_norm': 0.6202486157417297, 'learning_rate': 4.221387750338947e-06, 'epoch': 1.78}

 59%|█████▉    | 7921/13317 [1:21:38<55:17,  1.63it/s]
 59%|█████▉    | 7922/13317 [1:21:39<55:18,  1.63it/s]
                                                      
{'loss': 0.1678, 'grad_norm': 0.49656710028648376, 'learning_rate': 4.220093130506438e-06, 'epoch': 1.78}

 59%|█████▉    | 7922/13317 [1:21:39<55:18,  1.63it/s]
 59%|█████▉    | 7923/13317 [1:21:39<55:15,  1.63it/s]
                                                      
{'loss': 0.2613, 'grad_norm': 0.638519823551178, 'learning_rate': 4.21879856426179e-06, 'epoch': 1.78}

 59%|█████▉    | 7923/13317 [1:21:40<55:15,  1.63it/s]
 60%|█████▉    | 7924/13317 [1:21:40<55:13,  1.63it/s]
                                                      
{'loss': 0.1993, 'grad_norm': 0.5392350554466248, 'learning_rate': 4.217504051693952e-06, 'epoch': 1.79}

 60%|█████▉    | 7924/13317 [1:21:40<55:13,  1.63it/s]
 60%|█████▉    | 7925/13317 [1:21:41<55:11,  1.63it/s]
                                                      
{'loss': 0.1782, 'grad_norm': 0.4797538220882416, 'learning_rate': 4.21620959289187e-06, 'epoch': 1.79}

 60%|█████▉    | 7925/13317 [1:21:41<55:11,  1.63it/s]
 60%|█████▉    | 7926/13317 [1:21:41<55:12,  1.63it/s]
                                                      
{'loss': 0.1805, 'grad_norm': 0.5499023199081421, 'learning_rate': 4.21491518794449e-06, 'epoch': 1.79}

 60%|█████▉    | 7926/13317 [1:21:41<55:12,  1.63it/s]
 60%|█████▉    | 7927/13317 [1:21:42<55:12,  1.63it/s]
                                                      
{'loss': 0.2295, 'grad_norm': 0.5503281950950623, 'learning_rate': 4.213620836940746e-06, 'epoch': 1.79}

 60%|█████▉    | 7927/13317 [1:21:42<55:12,  1.63it/s]
 60%|█████▉    | 7928/13317 [1:21:42<55:11,  1.63it/s]
                                                      
{'loss': 0.169, 'grad_norm': 0.4362323582172394, 'learning_rate': 4.212326539969578e-06, 'epoch': 1.79}

 60%|█████▉    | 7928/13317 [1:21:43<55:11,  1.63it/s]
 60%|█████▉    | 7929/13317 [1:21:43<55:08,  1.63it/s]
                                                      
{'loss': 0.1983, 'grad_norm': 0.6299642324447632, 'learning_rate': 4.211032297119917e-06, 'epoch': 1.79}

 60%|█████▉    | 7929/13317 [1:21:43<55:08,  1.63it/s]
 60%|█████▉    | 7930/13317 [1:21:44<55:07,  1.63it/s]
                                                      
{'loss': 0.2112, 'grad_norm': 0.6192761659622192, 'learning_rate': 4.209738108480692e-06, 'epoch': 1.79}

 60%|█████▉    | 7930/13317 [1:21:44<55:07,  1.63it/s]
 60%|█████▉    | 7931/13317 [1:21:44<55:05,  1.63it/s]
                                                      
{'loss': 0.187, 'grad_norm': 0.5071905255317688, 'learning_rate': 4.208443974140825e-06, 'epoch': 1.79}

 60%|█████▉    | 7931/13317 [1:21:44<55:05,  1.63it/s]
 60%|█████▉    | 7932/13317 [1:21:45<55:04,  1.63it/s]
                                                      
{'loss': 0.24, 'grad_norm': 0.6137428283691406, 'learning_rate': 4.207149894189237e-06, 'epoch': 1.79}

 60%|█████▉    | 7932/13317 [1:21:45<55:04,  1.63it/s]
 60%|█████▉    | 7933/13317 [1:21:46<55:03,  1.63it/s]
                                                      
{'loss': 0.1991, 'grad_norm': 0.5139570236206055, 'learning_rate': 4.2058558687148495e-06, 'epoch': 1.79}

 60%|█████▉    | 7933/13317 [1:21:46<55:03,  1.63it/s]
 60%|█████▉    | 7934/13317 [1:21:46<55:01,  1.63it/s]
                                                      
{'loss': 0.1703, 'grad_norm': 0.5074630975723267, 'learning_rate': 4.204561897806569e-06, 'epoch': 1.79}

 60%|█████▉    | 7934/13317 [1:21:46<55:01,  1.63it/s]
 60%|█████▉    | 7935/13317 [1:21:47<55:05,  1.63it/s]
                                                      
{'loss': 0.21, 'grad_norm': 0.5928581357002258, 'learning_rate': 4.20326798155331e-06, 'epoch': 1.79}

 60%|█████▉    | 7935/13317 [1:21:47<55:05,  1.63it/s]
 60%|█████▉    | 7936/13317 [1:21:47<55:02,  1.63it/s]
                                                      
{'loss': 0.1806, 'grad_norm': 0.5271252393722534, 'learning_rate': 4.201974120043978e-06, 'epoch': 1.79}

 60%|█████▉    | 7936/13317 [1:21:48<55:02,  1.63it/s]
 60%|█████▉    | 7937/13317 [1:21:48<55:02,  1.63it/s]
                                                      
{'loss': 0.1738, 'grad_norm': 0.5229293704032898, 'learning_rate': 4.200680313367471e-06, 'epoch': 1.79}

 60%|█████▉    | 7937/13317 [1:21:48<55:02,  1.63it/s]
 60%|█████▉    | 7938/13317 [1:21:49<55:01,  1.63it/s]
                                                      
{'loss': 0.1742, 'grad_norm': 0.42781898379325867, 'learning_rate': 4.199386561612691e-06, 'epoch': 1.79}

 60%|█████▉    | 7938/13317 [1:21:49<55:01,  1.63it/s]
 60%|█████▉    | 7939/13317 [1:21:49<54:57,  1.63it/s]
                                                      
{'loss': 0.2832, 'grad_norm': 0.5839696526527405, 'learning_rate': 4.19809286486853e-06, 'epoch': 1.79}

 60%|█████▉    | 7939/13317 [1:21:49<54:57,  1.63it/s]
 60%|█████▉    | 7940/13317 [1:21:50<54:57,  1.63it/s]
                                                      
{'loss': 0.2024, 'grad_norm': 0.5498126149177551, 'learning_rate': 4.196799223223882e-06, 'epoch': 1.79}

 60%|█████▉    | 7940/13317 [1:21:50<54:57,  1.63it/s]
 60%|█████▉    | 7941/13317 [1:21:50<54:59,  1.63it/s]
                                                      
{'loss': 0.2273, 'grad_norm': 0.4986919164657593, 'learning_rate': 4.19550563676763e-06, 'epoch': 1.79}

 60%|█████▉    | 7941/13317 [1:21:51<54:59,  1.63it/s]
 60%|█████▉    | 7942/13317 [1:21:51<54:56,  1.63it/s]
                                                      
{'loss': 0.1946, 'grad_norm': 0.4921632409095764, 'learning_rate': 4.19421210558866e-06, 'epoch': 1.79}

 60%|█████▉    | 7942/13317 [1:21:51<54:56,  1.63it/s]
 60%|█████▉    | 7943/13317 [1:21:52<54:56,  1.63it/s]
                                                      
{'loss': 0.2506, 'grad_norm': 0.6588376760482788, 'learning_rate': 4.19291862977585e-06, 'epoch': 1.79}

 60%|█████▉    | 7943/13317 [1:21:52<54:56,  1.63it/s]
 60%|█████▉    | 7944/13317 [1:21:52<54:54,  1.63it/s]
                                                      
{'loss': 0.1909, 'grad_norm': 0.5185906887054443, 'learning_rate': 4.191625209418075e-06, 'epoch': 1.79}

 60%|█████▉    | 7944/13317 [1:21:52<54:54,  1.63it/s]
 60%|█████▉    | 7945/13317 [1:21:53<54:54,  1.63it/s]
                                                      
{'loss': 0.2054, 'grad_norm': 0.5113170146942139, 'learning_rate': 4.190331844604208e-06, 'epoch': 1.79}

 60%|█████▉    | 7945/13317 [1:21:53<54:54,  1.63it/s]
 60%|█████▉    | 7946/13317 [1:21:54<54:52,  1.63it/s]
                                                      
{'loss': 0.1699, 'grad_norm': 0.47051358222961426, 'learning_rate': 4.189038535423115e-06, 'epoch': 1.79}

 60%|█████▉    | 7946/13317 [1:21:54<54:52,  1.63it/s]
 60%|█████▉    | 7947/13317 [1:21:54<54:51,  1.63it/s]
                                                      
{'loss': 0.216, 'grad_norm': 0.48656633496284485, 'learning_rate': 4.1877452819636634e-06, 'epoch': 1.79}

 60%|█████▉    | 7947/13317 [1:21:54<54:51,  1.63it/s]
 60%|█████▉    | 7948/13317 [1:21:55<54:51,  1.63it/s]
                                                      
{'loss': 0.2029, 'grad_norm': 0.5891323685646057, 'learning_rate': 4.186452084314709e-06, 'epoch': 1.79}

 60%|█████▉    | 7948/13317 [1:21:55<54:51,  1.63it/s]
 60%|█████▉    | 7949/13317 [1:21:55<54:50,  1.63it/s]
                                                      
{'loss': 0.1796, 'grad_norm': 0.5463069677352905, 'learning_rate': 4.185158942565111e-06, 'epoch': 1.79}

 60%|█████▉    | 7949/13317 [1:21:55<54:50,  1.63it/s]
 60%|█████▉    | 7950/13317 [1:21:56<54:49,  1.63it/s]
                                                      
{'loss': 0.2209, 'grad_norm': 0.65427565574646, 'learning_rate': 4.183865856803723e-06, 'epoch': 1.79}

 60%|█████▉    | 7950/13317 [1:21:56<54:49,  1.63it/s]
 60%|█████▉    | 7951/13317 [1:21:57<54:50,  1.63it/s]
                                                      
{'loss': 0.1857, 'grad_norm': 0.5051917433738708, 'learning_rate': 4.18257282711939e-06, 'epoch': 1.79}

 60%|█████▉    | 7951/13317 [1:21:57<54:50,  1.63it/s]
 60%|█████▉    | 7952/13317 [1:21:57<54:51,  1.63it/s]
                                                      
{'loss': 0.2014, 'grad_norm': 0.5566297769546509, 'learning_rate': 4.181279853600959e-06, 'epoch': 1.79}

 60%|█████▉    | 7952/13317 [1:21:57<54:51,  1.63it/s]
 60%|█████▉    | 7953/13317 [1:21:58<54:52,  1.63it/s]
                                                      
{'loss': 0.1874, 'grad_norm': 0.4979874789714813, 'learning_rate': 4.179986936337272e-06, 'epoch': 1.79}

 60%|█████▉    | 7953/13317 [1:21:58<54:52,  1.63it/s]
 60%|█████▉    | 7954/13317 [1:21:58<54:52,  1.63it/s]
                                                      
{'loss': 0.2139, 'grad_norm': 0.5587602257728577, 'learning_rate': 4.178694075417164e-06, 'epoch': 1.79}

 60%|█████▉    | 7954/13317 [1:21:59<54:52,  1.63it/s]
 60%|█████▉    | 7955/13317 [1:21:59<54:50,  1.63it/s]
                                                      
{'loss': 0.1677, 'grad_norm': 0.4586150050163269, 'learning_rate': 4.1774012709294685e-06, 'epoch': 1.79}

 60%|█████▉    | 7955/13317 [1:21:59<54:50,  1.63it/s]
 60%|█████▉    | 7956/13317 [1:22:00<54:50,  1.63it/s]
                                                      
{'loss': 0.229, 'grad_norm': 0.53897625207901, 'learning_rate': 4.176108522963016e-06, 'epoch': 1.79}

 60%|█████▉    | 7956/13317 [1:22:00<54:50,  1.63it/s]
 60%|█████▉    | 7957/13317 [1:22:00<54:54,  1.63it/s]
                                                      
{'loss': 0.2055, 'grad_norm': 0.5035035014152527, 'learning_rate': 4.174815831606634e-06, 'epoch': 1.79}

 60%|█████▉    | 7957/13317 [1:22:00<54:54,  1.63it/s]
 60%|█████▉    | 7958/13317 [1:22:01<54:50,  1.63it/s]
                                                      
{'loss': 0.1852, 'grad_norm': 0.5525067448616028, 'learning_rate': 4.173523196949138e-06, 'epoch': 1.79}

 60%|█████▉    | 7958/13317 [1:22:01<54:50,  1.63it/s]
 60%|█████▉    | 7959/13317 [1:22:01<54:52,  1.63it/s]
                                                      
{'loss': 0.2244, 'grad_norm': 0.5944509506225586, 'learning_rate': 4.17223061907935e-06, 'epoch': 1.79}

 60%|█████▉    | 7959/13317 [1:22:02<54:52,  1.63it/s]
 60%|█████▉    | 7960/13317 [1:22:02<54:49,  1.63it/s]
                                                      
{'loss': 0.2867, 'grad_norm': 0.7348896265029907, 'learning_rate': 4.170938098086086e-06, 'epoch': 1.79}

 60%|█████▉    | 7960/13317 [1:22:02<54:49,  1.63it/s]
 60%|█████▉    | 7961/13317 [1:22:03<54:47,  1.63it/s]
                                                      
{'loss': 0.1772, 'grad_norm': 0.4855683445930481, 'learning_rate': 4.169645634058149e-06, 'epoch': 1.79}

 60%|█████▉    | 7961/13317 [1:22:03<54:47,  1.63it/s]
 60%|█████▉    | 7962/13317 [1:22:03<54:48,  1.63it/s]
                                                      
{'loss': 0.233, 'grad_norm': 0.5702263116836548, 'learning_rate': 4.1683532270843505e-06, 'epoch': 1.79}

 60%|█████▉    | 7962/13317 [1:22:03<54:48,  1.63it/s]
 60%|█████▉    | 7963/13317 [1:22:04<54:47,  1.63it/s]
                                                      
{'loss': 0.1818, 'grad_norm': 0.4898233115673065, 'learning_rate': 4.1670608772534904e-06, 'epoch': 1.79}

 60%|█████▉    | 7963/13317 [1:22:04<54:47,  1.63it/s]
 60%|█████▉    | 7964/13317 [1:22:05<54:45,  1.63it/s]
                                                      
{'loss': 0.2046, 'grad_norm': 0.5039781332015991, 'learning_rate': 4.165768584654369e-06, 'epoch': 1.79}

 60%|█████▉    | 7964/13317 [1:22:05<54:45,  1.63it/s]
 60%|█████▉    | 7965/13317 [1:22:05<54:43,  1.63it/s]
                                                      
{'loss': 0.1386, 'grad_norm': 0.3977975845336914, 'learning_rate': 4.164476349375777e-06, 'epoch': 1.79}

 60%|█████▉    | 7965/13317 [1:22:05<54:43,  1.63it/s]
 60%|█████▉    | 7966/13317 [1:22:06<54:43,  1.63it/s]
                                                      
{'loss': 0.1612, 'grad_norm': 0.4533793330192566, 'learning_rate': 4.163184171506506e-06, 'epoch': 1.79}

 60%|█████▉    | 7966/13317 [1:22:06<54:43,  1.63it/s]
 60%|█████▉    | 7967/13317 [1:22:06<54:45,  1.63it/s]
                                                      
{'loss': 0.1679, 'grad_norm': 0.4567244350910187, 'learning_rate': 4.161892051135346e-06, 'epoch': 1.79}

 60%|█████▉    | 7967/13317 [1:22:07<54:45,  1.63it/s]
 60%|█████▉    | 7968/13317 [1:22:07<54:45,  1.63it/s]
                                                      
{'loss': 0.1912, 'grad_norm': 0.4925084114074707, 'learning_rate': 4.160599988351073e-06, 'epoch': 1.79}

 60%|█████▉    | 7968/13317 [1:22:07<54:45,  1.63it/s]
 60%|█████▉    | 7969/13317 [1:22:08<54:43,  1.63it/s]
                                                      
{'loss': 0.1448, 'grad_norm': 0.4460769295692444, 'learning_rate': 4.159307983242469e-06, 'epoch': 1.8}

 60%|█████▉    | 7969/13317 [1:22:08<54:43,  1.63it/s]
 60%|█████▉    | 7970/13317 [1:22:08<54:40,  1.63it/s]
                                                      
{'loss': 0.2019, 'grad_norm': 0.46227625012397766, 'learning_rate': 4.1580160358983094e-06, 'epoch': 1.8}

 60%|█████▉    | 7970/13317 [1:22:08<54:40,  1.63it/s]
 60%|█████▉    | 7971/13317 [1:22:09<54:40,  1.63it/s]
                                                      
{'loss': 0.1889, 'grad_norm': 0.5325335264205933, 'learning_rate': 4.1567241464073615e-06, 'epoch': 1.8}

 60%|█████▉    | 7971/13317 [1:22:09<54:40,  1.63it/s]
 60%|█████▉    | 7972/13317 [1:22:09<54:43,  1.63it/s]
                                                      
{'loss': 0.2004, 'grad_norm': 0.4936160445213318, 'learning_rate': 4.155432314858394e-06, 'epoch': 1.8}

 60%|█████▉    | 7972/13317 [1:22:10<54:43,  1.63it/s]
 60%|█████▉    | 7973/13317 [1:22:10<54:43,  1.63it/s]
                                                      
{'loss': 0.1841, 'grad_norm': 0.5059345364570618, 'learning_rate': 4.154140541340168e-06, 'epoch': 1.8}

 60%|█████▉    | 7973/13317 [1:22:10<54:43,  1.63it/s]
 60%|█████▉    | 7974/13317 [1:22:11<54:40,  1.63it/s]
                                                      
{'loss': 0.193, 'grad_norm': 0.5129794478416443, 'learning_rate': 4.152848825941445e-06, 'epoch': 1.8}

 60%|█████▉    | 7974/13317 [1:22:11<54:40,  1.63it/s]
 60%|█████▉    | 7975/13317 [1:22:11<54:38,  1.63it/s]
                                                      
{'loss': 0.1839, 'grad_norm': 0.54285728931427, 'learning_rate': 4.151557168750977e-06, 'epoch': 1.8}

 60%|█████▉    | 7975/13317 [1:22:11<54:38,  1.63it/s]
 60%|█████▉    | 7976/13317 [1:22:12<54:37,  1.63it/s]
                                                      
{'loss': 0.2195, 'grad_norm': 0.6675556302070618, 'learning_rate': 4.150265569857513e-06, 'epoch': 1.8}

 60%|█████▉    | 7976/13317 [1:22:12<54:37,  1.63it/s]
 60%|█████▉    | 7977/13317 [1:22:13<54:54,  1.62it/s]
                                                      
{'loss': 0.1724, 'grad_norm': 0.5305349826812744, 'learning_rate': 4.148974029349804e-06, 'epoch': 1.8}

 60%|█████▉    | 7977/13317 [1:22:13<54:54,  1.62it/s]
 60%|█████▉    | 7978/13317 [1:22:13<54:50,  1.62it/s]
                                                      
{'loss': 0.2196, 'grad_norm': 0.6846964359283447, 'learning_rate': 4.147682547316589e-06, 'epoch': 1.8}

 60%|█████▉    | 7978/13317 [1:22:13<54:50,  1.62it/s]
 60%|█████▉    | 7979/13317 [1:22:14<54:44,  1.63it/s]
                                                      
{'loss': 0.163, 'grad_norm': 0.4432225823402405, 'learning_rate': 4.146391123846607e-06, 'epoch': 1.8}

 60%|█████▉    | 7979/13317 [1:22:14<54:44,  1.63it/s]
 60%|█████▉    | 7980/13317 [1:22:14<54:40,  1.63it/s]
                                                      
{'loss': 0.2517, 'grad_norm': 0.5309733748435974, 'learning_rate': 4.145099759028595e-06, 'epoch': 1.8}

 60%|█████▉    | 7980/13317 [1:22:15<54:40,  1.63it/s]
 60%|█████▉    | 7981/13317 [1:22:15<54:39,  1.63it/s]
                                                      
{'loss': 0.1586, 'grad_norm': 0.44415074586868286, 'learning_rate': 4.143808452951278e-06, 'epoch': 1.8}

 60%|█████▉    | 7981/13317 [1:22:15<54:39,  1.63it/s]
 60%|█████▉    | 7982/13317 [1:22:16<54:54,  1.62it/s]
                                                      
{'loss': 0.1813, 'grad_norm': 0.4788772165775299, 'learning_rate': 4.142517205703387e-06, 'epoch': 1.8}

 60%|█████▉    | 7982/13317 [1:22:16<54:54,  1.62it/s]
 60%|█████▉    | 7983/13317 [1:22:16<54:46,  1.62it/s]
                                                      
{'loss': 0.2151, 'grad_norm': 0.5184949636459351, 'learning_rate': 4.141226017373644e-06, 'epoch': 1.8}

 60%|█████▉    | 7983/13317 [1:22:16<54:46,  1.62it/s]
 60%|█████▉    | 7984/13317 [1:22:17<54:42,  1.62it/s]
                                                      
{'loss': 0.1815, 'grad_norm': 0.5418878793716431, 'learning_rate': 4.139934888050766e-06, 'epoch': 1.8}

 60%|█████▉    | 7984/13317 [1:22:17<54:42,  1.62it/s]
 60%|█████▉    | 7985/13317 [1:22:17<54:38,  1.63it/s]
                                                      
{'loss': 0.2318, 'grad_norm': 0.5317277312278748, 'learning_rate': 4.138643817823468e-06, 'epoch': 1.8}

 60%|█████▉    | 7985/13317 [1:22:18<54:38,  1.63it/s]
 60%|█████▉    | 7986/13317 [1:22:18<54:35,  1.63it/s]
                                                      
{'loss': 0.1498, 'grad_norm': 0.43536481261253357, 'learning_rate': 4.137352806780459e-06, 'epoch': 1.8}

 60%|█████▉    | 7986/13317 [1:22:18<54:35,  1.63it/s]
 60%|█████▉    | 7987/13317 [1:22:19<54:45,  1.62it/s]
                                                      
{'loss': 0.2507, 'grad_norm': 0.5868749618530273, 'learning_rate': 4.1360618550104485e-06, 'epoch': 1.8}

 60%|█████▉    | 7987/13317 [1:22:19<54:45,  1.62it/s]
 60%|█████▉    | 7988/13317 [1:22:19<54:40,  1.62it/s]
                                                      
{'loss': 0.1554, 'grad_norm': 0.4664444923400879, 'learning_rate': 4.134770962602133e-06, 'epoch': 1.8}

 60%|█████▉    | 7988/13317 [1:22:19<54:40,  1.62it/s]
 60%|█████▉    | 7989/13317 [1:22:20<54:36,  1.63it/s]
                                                      
{'loss': 0.3076, 'grad_norm': 0.6240715980529785, 'learning_rate': 4.133480129644215e-06, 'epoch': 1.8}

 60%|█████▉    | 7989/13317 [1:22:20<54:36,  1.63it/s]
 60%|█████▉    | 7990/13317 [1:22:21<54:33,  1.63it/s]
                                                      
{'loss': 0.1572, 'grad_norm': 0.49501705169677734, 'learning_rate': 4.132189356225387e-06, 'epoch': 1.8}

 60%|█████▉    | 7990/13317 [1:22:21<54:33,  1.63it/s]
 60%|██████    | 7991/13317 [1:22:21<54:30,  1.63it/s]
                                                      
{'loss': 0.1748, 'grad_norm': 0.4405801594257355, 'learning_rate': 4.130898642434339e-06, 'epoch': 1.8}

 60%|██████    | 7991/13317 [1:22:21<54:30,  1.63it/s]
 60%|██████    | 7992/13317 [1:22:22<54:31,  1.63it/s]
                                                      
{'loss': 0.1738, 'grad_norm': 0.4790673851966858, 'learning_rate': 4.129607988359755e-06, 'epoch': 1.8}

 60%|██████    | 7992/13317 [1:22:22<54:31,  1.63it/s]
 60%|██████    | 7993/13317 [1:22:22<54:29,  1.63it/s]
                                                      
{'loss': 0.2204, 'grad_norm': 0.5768387913703918, 'learning_rate': 4.128317394090318e-06, 'epoch': 1.8}

 60%|██████    | 7993/13317 [1:22:23<54:29,  1.63it/s]
 60%|██████    | 7994/13317 [1:22:23<54:28,  1.63it/s]
                                                      
{'loss': 0.1751, 'grad_norm': 0.5182169675827026, 'learning_rate': 4.127026859714709e-06, 'epoch': 1.8}

 60%|██████    | 7994/13317 [1:22:23<54:28,  1.63it/s]
 60%|██████    | 7995/13317 [1:22:24<54:25,  1.63it/s]
                                                      
{'loss': 0.2652, 'grad_norm': 0.6841035485267639, 'learning_rate': 4.125736385321595e-06, 'epoch': 1.8}

 60%|██████    | 7995/13317 [1:22:24<54:25,  1.63it/s]
 60%|██████    | 7996/13317 [1:22:24<54:31,  1.63it/s]
                                                      
{'loss': 0.2014, 'grad_norm': 0.5080269575119019, 'learning_rate': 4.124445970999648e-06, 'epoch': 1.8}

 60%|██████    | 7996/13317 [1:22:24<54:31,  1.63it/s]
 60%|██████    | 7997/13317 [1:22:25<54:30,  1.63it/s]
                                                      
{'loss': 0.1657, 'grad_norm': 0.47189804911613464, 'learning_rate': 4.123155616837536e-06, 'epoch': 1.8}

 60%|██████    | 7997/13317 [1:22:25<54:30,  1.63it/s]
 60%|██████    | 7998/13317 [1:22:25<54:28,  1.63it/s]
                                                      
{'loss': 0.2152, 'grad_norm': 0.5640479326248169, 'learning_rate': 4.121865322923914e-06, 'epoch': 1.8}

 60%|██████    | 7998/13317 [1:22:26<54:28,  1.63it/s]
 60%|██████    | 7999/13317 [1:22:26<54:27,  1.63it/s]
                                                      
{'loss': 0.1698, 'grad_norm': 0.44674891233444214, 'learning_rate': 4.120575089347444e-06, 'epoch': 1.8}

 60%|██████    | 7999/13317 [1:22:26<54:27,  1.63it/s]
 60%|██████    | 8000/13317 [1:22:27<54:28,  1.63it/s]
                                                      
{'loss': 0.1792, 'grad_norm': 0.5092365145683289, 'learning_rate': 4.119284916196775e-06, 'epoch': 1.8}

 60%|██████    | 8000/13317 [1:22:27<54:28,  1.63it/s]

  0%|          | 0/13 [00:00<?, ?it/s]

 15%|█▌        | 2/13 [00:00<00:00, 13.36it/s]

 31%|███       | 4/13 [00:00<00:01,  8.40it/s]

 38%|███▊      | 5/13 [00:00<00:01,  7.80it/s]

 46%|████▌     | 6/13 [00:00<00:00,  7.42it/s]

 54%|█████▍    | 7/13 [00:00<00:00,  7.18it/s]

 62%|██████▏   | 8/13 [00:01<00:00,  7.02it/s]

 69%|██████▉   | 9/13 [00:01<00:00,  6.91it/s]

 77%|███████▋  | 10/13 [00:01<00:00,  6.83it/s]

 85%|████████▍ | 11/13 [00:01<00:00,  6.78it/s]

 92%|█████████▏| 12/13 [00:01<00:00,  6.75it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]
                                                      

                                               
{'eval_loss': 0.5382644534111023, 'eval_runtime': 1.9525, 'eval_samples_per_second': 102.434, 'eval_steps_per_second': 6.658, 'epoch': 1.8}

 60%|██████    | 8000/13317 [1:22:29<54:28,  1.63it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]

                                               
 60%|██████    | 8001/13317 [1:22:29<1:46:28,  1.20s/it]
                                                        
{'loss': 0.1945, 'grad_norm': 0.4963882863521576, 'learning_rate': 4.11799480356056e-06, 'epoch': 1.8}

 60%|██████    | 8001/13317 [1:22:29<1:46:28,  1.20s/it]
 60%|██████    | 8002/13317 [1:22:30<1:30:49,  1.03s/it]
                                                        
{'loss': 0.1537, 'grad_norm': 0.46161434054374695, 'learning_rate': 4.116704751527439e-06, 'epoch': 1.8}

 60%|██████    | 8002/13317 [1:22:30<1:30:49,  1.03s/it]
 60%|██████    | 8003/13317 [1:22:30<1:19:52,  1.11it/s]
                                                        
{'loss': 0.2613, 'grad_norm': 0.6590343713760376, 'learning_rate': 4.115414760186055e-06, 'epoch': 1.8}

 60%|██████    | 8003/13317 [1:22:31<1:19:52,  1.11it/s]
 60%|██████    | 8004/13317 [1:22:31<1:12:11,  1.23it/s]
                                                        
{'loss': 0.2107, 'grad_norm': 0.5213243961334229, 'learning_rate': 4.114124829625043e-06, 'epoch': 1.8}

 60%|██████    | 8004/13317 [1:22:31<1:12:11,  1.23it/s]
 60%|██████    | 8005/13317 [1:22:32<1:06:50,  1.32it/s]
                                                        
{'loss': 0.1896, 'grad_norm': 0.5548965334892273, 'learning_rate': 4.112834959933035e-06, 'epoch': 1.8}

 60%|██████    | 8005/13317 [1:22:32<1:06:50,  1.32it/s]
 60%|██████    | 8006/13317 [1:22:32<1:03:04,  1.40it/s]
                                                        
{'loss': 0.2312, 'grad_norm': 0.5602377653121948, 'learning_rate': 4.111545151198657e-06, 'epoch': 1.8}

 60%|██████    | 8006/13317 [1:22:32<1:03:04,  1.40it/s]
 60%|██████    | 8007/13317 [1:22:33<1:00:28,  1.46it/s]
                                                        
{'loss': 0.2115, 'grad_norm': 0.565655529499054, 'learning_rate': 4.110255403510535e-06, 'epoch': 1.8}

 60%|██████    | 8007/13317 [1:22:33<1:00:28,  1.46it/s]
 60%|██████    | 8008/13317 [1:22:34<58:37,  1.51it/s]  
                                                      
{'loss': 0.2957, 'grad_norm': 0.5951086282730103, 'learning_rate': 4.108965716957289e-06, 'epoch': 1.8}

 60%|██████    | 8008/13317 [1:22:34<58:37,  1.51it/s]
 60%|██████    | 8009/13317 [1:22:34<57:19,  1.54it/s]
                                                      
{'loss': 0.22, 'grad_norm': 0.686813473701477, 'learning_rate': 4.107676091627532e-06, 'epoch': 1.8}

 60%|██████    | 8009/13317 [1:22:34<57:19,  1.54it/s]
 60%|██████    | 8010/13317 [1:22:35<56:23,  1.57it/s]
                                                      
{'loss': 0.1889, 'grad_norm': 0.47467270493507385, 'learning_rate': 4.106386527609876e-06, 'epoch': 1.8}

 60%|██████    | 8010/13317 [1:22:35<56:23,  1.57it/s]
 60%|██████    | 8011/13317 [1:22:35<55:44,  1.59it/s]
                                                      
{'loss': 0.2108, 'grad_norm': 0.5399473309516907, 'learning_rate': 4.105097024992928e-06, 'epoch': 1.8}

 60%|██████    | 8011/13317 [1:22:36<55:44,  1.59it/s]
 60%|██████    | 8012/13317 [1:22:36<55:19,  1.60it/s]
                                                      
{'loss': 0.1395, 'grad_norm': 0.4440370202064514, 'learning_rate': 4.103807583865288e-06, 'epoch': 1.8}

 60%|██████    | 8012/13317 [1:22:36<55:19,  1.60it/s]
 60%|██████    | 8013/13317 [1:22:37<54:58,  1.61it/s]
                                                      
{'loss': 0.1663, 'grad_norm': 0.4443536698818207, 'learning_rate': 4.102518204315555e-06, 'epoch': 1.81}

 60%|██████    | 8013/13317 [1:22:37<54:58,  1.61it/s]
 60%|██████    | 8014/13317 [1:22:37<54:44,  1.61it/s]
                                                      
{'loss': 0.2354, 'grad_norm': 0.5513588190078735, 'learning_rate': 4.101228886432327e-06, 'epoch': 1.81}

 60%|██████    | 8014/13317 [1:22:37<54:44,  1.61it/s]
 60%|██████    | 8015/13317 [1:22:38<54:34,  1.62it/s]
                                                      
{'loss': 0.1821, 'grad_norm': 0.47890082001686096, 'learning_rate': 4.0999396303041885e-06, 'epoch': 1.81}

 60%|██████    | 8015/13317 [1:22:38<54:34,  1.62it/s]
 60%|██████    | 8016/13317 [1:22:38<54:26,  1.62it/s]
                                                      
{'loss': 0.1654, 'grad_norm': 0.47036704421043396, 'learning_rate': 4.098650436019728e-06, 'epoch': 1.81}

 60%|██████    | 8016/13317 [1:22:39<54:26,  1.62it/s]
 60%|██████    | 8017/13317 [1:22:39<54:25,  1.62it/s]
                                                      
{'loss': 0.1734, 'grad_norm': 0.6047675013542175, 'learning_rate': 4.097361303667527e-06, 'epoch': 1.81}

 60%|██████    | 8017/13317 [1:22:39<54:25,  1.62it/s]
 60%|██████    | 8018/13317 [1:22:40<54:21,  1.62it/s]
                                                      
{'loss': 0.149, 'grad_norm': 0.40022245049476624, 'learning_rate': 4.096072233336162e-06, 'epoch': 1.81}

 60%|██████    | 8018/13317 [1:22:40<54:21,  1.62it/s]
 60%|██████    | 8019/13317 [1:22:40<54:18,  1.63it/s]
                                                      
{'loss': 0.1968, 'grad_norm': 0.523635745048523, 'learning_rate': 4.094783225114204e-06, 'epoch': 1.81}

 60%|██████    | 8019/13317 [1:22:40<54:18,  1.63it/s]
 60%|██████    | 8020/13317 [1:22:41<54:14,  1.63it/s]
                                                      
{'loss': 0.1886, 'grad_norm': 0.4819149672985077, 'learning_rate': 4.093494279090223e-06, 'epoch': 1.81}

 60%|██████    | 8020/13317 [1:22:41<54:14,  1.63it/s]
 60%|██████    | 8021/13317 [1:22:42<54:13,  1.63it/s]
                                                      
{'loss': 0.2457, 'grad_norm': 0.6210206151008606, 'learning_rate': 4.092205395352786e-06, 'epoch': 1.81}

 60%|██████    | 8021/13317 [1:22:42<54:13,  1.63it/s]
 60%|██████    | 8022/13317 [1:22:42<54:13,  1.63it/s]
                                                      
{'loss': 0.1961, 'grad_norm': 0.5748053193092346, 'learning_rate': 4.0909165739904474e-06, 'epoch': 1.81}

 60%|██████    | 8022/13317 [1:22:42<54:13,  1.63it/s]
 60%|██████    | 8023/13317 [1:22:43<54:11,  1.63it/s]
                                                      
{'loss': 0.2274, 'grad_norm': 0.6035304069519043, 'learning_rate': 4.089627815091766e-06, 'epoch': 1.81}

 60%|██████    | 8023/13317 [1:22:43<54:11,  1.63it/s]
 60%|██████    | 8024/13317 [1:22:43<54:09,  1.63it/s]
                                                      
{'loss': 0.2105, 'grad_norm': 0.5568983554840088, 'learning_rate': 4.088339118745293e-06, 'epoch': 1.81}

 60%|██████    | 8024/13317 [1:22:44<54:09,  1.63it/s]
 60%|██████    | 8025/13317 [1:22:44<54:09,  1.63it/s]
                                                      
{'loss': 0.2178, 'grad_norm': 0.6019482612609863, 'learning_rate': 4.087050485039577e-06, 'epoch': 1.81}

 60%|██████    | 8025/13317 [1:22:44<54:09,  1.63it/s]
 60%|██████    | 8026/13317 [1:22:45<54:07,  1.63it/s]
                                                      
{'loss': 0.2282, 'grad_norm': 0.5402526259422302, 'learning_rate': 4.085761914063158e-06, 'epoch': 1.81}

 60%|██████    | 8026/13317 [1:22:45<54:07,  1.63it/s]
 60%|██████    | 8027/13317 [1:22:45<54:08,  1.63it/s]
                                                      
{'loss': 0.2799, 'grad_norm': 0.5905720591545105, 'learning_rate': 4.084473405904574e-06, 'epoch': 1.81}

 60%|██████    | 8027/13317 [1:22:45<54:08,  1.63it/s]
 60%|██████    | 8028/13317 [1:22:46<54:07,  1.63it/s]
                                                      
{'loss': 0.1866, 'grad_norm': 0.4678596258163452, 'learning_rate': 4.083184960652363e-06, 'epoch': 1.81}

 60%|██████    | 8028/13317 [1:22:46<54:07,  1.63it/s]
 60%|██████    | 8029/13317 [1:22:46<54:05,  1.63it/s]
                                                      
{'loss': 0.1919, 'grad_norm': 0.5920097231864929, 'learning_rate': 4.081896578395053e-06, 'epoch': 1.81}

 60%|██████    | 8029/13317 [1:22:47<54:05,  1.63it/s]
 60%|██████    | 8030/13317 [1:22:47<54:05,  1.63it/s]
                                                      
{'loss': 0.1941, 'grad_norm': 0.5684961080551147, 'learning_rate': 4.080608259221167e-06, 'epoch': 1.81}

 60%|██████    | 8030/13317 [1:22:47<54:05,  1.63it/s]
 60%|██████    | 8031/13317 [1:22:48<54:04,  1.63it/s]
                                                      
{'loss': 0.1786, 'grad_norm': 0.5096432566642761, 'learning_rate': 4.079320003219231e-06, 'epoch': 1.81}

 60%|██████    | 8031/13317 [1:22:48<54:04,  1.63it/s]
 60%|██████    | 8032/13317 [1:22:48<54:07,  1.63it/s]
                                                      
{'loss': 0.157, 'grad_norm': 0.4776594042778015, 'learning_rate': 4.078031810477757e-06, 'epoch': 1.81}

 60%|██████    | 8032/13317 [1:22:48<54:07,  1.63it/s]
 60%|██████    | 8033/13317 [1:22:49<54:04,  1.63it/s]
                                                      
{'loss': 0.2197, 'grad_norm': 0.5546208620071411, 'learning_rate': 4.076743681085259e-06, 'epoch': 1.81}

 60%|██████    | 8033/13317 [1:22:49<54:04,  1.63it/s]
 60%|██████    | 8034/13317 [1:22:50<54:03,  1.63it/s]
                                                      
{'loss': 0.1762, 'grad_norm': 0.5310531258583069, 'learning_rate': 4.075455615130246e-06, 'epoch': 1.81}

 60%|██████    | 8034/13317 [1:22:50<54:03,  1.63it/s]
 60%|██████    | 8035/13317 [1:22:50<54:01,  1.63it/s]
                                                      
{'loss': 0.2123, 'grad_norm': 0.6065061688423157, 'learning_rate': 4.074167612701223e-06, 'epoch': 1.81}

 60%|██████    | 8035/13317 [1:22:50<54:01,  1.63it/s]
 60%|██████    | 8036/13317 [1:22:51<54:01,  1.63it/s]
                                                      
{'loss': 0.2366, 'grad_norm': 0.5692067742347717, 'learning_rate': 4.0728796738866875e-06, 'epoch': 1.81}

 60%|██████    | 8036/13317 [1:22:51<54:01,  1.63it/s]
 60%|██████    | 8037/13317 [1:22:51<54:03,  1.63it/s]
                                                      
{'loss': 0.1901, 'grad_norm': 0.4561771750450134, 'learning_rate': 4.071591798775134e-06, 'epoch': 1.81}

 60%|██████    | 8037/13317 [1:22:51<54:03,  1.63it/s]
 60%|██████    | 8038/13317 [1:22:52<54:00,  1.63it/s]
                                                      
{'loss': 0.1732, 'grad_norm': 0.4622653126716614, 'learning_rate': 4.070303987455055e-06, 'epoch': 1.81}

 60%|██████    | 8038/13317 [1:22:52<54:00,  1.63it/s]
 60%|██████    | 8039/13317 [1:22:53<54:02,  1.63it/s]
                                                      
{'loss': 0.1806, 'grad_norm': 0.5762357115745544, 'learning_rate': 4.069016240014935e-06, 'epoch': 1.81}

 60%|██████    | 8039/13317 [1:22:53<54:02,  1.63it/s]
 60%|██████    | 8040/13317 [1:22:53<54:00,  1.63it/s]
                                                      
{'loss': 0.2064, 'grad_norm': 0.5377437472343445, 'learning_rate': 4.067728556543256e-06, 'epoch': 1.81}

 60%|██████    | 8040/13317 [1:22:53<54:00,  1.63it/s]
 60%|██████    | 8041/13317 [1:22:54<53:58,  1.63it/s]
                                                      
{'loss': 0.2454, 'grad_norm': 0.5402572154998779, 'learning_rate': 4.0664409371284994e-06, 'epoch': 1.81}

 60%|██████    | 8041/13317 [1:22:54<53:58,  1.63it/s]
 60%|██████    | 8042/13317 [1:22:54<53:58,  1.63it/s]
                                                      
{'loss': 0.1637, 'grad_norm': 0.4819873869419098, 'learning_rate': 4.065153381859132e-06, 'epoch': 1.81}

 60%|██████    | 8042/13317 [1:22:55<53:58,  1.63it/s]
 60%|██████    | 8043/13317 [1:22:55<53:57,  1.63it/s]
                                                      
{'loss': 0.1503, 'grad_norm': 0.46468761563301086, 'learning_rate': 4.063865890823626e-06, 'epoch': 1.81}

 60%|██████    | 8043/13317 [1:22:55<53:57,  1.63it/s]
 60%|██████    | 8044/13317 [1:22:56<53:57,  1.63it/s]
                                                      
{'loss': 0.1951, 'grad_norm': 0.48672062158584595, 'learning_rate': 4.062578464110446e-06, 'epoch': 1.81}

 60%|██████    | 8044/13317 [1:22:56<53:57,  1.63it/s]
 60%|██████    | 8045/13317 [1:22:56<53:58,  1.63it/s]
                                                      
{'loss': 0.2111, 'grad_norm': 0.6084371209144592, 'learning_rate': 4.061291101808051e-06, 'epoch': 1.81}

 60%|██████    | 8045/13317 [1:22:56<53:58,  1.63it/s]
 60%|██████    | 8046/13317 [1:22:57<53:57,  1.63it/s]
                                                      
{'loss': 0.1623, 'grad_norm': 0.4522285461425781, 'learning_rate': 4.060003804004897e-06, 'epoch': 1.81}

 60%|██████    | 8046/13317 [1:22:57<53:57,  1.63it/s]
 60%|██████    | 8047/13317 [1:22:58<54:01,  1.63it/s]
                                                      
{'loss': 0.2378, 'grad_norm': 0.547892153263092, 'learning_rate': 4.0587165707894326e-06, 'epoch': 1.81}

 60%|██████    | 8047/13317 [1:22:58<54:01,  1.63it/s]
 60%|██████    | 8048/13317 [1:22:58<53:57,  1.63it/s]
                                                      
{'loss': 0.1923, 'grad_norm': 0.4754466712474823, 'learning_rate': 4.057429402250109e-06, 'epoch': 1.81}

 60%|██████    | 8048/13317 [1:22:58<53:57,  1.63it/s]
 60%|██████    | 8049/13317 [1:22:59<53:55,  1.63it/s]
                                                      
{'loss': 0.196, 'grad_norm': 0.5287410020828247, 'learning_rate': 4.056142298475363e-06, 'epoch': 1.81}

 60%|██████    | 8049/13317 [1:22:59<53:55,  1.63it/s]
 60%|██████    | 8050/13317 [1:22:59<53:53,  1.63it/s]
                                                      
{'loss': 0.194, 'grad_norm': 0.4859565198421478, 'learning_rate': 4.054855259553636e-06, 'epoch': 1.81}

 60%|██████    | 8050/13317 [1:22:59<53:53,  1.63it/s]
 60%|██████    | 8051/13317 [1:23:00<53:54,  1.63it/s]
                                                      
{'loss': 0.1594, 'grad_norm': 0.4591328203678131, 'learning_rate': 4.05356828557336e-06, 'epoch': 1.81}

 60%|██████    | 8051/13317 [1:23:00<53:54,  1.63it/s]
 60%|██████    | 8052/13317 [1:23:01<53:59,  1.63it/s]
                                                      
{'loss': 0.2189, 'grad_norm': 0.5648224353790283, 'learning_rate': 4.052281376622964e-06, 'epoch': 1.81}

 60%|██████    | 8052/13317 [1:23:01<53:59,  1.63it/s]
 60%|██████    | 8053/13317 [1:23:01<53:55,  1.63it/s]
                                                      
{'loss': 0.2094, 'grad_norm': 0.4555523693561554, 'learning_rate': 4.050994532790871e-06, 'epoch': 1.81}

 60%|██████    | 8053/13317 [1:23:01<53:55,  1.63it/s]
 60%|██████    | 8054/13317 [1:23:02<53:53,  1.63it/s]
                                                      
{'loss': 0.1649, 'grad_norm': 0.49330174922943115, 'learning_rate': 4.049707754165503e-06, 'epoch': 1.81}

 60%|██████    | 8054/13317 [1:23:02<53:53,  1.63it/s]
 60%|██████    | 8055/13317 [1:23:02<53:51,  1.63it/s]
                                                      
{'loss': 0.2154, 'grad_norm': 0.5100939869880676, 'learning_rate': 4.048421040835275e-06, 'epoch': 1.81}

 60%|██████    | 8055/13317 [1:23:03<53:51,  1.63it/s]
 60%|██████    | 8056/13317 [1:23:03<53:49,  1.63it/s]
                                                      
{'loss': 0.1577, 'grad_norm': 0.47121918201446533, 'learning_rate': 4.047134392888595e-06, 'epoch': 1.81}

 60%|██████    | 8056/13317 [1:23:03<53:49,  1.63it/s]
 61%|██████    | 8057/13317 [1:23:04<53:52,  1.63it/s]
                                                      
{'loss': 0.2219, 'grad_norm': 0.5647776126861572, 'learning_rate': 4.045847810413871e-06, 'epoch': 1.82}

 61%|██████    | 8057/13317 [1:23:04<53:52,  1.63it/s]
 61%|██████    | 8058/13317 [1:23:04<53:51,  1.63it/s]
                                                      
{'loss': 0.2011, 'grad_norm': 0.5220916271209717, 'learning_rate': 4.0445612934995075e-06, 'epoch': 1.82}

 61%|██████    | 8058/13317 [1:23:04<53:51,  1.63it/s]
 61%|██████    | 8059/13317 [1:23:05<53:48,  1.63it/s]
                                                      
{'loss': 0.261, 'grad_norm': 0.6117851138114929, 'learning_rate': 4.043274842233899e-06, 'epoch': 1.82}

 61%|██████    | 8059/13317 [1:23:05<53:48,  1.63it/s]
 61%|██████    | 8060/13317 [1:23:05<53:46,  1.63it/s]
                                                      
{'loss': 0.133, 'grad_norm': 0.41205281019210815, 'learning_rate': 4.041988456705436e-06, 'epoch': 1.82}

 61%|██████    | 8060/13317 [1:23:06<53:46,  1.63it/s]
 61%|██████    | 8061/13317 [1:23:06<53:46,  1.63it/s]
                                                      
{'loss': 0.1741, 'grad_norm': 0.4741479158401489, 'learning_rate': 4.040702137002511e-06, 'epoch': 1.82}

 61%|██████    | 8061/13317 [1:23:06<53:46,  1.63it/s]
 61%|██████    | 8062/13317 [1:23:07<54:03,  1.62it/s]
                                                      
{'loss': 0.1412, 'grad_norm': 0.40167036652565, 'learning_rate': 4.039415883213508e-06, 'epoch': 1.82}

 61%|██████    | 8062/13317 [1:23:07<54:03,  1.62it/s]
 61%|██████    | 8063/13317 [1:23:07<53:57,  1.62it/s]
                                                      
{'loss': 0.2198, 'grad_norm': 0.48966184258461, 'learning_rate': 4.038129695426804e-06, 'epoch': 1.82}

 61%|██████    | 8063/13317 [1:23:07<53:57,  1.62it/s]
 61%|██████    | 8064/13317 [1:23:08<53:52,  1.63it/s]
                                                      
{'loss': 0.1788, 'grad_norm': 0.45835062861442566, 'learning_rate': 4.036843573730774e-06, 'epoch': 1.82}

 61%|██████    | 8064/13317 [1:23:08<53:52,  1.63it/s]
 61%|██████    | 8065/13317 [1:23:09<53:48,  1.63it/s]
                                                      
{'loss': 0.2245, 'grad_norm': 0.4905811846256256, 'learning_rate': 4.035557518213789e-06, 'epoch': 1.82}

 61%|██████    | 8065/13317 [1:23:09<53:48,  1.63it/s]
 61%|██████    | 8066/13317 [1:23:09<53:45,  1.63it/s]
                                                      
{'loss': 0.1718, 'grad_norm': 0.47821399569511414, 'learning_rate': 4.034271528964214e-06, 'epoch': 1.82}

 61%|██████    | 8066/13317 [1:23:09<53:45,  1.63it/s]
 61%|██████    | 8067/13317 [1:23:10<53:47,  1.63it/s]
                                                      
{'loss': 0.1639, 'grad_norm': 0.47232168912887573, 'learning_rate': 4.032985606070409e-06, 'epoch': 1.82}

 61%|██████    | 8067/13317 [1:23:10<53:47,  1.63it/s]
 61%|██████    | 8068/13317 [1:23:10<53:45,  1.63it/s]
                                                      
{'loss': 0.1841, 'grad_norm': 0.4840155839920044, 'learning_rate': 4.0316997496207326e-06, 'epoch': 1.82}

 61%|██████    | 8068/13317 [1:23:11<53:45,  1.63it/s]
 61%|██████    | 8069/13317 [1:23:11<53:44,  1.63it/s]
                                                      
{'loss': 0.2226, 'grad_norm': 0.5786517858505249, 'learning_rate': 4.030413959703536e-06, 'epoch': 1.82}

 61%|██████    | 8069/13317 [1:23:11<53:44,  1.63it/s]
 61%|██████    | 8070/13317 [1:23:12<53:42,  1.63it/s]
                                                      
{'loss': 0.1906, 'grad_norm': 0.5664747953414917, 'learning_rate': 4.0291282364071675e-06, 'epoch': 1.82}

 61%|██████    | 8070/13317 [1:23:12<53:42,  1.63it/s]
 61%|██████    | 8071/13317 [1:23:12<53:40,  1.63it/s]
                                                      
{'loss': 0.1938, 'grad_norm': 0.5633617639541626, 'learning_rate': 4.027842579819967e-06, 'epoch': 1.82}

 61%|██████    | 8071/13317 [1:23:12<53:40,  1.63it/s]
 61%|██████    | 8072/13317 [1:23:13<53:44,  1.63it/s]
                                                      
{'loss': 0.1901, 'grad_norm': 0.5160545706748962, 'learning_rate': 4.026556990030277e-06, 'epoch': 1.82}

 61%|██████    | 8072/13317 [1:23:13<53:44,  1.63it/s]
 61%|██████    | 8073/13317 [1:23:13<53:41,  1.63it/s]
                                                      
{'loss': 0.1742, 'grad_norm': 0.4434663951396942, 'learning_rate': 4.025271467126427e-06, 'epoch': 1.82}

 61%|██████    | 8073/13317 [1:23:14<53:41,  1.63it/s]
 61%|██████    | 8074/13317 [1:23:14<53:39,  1.63it/s]
                                                      
{'loss': 0.1848, 'grad_norm': 0.5113449096679688, 'learning_rate': 4.023986011196747e-06, 'epoch': 1.82}

 61%|██████    | 8074/13317 [1:23:14<53:39,  1.63it/s]
 61%|██████    | 8075/13317 [1:23:15<53:39,  1.63it/s]
                                                      
{'loss': 0.1971, 'grad_norm': 0.5597386360168457, 'learning_rate': 4.022700622329565e-06, 'epoch': 1.82}

 61%|██████    | 8075/13317 [1:23:15<53:39,  1.63it/s]
 61%|██████    | 8076/13317 [1:23:15<53:40,  1.63it/s]
                                                      
{'loss': 0.1661, 'grad_norm': 0.4933639466762543, 'learning_rate': 4.021415300613196e-06, 'epoch': 1.82}

 61%|██████    | 8076/13317 [1:23:15<53:40,  1.63it/s]
 61%|██████    | 8077/13317 [1:23:16<53:43,  1.63it/s]
                                                      
{'loss': 0.2129, 'grad_norm': 0.5942857265472412, 'learning_rate': 4.020130046135957e-06, 'epoch': 1.82}

 61%|██████    | 8077/13317 [1:23:16<53:43,  1.63it/s]
 61%|██████    | 8078/13317 [1:23:17<53:40,  1.63it/s]
                                                      
{'loss': 0.2244, 'grad_norm': 0.5503771901130676, 'learning_rate': 4.01884485898616e-06, 'epoch': 1.82}

 61%|██████    | 8078/13317 [1:23:17<53:40,  1.63it/s]
 61%|██████    | 8079/13317 [1:23:17<53:38,  1.63it/s]
                                                      
{'loss': 0.1885, 'grad_norm': 0.45732352137565613, 'learning_rate': 4.01755973925211e-06, 'epoch': 1.82}

 61%|██████    | 8079/13317 [1:23:17<53:38,  1.63it/s]
 61%|██████    | 8080/13317 [1:23:18<53:41,  1.63it/s]
                                                      
{'loss': 0.2233, 'grad_norm': 0.6237924695014954, 'learning_rate': 4.016274687022107e-06, 'epoch': 1.82}

 61%|██████    | 8080/13317 [1:23:18<53:41,  1.63it/s]
 61%|██████    | 8081/13317 [1:23:18<53:39,  1.63it/s]
                                                      
{'loss': 0.1537, 'grad_norm': 0.4765283167362213, 'learning_rate': 4.014989702384449e-06, 'epoch': 1.82}

 61%|██████    | 8081/13317 [1:23:19<53:39,  1.63it/s]
 61%|██████    | 8082/13317 [1:23:19<53:41,  1.63it/s]
                                                      
{'loss': 0.1578, 'grad_norm': 0.469203382730484, 'learning_rate': 4.013704785427429e-06, 'epoch': 1.82}

 61%|██████    | 8082/13317 [1:23:19<53:41,  1.63it/s]
 61%|██████    | 8083/13317 [1:23:20<53:38,  1.63it/s]
                                                      
{'loss': 0.2011, 'grad_norm': 0.5658526420593262, 'learning_rate': 4.012419936239333e-06, 'epoch': 1.82}

 61%|██████    | 8083/13317 [1:23:20<53:38,  1.63it/s]
 61%|██████    | 8084/13317 [1:23:20<53:35,  1.63it/s]
                                                      
{'loss': 0.1746, 'grad_norm': 0.5056454539299011, 'learning_rate': 4.0111351549084435e-06, 'epoch': 1.82}

 61%|██████    | 8084/13317 [1:23:20<53:35,  1.63it/s]
 61%|██████    | 8085/13317 [1:23:21<53:38,  1.63it/s]
                                                      
{'loss': 0.238, 'grad_norm': 0.5875850915908813, 'learning_rate': 4.00985044152304e-06, 'epoch': 1.82}

 61%|██████    | 8085/13317 [1:23:21<53:38,  1.63it/s]
 61%|██████    | 8086/13317 [1:23:21<53:38,  1.63it/s]
                                                      
{'loss': 0.1802, 'grad_norm': 0.451984703540802, 'learning_rate': 4.008565796171394e-06, 'epoch': 1.82}

 61%|██████    | 8086/13317 [1:23:22<53:38,  1.63it/s]
 61%|██████    | 8087/13317 [1:23:22<53:41,  1.62it/s]
                                                      
{'loss': 0.1806, 'grad_norm': 0.4699399471282959, 'learning_rate': 4.0072812189417744e-06, 'epoch': 1.82}

 61%|██████    | 8087/13317 [1:23:22<53:41,  1.62it/s]
 61%|██████    | 8088/13317 [1:23:23<53:36,  1.63it/s]
                                                      
{'loss': 0.2047, 'grad_norm': 0.5508679151535034, 'learning_rate': 4.005996709922448e-06, 'epoch': 1.82}

 61%|██████    | 8088/13317 [1:23:23<53:36,  1.63it/s]
 61%|██████    | 8089/13317 [1:23:23<53:33,  1.63it/s]
                                                      
{'loss': 0.1922, 'grad_norm': 0.49577581882476807, 'learning_rate': 4.004712269201672e-06, 'epoch': 1.82}

 61%|██████    | 8089/13317 [1:23:23<53:33,  1.63it/s]
 61%|██████    | 8090/13317 [1:23:24<53:35,  1.63it/s]
                                                      
{'loss': 0.2011, 'grad_norm': 0.5371473431587219, 'learning_rate': 4.003427896867701e-06, 'epoch': 1.82}

 61%|██████    | 8090/13317 [1:23:24<53:35,  1.63it/s]
 61%|██████    | 8091/13317 [1:23:25<53:32,  1.63it/s]
                                                      
{'loss': 0.1627, 'grad_norm': 0.44351235032081604, 'learning_rate': 4.002143593008786e-06, 'epoch': 1.82}

 61%|██████    | 8091/13317 [1:23:25<53:32,  1.63it/s]
 61%|██████    | 8092/13317 [1:23:25<53:35,  1.63it/s]
                                                      
{'loss': 0.2186, 'grad_norm': 0.6344464421272278, 'learning_rate': 4.000859357713171e-06, 'epoch': 1.82}

 61%|██████    | 8092/13317 [1:23:25<53:35,  1.63it/s]
 61%|██████    | 8093/13317 [1:23:26<53:31,  1.63it/s]
                                                      
{'loss': 0.1494, 'grad_norm': 0.4264751672744751, 'learning_rate': 3.999575191069098e-06, 'epoch': 1.82}

 61%|██████    | 8093/13317 [1:23:26<53:31,  1.63it/s]
 61%|██████    | 8094/13317 [1:23:26<53:27,  1.63it/s]
                                                      
{'loss': 0.3671, 'grad_norm': 0.627530038356781, 'learning_rate': 3.9982910931648e-06, 'epoch': 1.82}

 61%|██████    | 8094/13317 [1:23:27<53:27,  1.63it/s]
 61%|██████    | 8095/13317 [1:23:27<53:26,  1.63it/s]
                                                      
{'loss': 0.1938, 'grad_norm': 0.4769577383995056, 'learning_rate': 3.997007064088511e-06, 'epoch': 1.82}

 61%|██████    | 8095/13317 [1:23:27<53:26,  1.63it/s]
 61%|██████    | 8096/13317 [1:23:28<53:25,  1.63it/s]
                                                      
{'loss': 0.1633, 'grad_norm': 0.44069141149520874, 'learning_rate': 3.9957231039284576e-06, 'epoch': 1.82}

 61%|██████    | 8096/13317 [1:23:28<53:25,  1.63it/s]
 61%|██████    | 8097/13317 [1:23:28<53:28,  1.63it/s]
                                                      
{'loss': 0.1844, 'grad_norm': 0.5165771245956421, 'learning_rate': 3.99443921277286e-06, 'epoch': 1.82}

 61%|██████    | 8097/13317 [1:23:28<53:28,  1.63it/s]
 61%|██████    | 8098/13317 [1:23:29<53:24,  1.63it/s]
                                                      
{'loss': 0.1637, 'grad_norm': 0.46774256229400635, 'learning_rate': 3.993155390709935e-06, 'epoch': 1.82}

 61%|██████    | 8098/13317 [1:23:29<53:24,  1.63it/s]
 61%|██████    | 8099/13317 [1:23:29<53:23,  1.63it/s]
                                                      
{'loss': 0.1772, 'grad_norm': 0.4739975929260254, 'learning_rate': 3.991871637827897e-06, 'epoch': 1.82}

 61%|██████    | 8099/13317 [1:23:30<53:23,  1.63it/s]
 61%|██████    | 8100/13317 [1:23:30<53:23,  1.63it/s]
                                                      
{'loss': 0.2065, 'grad_norm': 0.5340814590454102, 'learning_rate': 3.99058795421495e-06, 'epoch': 1.82}

 61%|██████    | 8100/13317 [1:23:30<53:23,  1.63it/s]
 61%|██████    | 8101/13317 [1:23:31<53:23,  1.63it/s]
                                                      
{'loss': 0.2037, 'grad_norm': 0.5566793084144592, 'learning_rate': 3.989304339959298e-06, 'epoch': 1.82}

 61%|██████    | 8101/13317 [1:23:31<53:23,  1.63it/s]
 61%|██████    | 8102/13317 [1:23:31<53:28,  1.63it/s]
                                                      
{'loss': 0.2566, 'grad_norm': 0.7316734790802002, 'learning_rate': 3.98802079514914e-06, 'epoch': 1.83}

 61%|██████    | 8102/13317 [1:23:31<53:28,  1.63it/s]
 61%|██████    | 8103/13317 [1:23:32<53:25,  1.63it/s]
                                                      
{'loss': 0.1587, 'grad_norm': 0.49004364013671875, 'learning_rate': 3.986737319872669e-06, 'epoch': 1.83}

 61%|██████    | 8103/13317 [1:23:32<53:25,  1.63it/s]
 61%|██████    | 8104/13317 [1:23:33<53:24,  1.63it/s]
                                                      
{'loss': 0.1995, 'grad_norm': 0.47748613357543945, 'learning_rate': 3.985453914218072e-06, 'epoch': 1.83}

 61%|██████    | 8104/13317 [1:23:33<53:24,  1.63it/s]
 61%|██████    | 8105/13317 [1:23:33<53:26,  1.63it/s]
                                                      
{'loss': 0.2751, 'grad_norm': 0.5578868389129639, 'learning_rate': 3.984170578273533e-06, 'epoch': 1.83}

 61%|██████    | 8105/13317 [1:23:33<53:26,  1.63it/s]
 61%|██████    | 8106/13317 [1:23:34<53:24,  1.63it/s]
                                                      
{'loss': 0.2467, 'grad_norm': 0.603814959526062, 'learning_rate': 3.982887312127232e-06, 'epoch': 1.83}

 61%|██████    | 8106/13317 [1:23:34<53:24,  1.63it/s]
 61%|██████    | 8107/13317 [1:23:34<53:25,  1.63it/s]
                                                      
{'loss': 0.2414, 'grad_norm': 0.6380345225334167, 'learning_rate': 3.9816041158673405e-06, 'epoch': 1.83}

 61%|██████    | 8107/13317 [1:23:35<53:25,  1.63it/s]
 61%|██████    | 8108/13317 [1:23:35<53:26,  1.62it/s]
                                                      
{'loss': 0.1656, 'grad_norm': 0.4589832127094269, 'learning_rate': 3.98032098958203e-06, 'epoch': 1.83}

 61%|██████    | 8108/13317 [1:23:35<53:26,  1.62it/s]
 61%|██████    | 8109/13317 [1:23:36<53:29,  1.62it/s]
                                                      
{'loss': 0.1554, 'grad_norm': 0.42032870650291443, 'learning_rate': 3.979037933359465e-06, 'epoch': 1.83}

 61%|██████    | 8109/13317 [1:23:36<53:29,  1.62it/s]
 61%|██████    | 8110/13317 [1:23:36<53:29,  1.62it/s]
                                                      
{'loss': 0.2135, 'grad_norm': 0.5219689607620239, 'learning_rate': 3.977754947287803e-06, 'epoch': 1.83}

 61%|██████    | 8110/13317 [1:23:36<53:29,  1.62it/s]
 61%|██████    | 8111/13317 [1:23:37<53:27,  1.62it/s]
                                                      
{'loss': 0.2228, 'grad_norm': 0.5510971546173096, 'learning_rate': 3.9764720314552e-06, 'epoch': 1.83}

 61%|██████    | 8111/13317 [1:23:37<53:27,  1.62it/s]
 61%|██████    | 8112/13317 [1:23:37<53:29,  1.62it/s]
                                                      
{'loss': 0.1553, 'grad_norm': 0.4414577782154083, 'learning_rate': 3.975189185949805e-06, 'epoch': 1.83}

 61%|██████    | 8112/13317 [1:23:38<53:29,  1.62it/s]
 61%|██████    | 8113/13317 [1:23:38<53:27,  1.62it/s]
                                                      
{'loss': 0.1991, 'grad_norm': 0.4833833575248718, 'learning_rate': 3.973906410859767e-06, 'epoch': 1.83}

 61%|██████    | 8113/13317 [1:23:38<53:27,  1.62it/s]
 61%|██████    | 8114/13317 [1:23:39<53:22,  1.62it/s]
                                                      
{'loss': 0.1714, 'grad_norm': 0.490119993686676, 'learning_rate': 3.972623706273219e-06, 'epoch': 1.83}

 61%|██████    | 8114/13317 [1:23:39<53:22,  1.62it/s]
 61%|██████    | 8115/13317 [1:23:39<53:22,  1.62it/s]
                                                      
{'loss': 0.1953, 'grad_norm': 0.5276403427124023, 'learning_rate': 3.971341072278302e-06, 'epoch': 1.83}

 61%|██████    | 8115/13317 [1:23:39<53:22,  1.62it/s]
 61%|██████    | 8116/13317 [1:23:40<53:20,  1.63it/s]
                                                      
{'loss': 0.2421, 'grad_norm': 0.5847195982933044, 'learning_rate': 3.970058508963146e-06, 'epoch': 1.83}

 61%|██████    | 8116/13317 [1:23:40<53:20,  1.63it/s]
 61%|██████    | 8117/13317 [1:23:41<53:21,  1.62it/s]
                                                      
{'loss': 0.2493, 'grad_norm': 0.6793063282966614, 'learning_rate': 3.968776016415875e-06, 'epoch': 1.83}

 61%|██████    | 8117/13317 [1:23:41<53:21,  1.62it/s]
 61%|██████    | 8118/13317 [1:23:41<53:19,  1.62it/s]
                                                      
{'loss': 0.166, 'grad_norm': 0.5293914079666138, 'learning_rate': 3.9674935947246096e-06, 'epoch': 1.83}

 61%|██████    | 8118/13317 [1:23:41<53:19,  1.62it/s]
 61%|██████    | 8119/13317 [1:23:42<53:21,  1.62it/s]
                                                      
{'loss': 0.1766, 'grad_norm': 0.484803706407547, 'learning_rate': 3.966211243977468e-06, 'epoch': 1.83}

 61%|██████    | 8119/13317 [1:23:42<53:21,  1.62it/s]
 61%|██████    | 8120/13317 [1:23:42<53:17,  1.63it/s]
                                                      
{'loss': 0.204, 'grad_norm': 0.555545449256897, 'learning_rate': 3.964928964262559e-06, 'epoch': 1.83}

 61%|██████    | 8120/13317 [1:23:43<53:17,  1.63it/s]
 61%|██████    | 8121/13317 [1:23:43<53:13,  1.63it/s]
                                                      
{'loss': 0.2015, 'grad_norm': 0.5030943751335144, 'learning_rate': 3.963646755667988e-06, 'epoch': 1.83}

 61%|██████    | 8121/13317 [1:23:43<53:13,  1.63it/s]
 61%|██████    | 8122/13317 [1:23:44<53:17,  1.62it/s]
                                                      
{'loss': 0.1793, 'grad_norm': 0.49578559398651123, 'learning_rate': 3.96236461828186e-06, 'epoch': 1.83}

 61%|██████    | 8122/13317 [1:23:44<53:17,  1.62it/s]
 61%|██████    | 8123/13317 [1:23:44<53:15,  1.63it/s]
                                                      
{'loss': 0.176, 'grad_norm': 0.4824337065219879, 'learning_rate': 3.961082552192269e-06, 'epoch': 1.83}

 61%|██████    | 8123/13317 [1:23:44<53:15,  1.63it/s]
 61%|██████    | 8124/13317 [1:23:45<53:13,  1.63it/s]
                                                      
{'loss': 0.2399, 'grad_norm': 0.5944125056266785, 'learning_rate': 3.959800557487307e-06, 'epoch': 1.83}

 61%|██████    | 8124/13317 [1:23:45<53:13,  1.63it/s]
 61%|██████    | 8125/13317 [1:23:45<53:14,  1.63it/s]
                                                      
{'loss': 0.2415, 'grad_norm': 0.6436695456504822, 'learning_rate': 3.95851863425506e-06, 'epoch': 1.83}

 61%|██████    | 8125/13317 [1:23:46<53:14,  1.63it/s]
 61%|██████    | 8126/13317 [1:23:46<53:11,  1.63it/s]
                                                      
{'loss': 0.2012, 'grad_norm': 0.4874963164329529, 'learning_rate': 3.957236782583612e-06, 'epoch': 1.83}

 61%|██████    | 8126/13317 [1:23:46<53:11,  1.63it/s]
 61%|██████    | 8127/13317 [1:23:47<53:15,  1.62it/s]
                                                      
{'loss': 0.1836, 'grad_norm': 0.4639379382133484, 'learning_rate': 3.9559550025610364e-06, 'epoch': 1.83}

 61%|██████    | 8127/13317 [1:23:47<53:15,  1.62it/s]
 61%|██████    | 8128/13317 [1:23:47<53:11,  1.63it/s]
                                                      
{'loss': 0.2024, 'grad_norm': 0.5199019312858582, 'learning_rate': 3.9546732942754055e-06, 'epoch': 1.83}

 61%|██████    | 8128/13317 [1:23:47<53:11,  1.63it/s]
 61%|██████    | 8129/13317 [1:23:48<53:07,  1.63it/s]
                                                      
{'loss': 0.2474, 'grad_norm': 0.6053148508071899, 'learning_rate': 3.953391657814789e-06, 'epoch': 1.83}

 61%|██████    | 8129/13317 [1:23:48<53:07,  1.63it/s]
 61%|██████    | 8130/13317 [1:23:49<53:10,  1.63it/s]
                                                      
{'loss': 0.2382, 'grad_norm': 0.5007274150848389, 'learning_rate': 3.952110093267247e-06, 'epoch': 1.83}

 61%|██████    | 8130/13317 [1:23:49<53:10,  1.63it/s]
 61%|██████    | 8131/13317 [1:23:49<53:07,  1.63it/s]
                                                      
{'loss': 0.2401, 'grad_norm': 0.6441068053245544, 'learning_rate': 3.950828600720838e-06, 'epoch': 1.83}

 61%|██████    | 8131/13317 [1:23:49<53:07,  1.63it/s]
 61%|██████    | 8132/13317 [1:23:50<53:19,  1.62it/s]
                                                      
{'loss': 0.2713, 'grad_norm': 0.6491110920906067, 'learning_rate': 3.94954718026361e-06, 'epoch': 1.83}

 61%|██████    | 8132/13317 [1:23:50<53:19,  1.62it/s]
 61%|██████    | 8133/13317 [1:23:50<53:15,  1.62it/s]
                                                      
{'loss': 0.2165, 'grad_norm': 0.47641655802726746, 'learning_rate': 3.948265831983616e-06, 'epoch': 1.83}

 61%|██████    | 8133/13317 [1:23:51<53:15,  1.62it/s]
 61%|██████    | 8134/13317 [1:23:51<53:09,  1.62it/s]
                                                      
{'loss': 0.1858, 'grad_norm': 0.5703856348991394, 'learning_rate': 3.9469845559688935e-06, 'epoch': 1.83}

 61%|██████    | 8134/13317 [1:23:51<53:09,  1.62it/s]
 61%|██████    | 8135/13317 [1:23:52<53:08,  1.63it/s]
                                                      
{'loss': 0.2229, 'grad_norm': 0.5892605185508728, 'learning_rate': 3.945703352307482e-06, 'epoch': 1.83}

 61%|██████    | 8135/13317 [1:23:52<53:08,  1.63it/s]
 61%|██████    | 8136/13317 [1:23:52<53:07,  1.63it/s]
                                                      
{'loss': 0.1773, 'grad_norm': 0.459276020526886, 'learning_rate': 3.9444222210874146e-06, 'epoch': 1.83}

 61%|██████    | 8136/13317 [1:23:52<53:07,  1.63it/s]
 61%|██████    | 8137/13317 [1:23:53<53:09,  1.62it/s]
                                                      
{'loss': 0.2255, 'grad_norm': 0.6298724412918091, 'learning_rate': 3.943141162396716e-06, 'epoch': 1.83}

 61%|██████    | 8137/13317 [1:23:53<53:09,  1.62it/s]
 61%|██████    | 8138/13317 [1:23:53<53:06,  1.63it/s]
                                                      
{'loss': 0.1773, 'grad_norm': 0.4681467115879059, 'learning_rate': 3.94186017632341e-06, 'epoch': 1.83}

 61%|██████    | 8138/13317 [1:23:54<53:06,  1.63it/s]
 61%|██████    | 8139/13317 [1:23:54<53:23,  1.62it/s]
                                                      
{'loss': 0.1967, 'grad_norm': 0.5511682033538818, 'learning_rate': 3.940579262955512e-06, 'epoch': 1.83}

 61%|██████    | 8139/13317 [1:23:54<53:23,  1.62it/s]
 61%|██████    | 8140/13317 [1:23:55<53:26,  1.61it/s]
                                                      
{'loss': 0.1657, 'grad_norm': 0.4267832636833191, 'learning_rate': 3.93929842238104e-06, 'epoch': 1.83}

 61%|██████    | 8140/13317 [1:23:55<53:26,  1.61it/s]
 61%|██████    | 8141/13317 [1:23:55<53:25,  1.61it/s]
                                                      
{'loss': 0.2295, 'grad_norm': 0.5895342826843262, 'learning_rate': 3.9380176546879925e-06, 'epoch': 1.83}

 61%|██████    | 8141/13317 [1:23:55<53:25,  1.61it/s]
 61%|██████    | 8142/13317 [1:23:56<53:15,  1.62it/s]
                                                      
{'loss': 0.1987, 'grad_norm': 0.5532308220863342, 'learning_rate': 3.936736959964379e-06, 'epoch': 1.83}

 61%|██████    | 8142/13317 [1:23:56<53:15,  1.62it/s]
 61%|██████    | 8143/13317 [1:23:57<53:08,  1.62it/s]
                                                      
{'loss': 0.2002, 'grad_norm': 0.4956154227256775, 'learning_rate': 3.935456338298195e-06, 'epoch': 1.83}

 61%|██████    | 8143/13317 [1:23:57<53:08,  1.62it/s]
 61%|██████    | 8144/13317 [1:23:57<53:03,  1.62it/s]
                                                      
{'loss': 0.2258, 'grad_norm': 0.5273988246917725, 'learning_rate': 3.934175789777431e-06, 'epoch': 1.83}

 61%|██████    | 8144/13317 [1:23:57<53:03,  1.62it/s]
 61%|██████    | 8145/13317 [1:23:58<53:01,  1.63it/s]
                                                      
{'loss': 0.1979, 'grad_norm': 0.5907329320907593, 'learning_rate': 3.932895314490075e-06, 'epoch': 1.83}

 61%|██████    | 8145/13317 [1:23:58<53:01,  1.63it/s]
 61%|██████    | 8146/13317 [1:23:58<52:56,  1.63it/s]
                                                      
{'loss': 0.2216, 'grad_norm': 0.5588589310646057, 'learning_rate': 3.9316149125241114e-06, 'epoch': 1.84}

 61%|██████    | 8146/13317 [1:23:59<52:56,  1.63it/s]
 61%|██████    | 8147/13317 [1:23:59<52:57,  1.63it/s]
                                                      
{'loss': 0.189, 'grad_norm': 0.4430118799209595, 'learning_rate': 3.930334583967514e-06, 'epoch': 1.84}

 61%|██████    | 8147/13317 [1:23:59<52:57,  1.63it/s]
 61%|██████    | 8148/13317 [1:24:00<52:57,  1.63it/s]
                                                      
{'loss': 0.1501, 'grad_norm': 0.4146482050418854, 'learning_rate': 3.929054328908256e-06, 'epoch': 1.84}

 61%|██████    | 8148/13317 [1:24:00<52:57,  1.63it/s]
 61%|██████    | 8149/13317 [1:24:00<52:54,  1.63it/s]
                                                      
{'loss': 0.1632, 'grad_norm': 0.4505009949207306, 'learning_rate': 3.927774147434305e-06, 'epoch': 1.84}

 61%|██████    | 8149/13317 [1:24:00<52:54,  1.63it/s]
 61%|██████    | 8150/13317 [1:24:01<52:53,  1.63it/s]
                                                      
{'loss': 0.1622, 'grad_norm': 0.45226413011550903, 'learning_rate': 3.926494039633625e-06, 'epoch': 1.84}

 61%|██████    | 8150/13317 [1:24:01<52:53,  1.63it/s]
 61%|██████    | 8151/13317 [1:24:01<52:52,  1.63it/s]
                                                      
{'loss': 0.2409, 'grad_norm': 0.6398005485534668, 'learning_rate': 3.92521400559417e-06, 'epoch': 1.84}

 61%|██████    | 8151/13317 [1:24:02<52:52,  1.63it/s]
 61%|██████    | 8152/13317 [1:24:02<52:56,  1.63it/s]
                                                      
{'loss': 0.2022, 'grad_norm': 0.4952271282672882, 'learning_rate': 3.923934045403892e-06, 'epoch': 1.84}

 61%|██████    | 8152/13317 [1:24:02<52:56,  1.63it/s]
 61%|██████    | 8153/13317 [1:24:03<52:52,  1.63it/s]
                                                      
{'loss': 0.191, 'grad_norm': 0.5088985562324524, 'learning_rate': 3.9226541591507405e-06, 'epoch': 1.84}

 61%|██████    | 8153/13317 [1:24:03<52:52,  1.63it/s]
 61%|██████    | 8154/13317 [1:24:03<52:51,  1.63it/s]
                                                      
{'loss': 0.1674, 'grad_norm': 0.46948862075805664, 'learning_rate': 3.921374346922653e-06, 'epoch': 1.84}

 61%|██████    | 8154/13317 [1:24:03<52:51,  1.63it/s]
 61%|██████    | 8155/13317 [1:24:04<52:49,  1.63it/s]
                                                      
{'loss': 0.234, 'grad_norm': 0.5728417038917542, 'learning_rate': 3.9200946088075696e-06, 'epoch': 1.84}

 61%|██████    | 8155/13317 [1:24:04<52:49,  1.63it/s]
 61%|██████    | 8156/13317 [1:24:05<52:47,  1.63it/s]
                                                      
{'loss': 0.1791, 'grad_norm': 0.49097731709480286, 'learning_rate': 3.91881494489342e-06, 'epoch': 1.84}

 61%|██████    | 8156/13317 [1:24:05<52:47,  1.63it/s]
 61%|██████▏   | 8157/13317 [1:24:05<55:35,  1.55it/s]
                                                      
{'loss': 0.1642, 'grad_norm': 0.45376384258270264, 'learning_rate': 3.917535355268133e-06, 'epoch': 1.84}

 61%|██████▏   | 8157/13317 [1:24:05<55:35,  1.55it/s]
 61%|██████▏   | 8158/13317 [1:24:06<54:43,  1.57it/s]
                                                      
{'loss': 0.2142, 'grad_norm': 0.5372896790504456, 'learning_rate': 3.916255840019628e-06, 'epoch': 1.84}

 61%|██████▏   | 8158/13317 [1:24:06<54:43,  1.57it/s]
 61%|██████▏   | 8159/13317 [1:24:06<54:08,  1.59it/s]
                                                      
{'loss': 0.2763, 'grad_norm': 0.7759768962860107, 'learning_rate': 3.91497639923582e-06, 'epoch': 1.84}

 61%|██████▏   | 8159/13317 [1:24:07<54:08,  1.59it/s]
 61%|██████▏   | 8160/13317 [1:24:07<53:48,  1.60it/s]
                                                      
{'loss': 0.2394, 'grad_norm': 0.47915300726890564, 'learning_rate': 3.913697033004626e-06, 'epoch': 1.84}

 61%|██████▏   | 8160/13317 [1:24:07<53:48,  1.60it/s]
 61%|██████▏   | 8161/13317 [1:24:08<53:28,  1.61it/s]
                                                      
{'loss': 0.2655, 'grad_norm': 0.5911281704902649, 'learning_rate': 3.912417741413943e-06, 'epoch': 1.84}

 61%|██████▏   | 8161/13317 [1:24:08<53:28,  1.61it/s]
 61%|██████▏   | 8162/13317 [1:24:08<53:19,  1.61it/s]
                                                      
{'loss': 0.1904, 'grad_norm': 0.5030157566070557, 'learning_rate': 3.911138524551679e-06, 'epoch': 1.84}

 61%|██████▏   | 8162/13317 [1:24:09<53:19,  1.61it/s]
 61%|██████▏   | 8163/13317 [1:24:09<55:50,  1.54it/s]
                                                      
{'loss': 0.2414, 'grad_norm': 0.5519351363182068, 'learning_rate': 3.909859382505728e-06, 'epoch': 1.84}

 61%|██████▏   | 8163/13317 [1:24:09<55:50,  1.54it/s]
 61%|██████▏   | 8164/13317 [1:24:10<54:55,  1.56it/s]
                                                      
{'loss': 0.1595, 'grad_norm': 0.4431249499320984, 'learning_rate': 3.908580315363978e-06, 'epoch': 1.84}

 61%|██████▏   | 8164/13317 [1:24:10<54:55,  1.56it/s]
 61%|██████▏   | 8165/13317 [1:24:10<54:17,  1.58it/s]
                                                      
{'loss': 0.1543, 'grad_norm': 0.4522823393344879, 'learning_rate': 3.907301323214318e-06, 'epoch': 1.84}

 61%|██████▏   | 8165/13317 [1:24:10<54:17,  1.58it/s]
 61%|██████▏   | 8166/13317 [1:24:11<53:50,  1.59it/s]
                                                      
{'loss': 0.2075, 'grad_norm': 0.6511449813842773, 'learning_rate': 3.906022406144625e-06, 'epoch': 1.84}

 61%|██████▏   | 8166/13317 [1:24:11<53:50,  1.59it/s]
 61%|██████▏   | 8167/13317 [1:24:12<53:30,  1.60it/s]
                                                      
{'loss': 0.2158, 'grad_norm': 0.6230466365814209, 'learning_rate': 3.904743564242779e-06, 'epoch': 1.84}

 61%|██████▏   | 8167/13317 [1:24:12<53:30,  1.60it/s]
 61%|██████▏   | 8168/13317 [1:24:12<53:19,  1.61it/s]
                                                      
{'loss': 0.217, 'grad_norm': 0.5430140495300293, 'learning_rate': 3.903464797596643e-06, 'epoch': 1.84}

 61%|██████▏   | 8168/13317 [1:24:12<53:19,  1.61it/s]
 61%|██████▏   | 8169/13317 [1:24:13<53:06,  1.62it/s]
                                                      
{'loss': 0.2008, 'grad_norm': 0.5810076594352722, 'learning_rate': 3.902186106294088e-06, 'epoch': 1.84}

 61%|██████▏   | 8169/13317 [1:24:13<53:06,  1.62it/s]
 61%|██████▏   | 8170/13317 [1:24:13<53:00,  1.62it/s]
                                                      
{'loss': 0.168, 'grad_norm': 0.45044901967048645, 'learning_rate': 3.900907490422972e-06, 'epoch': 1.84}

 61%|██████▏   | 8170/13317 [1:24:13<53:00,  1.62it/s]
 61%|██████▏   | 8171/13317 [1:24:14<52:54,  1.62it/s]
                                                      
{'loss': 0.1822, 'grad_norm': 0.4959709346294403, 'learning_rate': 3.899628950071148e-06, 'epoch': 1.84}

 61%|██████▏   | 8171/13317 [1:24:14<52:54,  1.62it/s]
 61%|██████▏   | 8172/13317 [1:24:15<52:49,  1.62it/s]
                                                      
{'loss': 0.1893, 'grad_norm': 0.5563886165618896, 'learning_rate': 3.898350485326466e-06, 'epoch': 1.84}

 61%|██████▏   | 8172/13317 [1:24:15<52:49,  1.62it/s]
 61%|██████▏   | 8173/13317 [1:24:15<52:45,  1.62it/s]
                                                      
{'loss': 0.1791, 'grad_norm': 0.42403265833854675, 'learning_rate': 3.897072096276769e-06, 'epoch': 1.84}

 61%|██████▏   | 8173/13317 [1:24:15<52:45,  1.62it/s]
 61%|██████▏   | 8174/13317 [1:24:16<52:42,  1.63it/s]
                                                      
{'loss': 0.1822, 'grad_norm': 0.533417820930481, 'learning_rate': 3.895793783009899e-06, 'epoch': 1.84}

 61%|██████▏   | 8174/13317 [1:24:16<52:42,  1.63it/s]
 61%|██████▏   | 8175/13317 [1:24:16<53:02,  1.62it/s]
                                                      
{'loss': 0.1756, 'grad_norm': 0.4492940604686737, 'learning_rate': 3.894515545613686e-06, 'epoch': 1.84}

 61%|██████▏   | 8175/13317 [1:24:17<53:02,  1.62it/s]
 61%|██████▏   | 8176/13317 [1:24:17<52:53,  1.62it/s]
                                                      
{'loss': 0.1974, 'grad_norm': 0.4764334559440613, 'learning_rate': 3.8932373841759604e-06, 'epoch': 1.84}

 61%|██████▏   | 8176/13317 [1:24:17<52:53,  1.62it/s]
 61%|██████▏   | 8177/13317 [1:24:18<52:47,  1.62it/s]
                                                      
{'loss': 0.1879, 'grad_norm': 0.5103088617324829, 'learning_rate': 3.891959298784547e-06, 'epoch': 1.84}

 61%|██████▏   | 8177/13317 [1:24:18<52:47,  1.62it/s]
 61%|██████▏   | 8178/13317 [1:24:18<52:42,  1.62it/s]
                                                      
{'loss': 0.2212, 'grad_norm': 0.6251096725463867, 'learning_rate': 3.89068128952726e-06, 'epoch': 1.84}

 61%|██████▏   | 8178/13317 [1:24:18<52:42,  1.62it/s]
 61%|██████▏   | 8179/13317 [1:24:19<52:38,  1.63it/s]
                                                      
{'loss': 0.2058, 'grad_norm': 0.5176072716712952, 'learning_rate': 3.889403356491914e-06, 'epoch': 1.84}

 61%|██████▏   | 8179/13317 [1:24:19<52:38,  1.63it/s]
 61%|██████▏   | 8180/13317 [1:24:20<52:55,  1.62it/s]
                                                      
{'loss': 0.1932, 'grad_norm': 0.5210752487182617, 'learning_rate': 3.888125499766318e-06, 'epoch': 1.84}

 61%|██████▏   | 8180/13317 [1:24:20<52:55,  1.62it/s]
 61%|██████▏   | 8181/13317 [1:24:20<52:47,  1.62it/s]
                                                      
{'loss': 0.1877, 'grad_norm': 0.5087077021598816, 'learning_rate': 3.886847719438271e-06, 'epoch': 1.84}

 61%|██████▏   | 8181/13317 [1:24:20<52:47,  1.62it/s]
 61%|██████▏   | 8182/13317 [1:24:21<52:42,  1.62it/s]
                                                      
{'loss': 0.1719, 'grad_norm': 0.4873683750629425, 'learning_rate': 3.8855700155955725e-06, 'epoch': 1.84}

 61%|██████▏   | 8182/13317 [1:24:21<52:42,  1.62it/s]
 61%|██████▏   | 8183/13317 [1:24:21<52:42,  1.62it/s]
                                                      
{'loss': 0.2343, 'grad_norm': 0.6180227398872375, 'learning_rate': 3.884292388326014e-06, 'epoch': 1.84}

 61%|██████▏   | 8183/13317 [1:24:22<52:42,  1.62it/s]
 61%|██████▏   | 8184/13317 [1:24:22<52:38,  1.63it/s]
                                                      
{'loss': 0.1979, 'grad_norm': 0.5426004528999329, 'learning_rate': 3.883014837717383e-06, 'epoch': 1.84}

 61%|██████▏   | 8184/13317 [1:24:22<52:38,  1.63it/s]
 61%|██████▏   | 8185/13317 [1:24:23<52:39,  1.62it/s]
                                                      
{'loss': 0.1645, 'grad_norm': 0.5023361444473267, 'learning_rate': 3.881737363857457e-06, 'epoch': 1.84}

 61%|██████▏   | 8185/13317 [1:24:23<52:39,  1.62it/s]
 61%|██████▏   | 8186/13317 [1:24:23<53:45,  1.59it/s]
                                                      
{'loss': 0.1336, 'grad_norm': 0.4317473769187927, 'learning_rate': 3.880459966834015e-06, 'epoch': 1.84}

 61%|██████▏   | 8186/13317 [1:24:23<53:45,  1.59it/s]
 61%|██████▏   | 8187/13317 [1:24:24<53:34,  1.60it/s]
                                                      
{'loss': 0.217, 'grad_norm': 0.6119077801704407, 'learning_rate': 3.87918264673483e-06, 'epoch': 1.84}

 61%|██████▏   | 8187/13317 [1:24:24<53:34,  1.60it/s]
 61%|██████▏   | 8188/13317 [1:24:24<53:14,  1.61it/s]
                                                      
{'loss': 0.1829, 'grad_norm': 0.50633305311203, 'learning_rate': 3.877905403647661e-06, 'epoch': 1.84}

 61%|██████▏   | 8188/13317 [1:24:25<53:14,  1.61it/s]
 61%|██████▏   | 8189/13317 [1:24:25<53:00,  1.61it/s]
                                                      
{'loss': 0.1704, 'grad_norm': 0.5231257677078247, 'learning_rate': 3.8766282376602725e-06, 'epoch': 1.84}

 61%|██████▏   | 8189/13317 [1:24:25<53:00,  1.61it/s]
 62%|██████▏   | 8190/13317 [1:24:26<52:55,  1.61it/s]
                                                      
{'loss': 0.184, 'grad_norm': 0.4861680865287781, 'learning_rate': 3.87535114886042e-06, 'epoch': 1.85}

 62%|██████▏   | 8190/13317 [1:24:26<52:55,  1.61it/s]
 62%|██████▏   | 8191/13317 [1:24:26<52:45,  1.62it/s]
                                                      
{'loss': 0.171, 'grad_norm': 0.44707930088043213, 'learning_rate': 3.874074137335851e-06, 'epoch': 1.85}

 62%|██████▏   | 8191/13317 [1:24:26<52:45,  1.62it/s]
 62%|██████▏   | 8192/13317 [1:24:27<52:38,  1.62it/s]
                                                      
{'loss': 0.2284, 'grad_norm': 0.7171827554702759, 'learning_rate': 3.87279720317431e-06, 'epoch': 1.85}

 62%|██████▏   | 8192/13317 [1:24:27<52:38,  1.62it/s]
 62%|██████▏   | 8193/13317 [1:24:28<52:39,  1.62it/s]
                                                      
{'loss': 0.2333, 'grad_norm': 0.6969907879829407, 'learning_rate': 3.8715203464635346e-06, 'epoch': 1.85}

 62%|██████▏   | 8193/13317 [1:24:28<52:39,  1.62it/s]
 62%|██████▏   | 8194/13317 [1:24:28<52:33,  1.62it/s]
                                                      
{'loss': 0.1955, 'grad_norm': 0.5433605313301086, 'learning_rate': 3.870243567291263e-06, 'epoch': 1.85}

 62%|██████▏   | 8194/13317 [1:24:28<52:33,  1.62it/s]
 62%|██████▏   | 8195/13317 [1:24:29<52:44,  1.62it/s]
                                                      
{'loss': 0.1958, 'grad_norm': 0.5271968245506287, 'learning_rate': 3.868966865745218e-06, 'epoch': 1.85}

 62%|██████▏   | 8195/13317 [1:24:29<52:44,  1.62it/s]
 62%|██████▏   | 8196/13317 [1:24:29<52:38,  1.62it/s]
                                                      
{'loss': 0.1854, 'grad_norm': 0.4892334043979645, 'learning_rate': 3.867690241913126e-06, 'epoch': 1.85}

 62%|██████▏   | 8196/13317 [1:24:30<52:38,  1.62it/s]
 62%|██████▏   | 8197/13317 [1:24:30<52:32,  1.62it/s]
                                                      
{'loss': 0.1541, 'grad_norm': 0.42244255542755127, 'learning_rate': 3.866413695882703e-06, 'epoch': 1.85}

 62%|██████▏   | 8197/13317 [1:24:30<52:32,  1.62it/s]
 62%|██████▏   | 8198/13317 [1:24:31<52:32,  1.62it/s]
                                                      
{'loss': 0.1322, 'grad_norm': 0.40954849123954773, 'learning_rate': 3.865137227741662e-06, 'epoch': 1.85}

 62%|██████▏   | 8198/13317 [1:24:31<52:32,  1.62it/s]
 62%|██████▏   | 8199/13317 [1:24:31<52:28,  1.63it/s]
                                                      
{'loss': 0.2037, 'grad_norm': 0.5647141337394714, 'learning_rate': 3.86386083757771e-06, 'epoch': 1.85}

 62%|██████▏   | 8199/13317 [1:24:31<52:28,  1.63it/s]
 62%|██████▏   | 8200/13317 [1:24:32<52:32,  1.62it/s]
                                                      
{'loss': 0.1796, 'grad_norm': 0.4890509843826294, 'learning_rate': 3.862584525478545e-06, 'epoch': 1.85}

 62%|██████▏   | 8200/13317 [1:24:32<52:32,  1.62it/s]
 62%|██████▏   | 8201/13317 [1:24:32<52:26,  1.63it/s]
                                                      
{'loss': 0.195, 'grad_norm': 0.5617706775665283, 'learning_rate': 3.861308291531872e-06, 'epoch': 1.85}

 62%|██████▏   | 8201/13317 [1:24:33<52:26,  1.63it/s]
 62%|██████▏   | 8202/13317 [1:24:33<52:23,  1.63it/s]
                                                      
{'loss': 0.2091, 'grad_norm': 0.6414452791213989, 'learning_rate': 3.8600321358253714e-06, 'epoch': 1.85}

 62%|██████▏   | 8202/13317 [1:24:33<52:23,  1.63it/s]
 62%|██████▏   | 8203/13317 [1:24:34<52:24,  1.63it/s]
                                                      
{'loss': 0.2053, 'grad_norm': 0.5722959041595459, 'learning_rate': 3.858756058446734e-06, 'epoch': 1.85}

 62%|██████▏   | 8203/13317 [1:24:34<52:24,  1.63it/s]
 62%|██████▏   | 8204/13317 [1:24:34<52:21,  1.63it/s]
                                                      
{'loss': 0.2188, 'grad_norm': 0.5788330435752869, 'learning_rate': 3.857480059483641e-06, 'epoch': 1.85}

 62%|██████▏   | 8204/13317 [1:24:34<52:21,  1.63it/s]
 62%|██████▏   | 8205/13317 [1:24:35<52:26,  1.62it/s]
                                                      
{'loss': 0.2239, 'grad_norm': 0.6256008744239807, 'learning_rate': 3.856204139023764e-06, 'epoch': 1.85}

 62%|██████▏   | 8205/13317 [1:24:35<52:26,  1.62it/s]
 62%|██████▏   | 8206/13317 [1:24:36<52:27,  1.62it/s]
                                                      
{'loss': 0.2488, 'grad_norm': 0.6403433084487915, 'learning_rate': 3.854928297154771e-06, 'epoch': 1.85}

 62%|██████▏   | 8206/13317 [1:24:36<52:27,  1.62it/s]
 62%|██████▏   | 8207/13317 [1:24:36<52:24,  1.62it/s]
                                                      
{'loss': 0.2375, 'grad_norm': 0.5486268997192383, 'learning_rate': 3.853652533964332e-06, 'epoch': 1.85}

 62%|██████▏   | 8207/13317 [1:24:36<52:24,  1.62it/s]
 62%|██████▏   | 8208/13317 [1:24:37<52:27,  1.62it/s]
                                                      
{'loss': 0.1745, 'grad_norm': 0.5260441303253174, 'learning_rate': 3.8523768495400995e-06, 'epoch': 1.85}

 62%|██████▏   | 8208/13317 [1:24:37<52:27,  1.62it/s]
 62%|██████▏   | 8209/13317 [1:24:37<52:23,  1.62it/s]
                                                      
{'loss': 0.1805, 'grad_norm': 0.4867093563079834, 'learning_rate': 3.851101243969729e-06, 'epoch': 1.85}

 62%|██████▏   | 8209/13317 [1:24:38<52:23,  1.62it/s]
 62%|██████▏   | 8210/13317 [1:24:38<52:23,  1.62it/s]
                                                      
{'loss': 0.2164, 'grad_norm': 0.654721200466156, 'learning_rate': 3.849825717340868e-06, 'epoch': 1.85}

 62%|██████▏   | 8210/13317 [1:24:38<52:23,  1.62it/s]
 62%|██████▏   | 8211/13317 [1:24:39<52:21,  1.63it/s]
                                                      
{'loss': 0.2133, 'grad_norm': 0.6090071201324463, 'learning_rate': 3.8485502697411596e-06, 'epoch': 1.85}

 62%|██████▏   | 8211/13317 [1:24:39<52:21,  1.63it/s]
 62%|██████▏   | 8212/13317 [1:24:39<52:21,  1.63it/s]
                                                      
{'loss': 0.2103, 'grad_norm': 0.6288977861404419, 'learning_rate': 3.847274901258237e-06, 'epoch': 1.85}

 62%|██████▏   | 8212/13317 [1:24:39<52:21,  1.63it/s]
 62%|██████▏   | 8213/13317 [1:24:40<52:21,  1.62it/s]
                                                      
{'loss': 0.2377, 'grad_norm': 0.5778616666793823, 'learning_rate': 3.845999611979735e-06, 'epoch': 1.85}

 62%|██████▏   | 8213/13317 [1:24:40<52:21,  1.62it/s]
 62%|██████▏   | 8214/13317 [1:24:40<52:21,  1.62it/s]
                                                      
{'loss': 0.1701, 'grad_norm': 0.44782623648643494, 'learning_rate': 3.8447244019932805e-06, 'epoch': 1.85}

 62%|██████▏   | 8214/13317 [1:24:41<52:21,  1.62it/s]
 62%|██████▏   | 8215/13317 [1:24:41<52:20,  1.62it/s]
                                                      
{'loss': 0.2297, 'grad_norm': 0.5658291578292847, 'learning_rate': 3.8434492713864885e-06, 'epoch': 1.85}

 62%|██████▏   | 8215/13317 [1:24:41<52:20,  1.62it/s]
 62%|██████▏   | 8216/13317 [1:24:42<52:16,  1.63it/s]
                                                      
{'loss': 0.1786, 'grad_norm': 0.5482926368713379, 'learning_rate': 3.842174220246978e-06, 'epoch': 1.85}

 62%|██████▏   | 8216/13317 [1:24:42<52:16,  1.63it/s]
 62%|██████▏   | 8217/13317 [1:24:42<52:13,  1.63it/s]
                                                      
{'loss': 0.148, 'grad_norm': 0.43863871693611145, 'learning_rate': 3.840899248662358e-06, 'epoch': 1.85}

 62%|██████▏   | 8217/13317 [1:24:42<52:13,  1.63it/s]
 62%|██████▏   | 8218/13317 [1:24:43<52:15,  1.63it/s]
                                                      
{'loss': 0.1951, 'grad_norm': 0.4966515600681305, 'learning_rate': 3.839624356720235e-06, 'epoch': 1.85}

 62%|██████▏   | 8218/13317 [1:24:43<52:15,  1.63it/s]
 62%|██████▏   | 8219/13317 [1:24:44<52:13,  1.63it/s]
                                                      
{'loss': 0.1886, 'grad_norm': 0.5046193599700928, 'learning_rate': 3.838349544508202e-06, 'epoch': 1.85}

 62%|██████▏   | 8219/13317 [1:24:44<52:13,  1.63it/s]
 62%|██████▏   | 8220/13317 [1:24:44<52:12,  1.63it/s]
                                                      
{'loss': 0.1908, 'grad_norm': 0.4862731099128723, 'learning_rate': 3.837074812113855e-06, 'epoch': 1.85}

 62%|██████▏   | 8220/13317 [1:24:44<52:12,  1.63it/s]
 62%|██████▏   | 8221/13317 [1:24:45<52:11,  1.63it/s]
                                                      
{'loss': 0.2368, 'grad_norm': 0.5807379484176636, 'learning_rate': 3.835800159624784e-06, 'epoch': 1.85}

 62%|██████▏   | 8221/13317 [1:24:45<52:11,  1.63it/s]
 62%|██████▏   | 8222/13317 [1:24:45<52:10,  1.63it/s]
                                                      
{'loss': 0.232, 'grad_norm': 0.6135410070419312, 'learning_rate': 3.834525587128567e-06, 'epoch': 1.85}

 62%|██████▏   | 8222/13317 [1:24:46<52:10,  1.63it/s]
 62%|██████▏   | 8223/13317 [1:24:46<52:07,  1.63it/s]
                                                      
{'loss': 0.2058, 'grad_norm': 0.583220362663269, 'learning_rate': 3.8332510947127825e-06, 'epoch': 1.85}

 62%|██████▏   | 8223/13317 [1:24:46<52:07,  1.63it/s]
 62%|██████▏   | 8224/13317 [1:24:47<52:08,  1.63it/s]
                                                      
{'loss': 0.2072, 'grad_norm': 0.5298234820365906, 'learning_rate': 3.831976682465002e-06, 'epoch': 1.85}

 62%|██████▏   | 8224/13317 [1:24:47<52:08,  1.63it/s]
 62%|██████▏   | 8225/13317 [1:24:47<52:11,  1.63it/s]
                                                      
{'loss': 0.1558, 'grad_norm': 0.4416065812110901, 'learning_rate': 3.830702350472791e-06, 'epoch': 1.85}

 62%|██████▏   | 8225/13317 [1:24:47<52:11,  1.63it/s]
 62%|██████▏   | 8226/13317 [1:24:48<52:09,  1.63it/s]
                                                      
{'loss': 0.1623, 'grad_norm': 0.46428343653678894, 'learning_rate': 3.829428098823708e-06, 'epoch': 1.85}

 62%|██████▏   | 8226/13317 [1:24:48<52:09,  1.63it/s]
 62%|██████▏   | 8227/13317 [1:24:48<52:10,  1.63it/s]
                                                      
{'loss': 0.202, 'grad_norm': 0.5152066946029663, 'learning_rate': 3.828153927605309e-06, 'epoch': 1.85}

 62%|██████▏   | 8227/13317 [1:24:49<52:10,  1.63it/s]
 62%|██████▏   | 8228/13317 [1:24:49<52:07,  1.63it/s]
                                                      
{'loss': 0.2094, 'grad_norm': 0.5232322216033936, 'learning_rate': 3.826879836905144e-06, 'epoch': 1.85}

 62%|██████▏   | 8228/13317 [1:24:49<52:07,  1.63it/s]
 62%|██████▏   | 8229/13317 [1:24:50<52:03,  1.63it/s]
                                                      
{'loss': 0.1784, 'grad_norm': 0.5137650370597839, 'learning_rate': 3.8256058268107545e-06, 'epoch': 1.85}

 62%|██████▏   | 8229/13317 [1:24:50<52:03,  1.63it/s]
 62%|██████▏   | 8230/13317 [1:24:50<52:08,  1.63it/s]
                                                      
{'loss': 0.2253, 'grad_norm': 0.5205638408660889, 'learning_rate': 3.824331897409679e-06, 'epoch': 1.85}

 62%|██████▏   | 8230/13317 [1:24:50<52:08,  1.63it/s]
 62%|██████▏   | 8231/13317 [1:24:51<52:04,  1.63it/s]
                                                      
{'loss': 0.1941, 'grad_norm': 0.5489000678062439, 'learning_rate': 3.823058048789452e-06, 'epoch': 1.85}

 62%|██████▏   | 8231/13317 [1:24:51<52:04,  1.63it/s]
 62%|██████▏   | 8232/13317 [1:24:52<52:02,  1.63it/s]
                                                      
{'loss': 0.2262, 'grad_norm': 0.5399333238601685, 'learning_rate': 3.821784281037598e-06, 'epoch': 1.85}

 62%|██████▏   | 8232/13317 [1:24:52<52:02,  1.63it/s]
 62%|██████▏   | 8233/13317 [1:24:52<52:02,  1.63it/s]
                                                      
{'loss': 0.1958, 'grad_norm': 0.47434714436531067, 'learning_rate': 3.8205105942416375e-06, 'epoch': 1.85}

 62%|██████▏   | 8233/13317 [1:24:52<52:02,  1.63it/s]
 62%|██████▏   | 8234/13317 [1:24:53<52:03,  1.63it/s]
                                                      
{'loss': 0.1709, 'grad_norm': 0.4865753948688507, 'learning_rate': 3.819236988489089e-06, 'epoch': 1.85}

 62%|██████▏   | 8234/13317 [1:24:53<52:03,  1.63it/s]
 62%|██████▏   | 8235/13317 [1:24:53<52:03,  1.63it/s]
                                                      
{'loss': 0.2001, 'grad_norm': 0.5342336297035217, 'learning_rate': 3.817963463867462e-06, 'epoch': 1.86}

 62%|██████▏   | 8235/13317 [1:24:54<52:03,  1.63it/s]
 62%|██████▏   | 8236/13317 [1:24:54<52:01,  1.63it/s]
                                                      
{'loss': 0.2445, 'grad_norm': 0.5533227920532227, 'learning_rate': 3.81669002046426e-06, 'epoch': 1.86}

 62%|██████▏   | 8236/13317 [1:24:54<52:01,  1.63it/s]
 62%|██████▏   | 8237/13317 [1:24:55<51:59,  1.63it/s]
                                                      
{'loss': 0.2421, 'grad_norm': 0.5225448608398438, 'learning_rate': 3.815416658366983e-06, 'epoch': 1.86}

 62%|██████▏   | 8237/13317 [1:24:55<51:59,  1.63it/s]
 62%|██████▏   | 8238/13317 [1:24:55<52:00,  1.63it/s]
                                                      
{'loss': 0.2046, 'grad_norm': 0.6212422847747803, 'learning_rate': 3.814143377663125e-06, 'epoch': 1.86}

 62%|██████▏   | 8238/13317 [1:24:55<52:00,  1.63it/s]
 62%|██████▏   | 8239/13317 [1:24:56<51:56,  1.63it/s]
                                                      
{'loss': 0.192, 'grad_norm': 0.5351972579956055, 'learning_rate': 3.8128701784401717e-06, 'epoch': 1.86}

 62%|██████▏   | 8239/13317 [1:24:56<51:56,  1.63it/s]
 62%|██████▏   | 8240/13317 [1:24:56<52:02,  1.63it/s]
                                                      
{'loss': 0.2059, 'grad_norm': 0.4802989959716797, 'learning_rate': 3.8115970607856057e-06, 'epoch': 1.86}

 62%|██████▏   | 8240/13317 [1:24:57<52:02,  1.63it/s]
 62%|██████▏   | 8241/13317 [1:24:57<51:58,  1.63it/s]
                                                      
{'loss': 0.1658, 'grad_norm': 0.4528028964996338, 'learning_rate': 3.8103240247869077e-06, 'epoch': 1.86}

 62%|██████▏   | 8241/13317 [1:24:57<51:58,  1.63it/s]
 62%|██████▏   | 8242/13317 [1:24:58<51:57,  1.63it/s]
                                                      
{'loss': 0.1906, 'grad_norm': 0.47356992959976196, 'learning_rate': 3.809051070531543e-06, 'epoch': 1.86}

 62%|██████▏   | 8242/13317 [1:24:58<51:57,  1.63it/s]
 62%|██████▏   | 8243/13317 [1:24:58<51:57,  1.63it/s]
                                                      
{'loss': 0.1571, 'grad_norm': 0.45293110609054565, 'learning_rate': 3.8077781981069804e-06, 'epoch': 1.86}

 62%|██████▏   | 8243/13317 [1:24:58<51:57,  1.63it/s]
 62%|██████▏   | 8244/13317 [1:24:59<51:54,  1.63it/s]
                                                      
{'loss': 0.1753, 'grad_norm': 0.45940443873405457, 'learning_rate': 3.806505407600679e-06, 'epoch': 1.86}

 62%|██████▏   | 8244/13317 [1:24:59<51:54,  1.63it/s]
 62%|██████▏   | 8245/13317 [1:25:00<51:53,  1.63it/s]
                                                      
{'loss': 0.1856, 'grad_norm': 0.4643320143222809, 'learning_rate': 3.8052326991000945e-06, 'epoch': 1.86}

 62%|██████▏   | 8245/13317 [1:25:00<51:53,  1.63it/s]
 62%|██████▏   | 8246/13317 [1:25:00<51:52,  1.63it/s]
                                                      
{'loss': 0.1722, 'grad_norm': 0.48007479310035706, 'learning_rate': 3.803960072692672e-06, 'epoch': 1.86}

 62%|██████▏   | 8246/13317 [1:25:00<51:52,  1.63it/s]
 62%|██████▏   | 8247/13317 [1:25:01<51:51,  1.63it/s]
                                                      
{'loss': 0.1677, 'grad_norm': 0.4767109751701355, 'learning_rate': 3.802687528465856e-06, 'epoch': 1.86}

 62%|██████▏   | 8247/13317 [1:25:01<51:51,  1.63it/s]
 62%|██████▏   | 8248/13317 [1:25:01<51:51,  1.63it/s]
                                                      
{'loss': 0.2071, 'grad_norm': 0.6287320256233215, 'learning_rate': 3.8014150665070877e-06, 'epoch': 1.86}

 62%|██████▏   | 8248/13317 [1:25:02<51:51,  1.63it/s]
 62%|██████▏   | 8249/13317 [1:25:02<51:50,  1.63it/s]
                                                      
{'loss': 0.1743, 'grad_norm': 0.5363152623176575, 'learning_rate': 3.8001426869037912e-06, 'epoch': 1.86}

 62%|██████▏   | 8249/13317 [1:25:02<51:50,  1.63it/s]
 62%|██████▏   | 8250/13317 [1:25:03<52:06,  1.62it/s]
                                                      
{'loss': 0.2076, 'grad_norm': 0.5510851144790649, 'learning_rate': 3.7988703897433974e-06, 'epoch': 1.86}

 62%|██████▏   | 8250/13317 [1:25:03<52:06,  1.62it/s]
 62%|██████▏   | 8251/13317 [1:25:03<52:03,  1.62it/s]
                                                      
{'loss': 0.2115, 'grad_norm': 0.5216919779777527, 'learning_rate': 3.797598175113327e-06, 'epoch': 1.86}

 62%|██████▏   | 8251/13317 [1:25:03<52:03,  1.62it/s]
 62%|██████▏   | 8252/13317 [1:25:04<51:58,  1.62it/s]
                                                      
{'loss': 0.2134, 'grad_norm': 0.5690755248069763, 'learning_rate': 3.796326043100993e-06, 'epoch': 1.86}

 62%|██████▏   | 8252/13317 [1:25:04<51:58,  1.62it/s]
 62%|██████▏   | 8253/13317 [1:25:04<52:00,  1.62it/s]
                                                      
{'loss': 0.219, 'grad_norm': 0.5512404441833496, 'learning_rate': 3.7950539937938035e-06, 'epoch': 1.86}

 62%|██████▏   | 8253/13317 [1:25:05<52:00,  1.62it/s]
 62%|██████▏   | 8254/13317 [1:25:05<51:55,  1.62it/s]
                                                      
{'loss': 0.1995, 'grad_norm': 0.5196918249130249, 'learning_rate': 3.7937820272791614e-06, 'epoch': 1.86}

 62%|██████▏   | 8254/13317 [1:25:05<51:55,  1.62it/s]
 62%|██████▏   | 8255/13317 [1:25:06<51:54,  1.63it/s]
                                                      
{'loss': 0.2183, 'grad_norm': 0.5647530555725098, 'learning_rate': 3.792510143644469e-06, 'epoch': 1.86}

 62%|██████▏   | 8255/13317 [1:25:06<51:54,  1.63it/s]
 62%|██████▏   | 8256/13317 [1:25:06<51:50,  1.63it/s]
                                                      
{'loss': 0.2435, 'grad_norm': 0.5008637309074402, 'learning_rate': 3.791238342977113e-06, 'epoch': 1.86}

 62%|██████▏   | 8256/13317 [1:25:06<51:50,  1.63it/s]
 62%|██████▏   | 8257/13317 [1:25:07<51:47,  1.63it/s]
                                                      
{'loss': 0.1667, 'grad_norm': 0.45051148533821106, 'learning_rate': 3.7899666253644813e-06, 'epoch': 1.86}

 62%|██████▏   | 8257/13317 [1:25:07<51:47,  1.63it/s]
 62%|██████▏   | 8258/13317 [1:25:08<51:51,  1.63it/s]
                                                      
{'loss': 0.1883, 'grad_norm': 0.6106361150741577, 'learning_rate': 3.7886949908939557e-06, 'epoch': 1.86}

 62%|██████▏   | 8258/13317 [1:25:08<51:51,  1.63it/s]
 62%|██████▏   | 8259/13317 [1:25:08<51:48,  1.63it/s]
                                                      
{'loss': 0.2651, 'grad_norm': 0.6013010740280151, 'learning_rate': 3.7874234396529086e-06, 'epoch': 1.86}

 62%|██████▏   | 8259/13317 [1:25:08<51:48,  1.63it/s]
 62%|██████▏   | 8260/13317 [1:25:09<51:49,  1.63it/s]
                                                      
{'loss': 0.1708, 'grad_norm': 0.4684356451034546, 'learning_rate': 3.7861519717287094e-06, 'epoch': 1.86}

 62%|██████▏   | 8260/13317 [1:25:09<51:49,  1.63it/s]
 62%|██████▏   | 8261/13317 [1:25:09<51:47,  1.63it/s]
                                                      
{'loss': 0.2061, 'grad_norm': 0.5561591982841492, 'learning_rate': 3.784880587208723e-06, 'epoch': 1.86}

 62%|██████▏   | 8261/13317 [1:25:10<51:47,  1.63it/s]
 62%|██████▏   | 8262/13317 [1:25:10<51:44,  1.63it/s]
                                                      
{'loss': 0.2173, 'grad_norm': 0.582297146320343, 'learning_rate': 3.7836092861803076e-06, 'epoch': 1.86}

 62%|██████▏   | 8262/13317 [1:25:10<51:44,  1.63it/s]
 62%|██████▏   | 8263/13317 [1:25:11<51:44,  1.63it/s]
                                                      
{'loss': 0.2802, 'grad_norm': 0.7074705958366394, 'learning_rate': 3.782338068730812e-06, 'epoch': 1.86}

 62%|██████▏   | 8263/13317 [1:25:11<51:44,  1.63it/s]
 62%|██████▏   | 8264/13317 [1:25:11<51:42,  1.63it/s]
                                                      
{'loss': 0.1705, 'grad_norm': 0.4481219947338104, 'learning_rate': 3.7810669349475845e-06, 'epoch': 1.86}

 62%|██████▏   | 8264/13317 [1:25:11<51:42,  1.63it/s]
 62%|██████▏   | 8265/13317 [1:25:12<51:46,  1.63it/s]
                                                      
{'loss': 0.1575, 'grad_norm': 0.43135276436805725, 'learning_rate': 3.779795884917965e-06, 'epoch': 1.86}

 62%|██████▏   | 8265/13317 [1:25:12<51:46,  1.63it/s]
 62%|██████▏   | 8266/13317 [1:25:12<51:44,  1.63it/s]
                                                      
{'loss': 0.2109, 'grad_norm': 0.5358863472938538, 'learning_rate': 3.778524918729287e-06, 'epoch': 1.86}

 62%|██████▏   | 8266/13317 [1:25:13<51:44,  1.63it/s]
 62%|██████▏   | 8267/13317 [1:25:13<51:42,  1.63it/s]
                                                      
{'loss': 0.2323, 'grad_norm': 0.6209167838096619, 'learning_rate': 3.7772540364688805e-06, 'epoch': 1.86}

 62%|██████▏   | 8267/13317 [1:25:13<51:42,  1.63it/s]
 62%|██████▏   | 8268/13317 [1:25:14<51:58,  1.62it/s]
                                                      
{'loss': 0.1834, 'grad_norm': 0.4962010681629181, 'learning_rate': 3.77598323822407e-06, 'epoch': 1.86}

 62%|██████▏   | 8268/13317 [1:25:14<51:58,  1.62it/s]
 62%|██████▏   | 8269/13317 [1:25:14<51:52,  1.62it/s]
                                                      
{'loss': 0.1816, 'grad_norm': 0.5590936541557312, 'learning_rate': 3.7747125240821687e-06, 'epoch': 1.86}

 62%|██████▏   | 8269/13317 [1:25:14<51:52,  1.62it/s]
 62%|██████▏   | 8270/13317 [1:25:15<51:53,  1.62it/s]
                                                      
{'loss': 0.2294, 'grad_norm': 0.5805097222328186, 'learning_rate': 3.7734418941304912e-06, 'epoch': 1.86}

 62%|██████▏   | 8270/13317 [1:25:15<51:53,  1.62it/s]
 62%|██████▏   | 8271/13317 [1:25:16<51:48,  1.62it/s]
                                                      
{'loss': 0.2016, 'grad_norm': 0.5263344645500183, 'learning_rate': 3.7721713484563427e-06, 'epoch': 1.86}

 62%|██████▏   | 8271/13317 [1:25:16<51:48,  1.62it/s]
 62%|██████▏   | 8272/13317 [1:25:16<51:44,  1.62it/s]
                                                      
{'loss': 0.2169, 'grad_norm': 0.6544471979141235, 'learning_rate': 3.7709008871470243e-06, 'epoch': 1.86}

 62%|██████▏   | 8272/13317 [1:25:16<51:44,  1.62it/s]
 62%|██████▏   | 8273/13317 [1:25:17<51:44,  1.62it/s]
                                                      
{'loss': 0.1986, 'grad_norm': 0.5114243030548096, 'learning_rate': 3.769630510289828e-06, 'epoch': 1.86}

 62%|██████▏   | 8273/13317 [1:25:17<51:44,  1.62it/s]
 62%|██████▏   | 8274/13317 [1:25:17<51:43,  1.63it/s]
                                                      
{'loss': 0.1896, 'grad_norm': 0.5134689807891846, 'learning_rate': 3.7683602179720412e-06, 'epoch': 1.86}

 62%|██████▏   | 8274/13317 [1:25:18<51:43,  1.63it/s]
 62%|██████▏   | 8275/13317 [1:25:18<51:44,  1.62it/s]
                                                      
{'loss': 0.1574, 'grad_norm': 0.4516183137893677, 'learning_rate': 3.7670900102809522e-06, 'epoch': 1.86}

 62%|██████▏   | 8275/13317 [1:25:18<51:44,  1.62it/s]
 62%|██████▏   | 8276/13317 [1:25:19<51:40,  1.63it/s]
                                                      
{'loss': 0.1882, 'grad_norm': 0.4949108362197876, 'learning_rate': 3.7658198873038306e-06, 'epoch': 1.86}

 62%|██████▏   | 8276/13317 [1:25:19<51:40,  1.63it/s]
 62%|██████▏   | 8277/13317 [1:25:19<51:36,  1.63it/s]
                                                      
{'loss': 0.1755, 'grad_norm': 0.4993392825126648, 'learning_rate': 3.7645498491279507e-06, 'epoch': 1.86}

 62%|██████▏   | 8277/13317 [1:25:19<51:36,  1.63it/s]
 62%|██████▏   | 8278/13317 [1:25:20<51:37,  1.63it/s]
                                                      
{'loss': 0.1602, 'grad_norm': 0.42069754004478455, 'learning_rate': 3.763279895840578e-06, 'epoch': 1.86}

 62%|██████▏   | 8278/13317 [1:25:20<51:37,  1.63it/s]
 62%|██████▏   | 8279/13317 [1:25:20<51:36,  1.63it/s]
                                                      
{'loss': 0.2197, 'grad_norm': 0.5712767243385315, 'learning_rate': 3.762010027528972e-06, 'epoch': 1.87}

 62%|██████▏   | 8279/13317 [1:25:21<51:36,  1.63it/s]
 62%|██████▏   | 8280/13317 [1:25:21<51:37,  1.63it/s]
                                                      
{'loss': 0.2311, 'grad_norm': 0.5651736259460449, 'learning_rate': 3.7607402442803837e-06, 'epoch': 1.87}

 62%|██████▏   | 8280/13317 [1:25:21<51:37,  1.63it/s]
 62%|██████▏   | 8281/13317 [1:25:22<51:33,  1.63it/s]
                                                      
{'loss': 0.1518, 'grad_norm': 0.4421301782131195, 'learning_rate': 3.759470546182061e-06, 'epoch': 1.87}

 62%|██████▏   | 8281/13317 [1:25:22<51:33,  1.63it/s]
 62%|██████▏   | 8282/13317 [1:25:22<51:32,  1.63it/s]
                                                      
{'loss': 0.2024, 'grad_norm': 0.5822128057479858, 'learning_rate': 3.7582009333212497e-06, 'epoch': 1.87}

 62%|██████▏   | 8282/13317 [1:25:22<51:32,  1.63it/s]
 62%|██████▏   | 8283/13317 [1:25:23<51:33,  1.63it/s]
                                                      
{'loss': 0.1958, 'grad_norm': 0.5154951214790344, 'learning_rate': 3.756931405785181e-06, 'epoch': 1.87}

 62%|██████▏   | 8283/13317 [1:25:23<51:33,  1.63it/s]
 62%|██████▏   | 8284/13317 [1:25:24<51:31,  1.63it/s]
                                                      
{'loss': 0.1944, 'grad_norm': 0.5267917513847351, 'learning_rate': 3.755661963661087e-06, 'epoch': 1.87}

 62%|██████▏   | 8284/13317 [1:25:24<51:31,  1.63it/s]
 62%|██████▏   | 8285/13317 [1:25:24<51:34,  1.63it/s]
                                                      
{'loss': 0.1554, 'grad_norm': 0.43385639786720276, 'learning_rate': 3.754392607036191e-06, 'epoch': 1.87}

 62%|██████▏   | 8285/13317 [1:25:24<51:34,  1.63it/s]
 62%|██████▏   | 8286/13317 [1:25:25<51:31,  1.63it/s]
                                                      
{'loss': 0.1772, 'grad_norm': 0.44160184264183044, 'learning_rate': 3.7531233359977122e-06, 'epoch': 1.87}

 62%|██████▏   | 8286/13317 [1:25:25<51:31,  1.63it/s]
 62%|██████▏   | 8287/13317 [1:25:25<51:30,  1.63it/s]
                                                      
{'loss': 0.2049, 'grad_norm': 0.49348336458206177, 'learning_rate': 3.751854150632861e-06, 'epoch': 1.87}

 62%|██████▏   | 8287/13317 [1:25:26<51:30,  1.63it/s]
 62%|██████▏   | 8288/13317 [1:25:26<51:31,  1.63it/s]
                                                      
{'loss': 0.2574, 'grad_norm': 0.5736554265022278, 'learning_rate': 3.7505850510288455e-06, 'epoch': 1.87}

 62%|██████▏   | 8288/13317 [1:25:26<51:31,  1.63it/s]
 62%|██████▏   | 8289/13317 [1:25:27<51:30,  1.63it/s]
                                                      
{'loss': 0.24, 'grad_norm': 0.5973424911499023, 'learning_rate': 3.7493160372728677e-06, 'epoch': 1.87}

 62%|██████▏   | 8289/13317 [1:25:27<51:30,  1.63it/s]
 62%|██████▏   | 8290/13317 [1:25:27<51:31,  1.63it/s]
                                                      
{'loss': 0.2048, 'grad_norm': 0.5637016892433167, 'learning_rate': 3.74804710945212e-06, 'epoch': 1.87}

 62%|██████▏   | 8290/13317 [1:25:27<51:31,  1.63it/s]
 62%|██████▏   | 8291/13317 [1:25:28<51:27,  1.63it/s]
                                                      
{'loss': 0.1729, 'grad_norm': 0.4264207184314728, 'learning_rate': 3.7467782676537907e-06, 'epoch': 1.87}

 62%|██████▏   | 8291/13317 [1:25:28<51:27,  1.63it/s]
 62%|██████▏   | 8292/13317 [1:25:28<51:25,  1.63it/s]
                                                      
{'loss': 0.2185, 'grad_norm': 0.5886661410331726, 'learning_rate': 3.745509511965065e-06, 'epoch': 1.87}

 62%|██████▏   | 8292/13317 [1:25:29<51:25,  1.63it/s]
 62%|██████▏   | 8293/13317 [1:25:29<51:28,  1.63it/s]
                                                      
{'loss': 0.1998, 'grad_norm': 0.5530877709388733, 'learning_rate': 3.744240842473118e-06, 'epoch': 1.87}

 62%|██████▏   | 8293/13317 [1:25:29<51:28,  1.63it/s]
 62%|██████▏   | 8294/13317 [1:25:30<51:27,  1.63it/s]
                                                      
{'loss': 0.2012, 'grad_norm': 0.5563821196556091, 'learning_rate': 3.7429722592651195e-06, 'epoch': 1.87}

 62%|██████▏   | 8294/13317 [1:25:30<51:27,  1.63it/s]
 62%|██████▏   | 8295/13317 [1:25:30<51:29,  1.63it/s]
                                                      
{'loss': 0.1881, 'grad_norm': 0.49328675866127014, 'learning_rate': 3.7417037624282382e-06, 'epoch': 1.87}

 62%|██████▏   | 8295/13317 [1:25:30<51:29,  1.63it/s]
 62%|██████▏   | 8296/13317 [1:25:31<51:26,  1.63it/s]
                                                      
{'loss': 0.2171, 'grad_norm': 0.6316263675689697, 'learning_rate': 3.740435352049633e-06, 'epoch': 1.87}

 62%|██████▏   | 8296/13317 [1:25:31<51:26,  1.63it/s]
 62%|██████▏   | 8297/13317 [1:25:32<51:24,  1.63it/s]
                                                      
{'loss': 0.1554, 'grad_norm': 0.4804697334766388, 'learning_rate': 3.7391670282164533e-06, 'epoch': 1.87}

 62%|██████▏   | 8297/13317 [1:25:32<51:24,  1.63it/s]
 62%|██████▏   | 8298/13317 [1:25:32<51:28,  1.63it/s]
                                                      
{'loss': 0.1843, 'grad_norm': 0.4852513372898102, 'learning_rate': 3.7378987910158495e-06, 'epoch': 1.87}

 62%|██████▏   | 8298/13317 [1:25:32<51:28,  1.63it/s]
 62%|██████▏   | 8299/13317 [1:25:33<51:26,  1.63it/s]
                                                      
{'loss': 0.2217, 'grad_norm': 0.5368065237998962, 'learning_rate': 3.736630640534963e-06, 'epoch': 1.87}

 62%|██████▏   | 8299/13317 [1:25:33<51:26,  1.63it/s]
 62%|██████▏   | 8300/13317 [1:25:33<51:28,  1.62it/s]
                                                      
{'loss': 0.1755, 'grad_norm': 0.4678193926811218, 'learning_rate': 3.735362576860927e-06, 'epoch': 1.87}

 62%|██████▏   | 8300/13317 [1:25:33<51:28,  1.62it/s]
 62%|██████▏   | 8301/13317 [1:25:34<51:28,  1.62it/s]
                                                      
{'loss': 0.2205, 'grad_norm': 0.5593596696853638, 'learning_rate': 3.7340946000808707e-06, 'epoch': 1.87}

 62%|██████▏   | 8301/13317 [1:25:34<51:28,  1.62it/s]
 62%|██████▏   | 8302/13317 [1:25:35<51:24,  1.63it/s]
                                                      
{'loss': 0.176, 'grad_norm': 0.49715763330459595, 'learning_rate': 3.732826710281923e-06, 'epoch': 1.87}

 62%|██████▏   | 8302/13317 [1:25:35<51:24,  1.63it/s]
 62%|██████▏   | 8303/13317 [1:25:35<51:24,  1.63it/s]
                                                      
{'loss': 0.2228, 'grad_norm': 0.5397194027900696, 'learning_rate': 3.7315589075511937e-06, 'epoch': 1.87}

 62%|██████▏   | 8303/13317 [1:25:35<51:24,  1.63it/s]
 62%|██████▏   | 8304/13317 [1:25:36<51:23,  1.63it/s]
                                                      
{'loss': 0.2046, 'grad_norm': 0.5237137079238892, 'learning_rate': 3.730291191975799e-06, 'epoch': 1.87}

 62%|██████▏   | 8304/13317 [1:25:36<51:23,  1.63it/s]
 62%|██████▏   | 8305/13317 [1:25:36<51:26,  1.62it/s]
                                                      
{'loss': 0.1612, 'grad_norm': 0.4644707143306732, 'learning_rate': 3.729023563642844e-06, 'epoch': 1.87}

 62%|██████▏   | 8305/13317 [1:25:37<51:26,  1.62it/s]
 62%|██████▏   | 8306/13317 [1:25:37<51:22,  1.63it/s]
                                                      
{'loss': 0.2259, 'grad_norm': 0.5565671920776367, 'learning_rate': 3.7277560226394278e-06, 'epoch': 1.87}

 62%|██████▏   | 8306/13317 [1:25:37<51:22,  1.63it/s]
 62%|██████▏   | 8307/13317 [1:25:38<51:18,  1.63it/s]
                                                      
{'loss': 0.2183, 'grad_norm': 0.5474469661712646, 'learning_rate': 3.7264885690526416e-06, 'epoch': 1.87}

 62%|██████▏   | 8307/13317 [1:25:38<51:18,  1.63it/s]
 62%|██████▏   | 8308/13317 [1:25:38<51:16,  1.63it/s]
                                                      
{'loss': 0.1646, 'grad_norm': 0.4214138388633728, 'learning_rate': 3.7252212029695757e-06, 'epoch': 1.87}

 62%|██████▏   | 8308/13317 [1:25:38<51:16,  1.63it/s]
 62%|██████▏   | 8309/13317 [1:25:39<51:14,  1.63it/s]
                                                      
{'loss': 0.2499, 'grad_norm': 0.5646467804908752, 'learning_rate': 3.723953924477313e-06, 'epoch': 1.87}

 62%|██████▏   | 8309/13317 [1:25:39<51:14,  1.63it/s]
 62%|██████▏   | 8310/13317 [1:25:40<51:16,  1.63it/s]
                                                      
{'loss': 0.1543, 'grad_norm': 0.471123605966568, 'learning_rate': 3.7226867336629236e-06, 'epoch': 1.87}

 62%|██████▏   | 8310/13317 [1:25:40<51:16,  1.63it/s]
 62%|██████▏   | 8311/13317 [1:25:40<51:12,  1.63it/s]
                                                      
{'loss': 0.2334, 'grad_norm': 0.5386635661125183, 'learning_rate': 3.7214196306134813e-06, 'epoch': 1.87}

 62%|██████▏   | 8311/13317 [1:25:40<51:12,  1.63it/s]
 62%|██████▏   | 8312/13317 [1:25:41<51:10,  1.63it/s]
                                                      
{'loss': 0.1968, 'grad_norm': 0.5283646583557129, 'learning_rate': 3.7201526154160496e-06, 'epoch': 1.87}

 62%|██████▏   | 8312/13317 [1:25:41<51:10,  1.63it/s]
 62%|██████▏   | 8313/13317 [1:25:41<51:09,  1.63it/s]
                                                      
{'loss': 0.1674, 'grad_norm': 0.47510820627212524, 'learning_rate': 3.7188856881576822e-06, 'epoch': 1.87}

 62%|██████▏   | 8313/13317 [1:25:41<51:09,  1.63it/s]
 62%|██████▏   | 8314/13317 [1:25:42<51:07,  1.63it/s]
                                                      
{'loss': 0.2064, 'grad_norm': 0.5024534463882446, 'learning_rate': 3.7176188489254326e-06, 'epoch': 1.87}

 62%|██████▏   | 8314/13317 [1:25:42<51:07,  1.63it/s]
 62%|██████▏   | 8315/13317 [1:25:43<51:20,  1.62it/s]
                                                      
{'loss': 0.2214, 'grad_norm': 0.5489963889122009, 'learning_rate': 3.7163520978063464e-06, 'epoch': 1.87}

 62%|██████▏   | 8315/13317 [1:25:43<51:20,  1.62it/s]
 62%|██████▏   | 8316/13317 [1:25:43<51:15,  1.63it/s]
                                                      
{'loss': 0.1661, 'grad_norm': 0.4704338014125824, 'learning_rate': 3.7150854348874645e-06, 'epoch': 1.87}

 62%|██████▏   | 8316/13317 [1:25:43<51:15,  1.63it/s]
 62%|██████▏   | 8317/13317 [1:25:44<51:11,  1.63it/s]
                                                      
{'loss': 0.1844, 'grad_norm': 0.5825283527374268, 'learning_rate': 3.713818860255817e-06, 'epoch': 1.87}

 62%|██████▏   | 8317/13317 [1:25:44<51:11,  1.63it/s]
 62%|██████▏   | 8318/13317 [1:25:44<51:09,  1.63it/s]
                                                      
{'loss': 0.1708, 'grad_norm': 0.4667249321937561, 'learning_rate': 3.7125523739984314e-06, 'epoch': 1.87}

 62%|██████▏   | 8318/13317 [1:25:45<51:09,  1.63it/s]
 62%|██████▏   | 8319/13317 [1:25:45<51:07,  1.63it/s]
                                                      
{'loss': 0.22, 'grad_norm': 0.5500086545944214, 'learning_rate': 3.711285976202331e-06, 'epoch': 1.87}

 62%|██████▏   | 8319/13317 [1:25:45<51:07,  1.63it/s]
 62%|██████▏   | 8320/13317 [1:25:46<51:11,  1.63it/s]
                                                      
{'loss': 0.2086, 'grad_norm': 0.527306079864502, 'learning_rate': 3.7100196669545274e-06, 'epoch': 1.87}

 62%|██████▏   | 8320/13317 [1:25:46<51:11,  1.63it/s]
 62%|██████▏   | 8321/13317 [1:25:46<51:08,  1.63it/s]
                                                      
{'loss': 0.1345, 'grad_norm': 0.43297597765922546, 'learning_rate': 3.708753446342031e-06, 'epoch': 1.87}

 62%|██████▏   | 8321/13317 [1:25:46<51:08,  1.63it/s]
 62%|██████▏   | 8322/13317 [1:25:47<51:06,  1.63it/s]
                                                      
{'loss': 0.189, 'grad_norm': 0.4603928029537201, 'learning_rate': 3.707487314451845e-06, 'epoch': 1.87}

 62%|██████▏   | 8322/13317 [1:25:47<51:06,  1.63it/s]
 62%|██████▏   | 8323/13317 [1:25:47<51:09,  1.63it/s]
                                                      
{'loss': 0.1763, 'grad_norm': 0.45595821738243103, 'learning_rate': 3.7062212713709678e-06, 'epoch': 1.87}

 62%|██████▏   | 8323/13317 [1:25:48<51:09,  1.63it/s]
 63%|██████▎   | 8324/13317 [1:25:48<51:08,  1.63it/s]
                                                      
{'loss': 0.2182, 'grad_norm': 0.5210493803024292, 'learning_rate': 3.704955317186387e-06, 'epoch': 1.88}

 63%|██████▎   | 8324/13317 [1:25:48<51:08,  1.63it/s]
 63%|██████▎   | 8325/13317 [1:25:49<51:07,  1.63it/s]
                                                      
{'loss': 0.1991, 'grad_norm': 0.49455779790878296, 'learning_rate': 3.7036894519850876e-06, 'epoch': 1.88}

 63%|██████▎   | 8325/13317 [1:25:49<51:07,  1.63it/s]
 63%|██████▎   | 8326/13317 [1:25:49<51:04,  1.63it/s]
                                                      
{'loss': 0.1893, 'grad_norm': 0.5344967842102051, 'learning_rate': 3.7024236758540493e-06, 'epoch': 1.88}

 63%|██████▎   | 8326/13317 [1:25:49<51:04,  1.63it/s]
 63%|██████▎   | 8327/13317 [1:25:50<51:03,  1.63it/s]
                                                      
{'loss': 0.1734, 'grad_norm': 0.47828608751296997, 'learning_rate': 3.701157988880243e-06, 'epoch': 1.88}

 63%|██████▎   | 8327/13317 [1:25:50<51:03,  1.63it/s]
 63%|██████▎   | 8328/13317 [1:25:51<51:16,  1.62it/s]
                                                      
{'loss': 0.14, 'grad_norm': 0.4224208891391754, 'learning_rate': 3.6998923911506333e-06, 'epoch': 1.88}

 63%|██████▎   | 8328/13317 [1:25:51<51:16,  1.62it/s]
 63%|██████▎   | 8329/13317 [1:25:51<51:11,  1.62it/s]
                                                      
{'loss': 0.1746, 'grad_norm': 0.4652714133262634, 'learning_rate': 3.6986268827521843e-06, 'epoch': 1.88}

 63%|██████▎   | 8329/13317 [1:25:51<51:11,  1.62it/s]
 63%|██████▎   | 8330/13317 [1:25:52<51:07,  1.63it/s]
                                                      
{'loss': 0.163, 'grad_norm': 0.46825939416885376, 'learning_rate': 3.697361463771847e-06, 'epoch': 1.88}

 63%|██████▎   | 8330/13317 [1:25:52<51:07,  1.63it/s]
 63%|██████▎   | 8331/13317 [1:25:52<51:05,  1.63it/s]
                                                      
{'loss': 0.1767, 'grad_norm': 0.5509874224662781, 'learning_rate': 3.6960961342965693e-06, 'epoch': 1.88}

 63%|██████▎   | 8331/13317 [1:25:53<51:05,  1.63it/s]
 63%|██████▎   | 8332/13317 [1:25:53<51:01,  1.63it/s]
                                                      
{'loss': 0.1688, 'grad_norm': 0.48681551218032837, 'learning_rate': 3.694830894413294e-06, 'epoch': 1.88}

 63%|██████▎   | 8332/13317 [1:25:53<51:01,  1.63it/s]
 63%|██████▎   | 8333/13317 [1:25:54<51:03,  1.63it/s]
                                                      
{'loss': 0.2138, 'grad_norm': 0.5992109775543213, 'learning_rate': 3.693565744208956e-06, 'epoch': 1.88}

 63%|██████▎   | 8333/13317 [1:25:54<51:03,  1.63it/s]
 63%|██████▎   | 8334/13317 [1:25:54<51:01,  1.63it/s]
                                                      
{'loss': 0.2257, 'grad_norm': 0.5929598808288574, 'learning_rate': 3.692300683770482e-06, 'epoch': 1.88}

 63%|██████▎   | 8334/13317 [1:25:54<51:01,  1.63it/s]
 63%|██████▎   | 8335/13317 [1:25:55<51:00,  1.63it/s]
                                                      
{'loss': 0.1581, 'grad_norm': 0.4698563516139984, 'learning_rate': 3.6910357131847986e-06, 'epoch': 1.88}

 63%|██████▎   | 8335/13317 [1:25:55<51:00,  1.63it/s]
 63%|██████▎   | 8336/13317 [1:25:55<50:59,  1.63it/s]
                                                      
{'loss': 0.2003, 'grad_norm': 0.5508965253829956, 'learning_rate': 3.6897708325388217e-06, 'epoch': 1.88}

 63%|██████▎   | 8336/13317 [1:25:56<50:59,  1.63it/s]
 63%|██████▎   | 8337/13317 [1:25:56<51:00,  1.63it/s]
                                                      
{'loss': 0.1887, 'grad_norm': 0.4932631552219391, 'learning_rate': 3.6885060419194607e-06, 'epoch': 1.88}

 63%|██████▎   | 8337/13317 [1:25:56<51:00,  1.63it/s]
 63%|██████▎   | 8338/13317 [1:25:57<51:00,  1.63it/s]
                                                      
{'loss': 0.1692, 'grad_norm': 0.5637333393096924, 'learning_rate': 3.687241341413621e-06, 'epoch': 1.88}

 63%|██████▎   | 8338/13317 [1:25:57<51:00,  1.63it/s]
 63%|██████▎   | 8339/13317 [1:25:57<50:58,  1.63it/s]
                                                      
{'loss': 0.1855, 'grad_norm': 0.49047794938087463, 'learning_rate': 3.6859767311082e-06, 'epoch': 1.88}

 63%|██████▎   | 8339/13317 [1:25:57<50:58,  1.63it/s]
 63%|██████▎   | 8340/13317 [1:25:58<51:12,  1.62it/s]
                                                      
{'loss': 0.2923, 'grad_norm': 0.8125869035720825, 'learning_rate': 3.6847122110900934e-06, 'epoch': 1.88}

 63%|██████▎   | 8340/13317 [1:25:58<51:12,  1.62it/s]
 63%|██████▎   | 8341/13317 [1:25:59<51:04,  1.62it/s]
                                                      
{'loss': 0.1828, 'grad_norm': 0.48473215103149414, 'learning_rate': 3.6834477814461824e-06, 'epoch': 1.88}

 63%|██████▎   | 8341/13317 [1:25:59<51:04,  1.62it/s]
 63%|██████▎   | 8342/13317 [1:25:59<51:01,  1.63it/s]
                                                      
{'loss': 0.197, 'grad_norm': 0.504342257976532, 'learning_rate': 3.6821834422633484e-06, 'epoch': 1.88}

 63%|██████▎   | 8342/13317 [1:25:59<51:01,  1.63it/s]
 63%|██████▎   | 8343/13317 [1:26:00<50:59,  1.63it/s]
                                                      
{'loss': 0.2529, 'grad_norm': 0.5418204665184021, 'learning_rate': 3.6809191936284665e-06, 'epoch': 1.88}

 63%|██████▎   | 8343/13317 [1:26:00<50:59,  1.63it/s]
 63%|██████▎   | 8344/13317 [1:26:00<50:56,  1.63it/s]
                                                      
{'loss': 0.1775, 'grad_norm': 0.49715372920036316, 'learning_rate': 3.679655035628402e-06, 'epoch': 1.88}

 63%|██████▎   | 8344/13317 [1:26:01<50:56,  1.63it/s]
 63%|██████▎   | 8345/13317 [1:26:01<51:01,  1.62it/s]
                                                      
{'loss': 0.1963, 'grad_norm': 0.4724016785621643, 'learning_rate': 3.678390968350016e-06, 'epoch': 1.88}

 63%|██████▎   | 8345/13317 [1:26:01<51:01,  1.62it/s]
 63%|██████▎   | 8346/13317 [1:26:02<50:58,  1.63it/s]
                                                      
{'loss': 0.2068, 'grad_norm': 0.5074856281280518, 'learning_rate': 3.6771269918801656e-06, 'epoch': 1.88}

 63%|██████▎   | 8346/13317 [1:26:02<50:58,  1.63it/s]
 63%|██████▎   | 8347/13317 [1:26:02<50:55,  1.63it/s]
                                                      
{'loss': 0.1666, 'grad_norm': 0.4528160095214844, 'learning_rate': 3.675863106305696e-06, 'epoch': 1.88}

 63%|██████▎   | 8347/13317 [1:26:02<50:55,  1.63it/s]
 63%|██████▎   | 8348/13317 [1:26:03<50:54,  1.63it/s]
                                                      
{'loss': 0.171, 'grad_norm': 0.46677979826927185, 'learning_rate': 3.6745993117134504e-06, 'epoch': 1.88}

 63%|██████▎   | 8348/13317 [1:26:03<50:54,  1.63it/s]
 63%|██████▎   | 8349/13317 [1:26:03<50:52,  1.63it/s]
                                                      
{'loss': 0.2407, 'grad_norm': 0.636364758014679, 'learning_rate': 3.673335608190267e-06, 'epoch': 1.88}

 63%|██████▎   | 8349/13317 [1:26:04<50:52,  1.63it/s]
 63%|██████▎   | 8350/13317 [1:26:04<50:53,  1.63it/s]
                                                      
{'loss': 0.1955, 'grad_norm': 0.4866369068622589, 'learning_rate': 3.6720719958229752e-06, 'epoch': 1.88}

 63%|██████▎   | 8350/13317 [1:26:04<50:53,  1.63it/s]
 63%|██████▎   | 8351/13317 [1:26:05<50:49,  1.63it/s]
                                                      
{'loss': 0.182, 'grad_norm': 0.576156735420227, 'learning_rate': 3.670808474698396e-06, 'epoch': 1.88}

 63%|██████▎   | 8351/13317 [1:26:05<50:49,  1.63it/s]
 63%|██████▎   | 8352/13317 [1:26:05<50:49,  1.63it/s]
                                                      
{'loss': 0.2161, 'grad_norm': 0.6455330848693848, 'learning_rate': 3.669545044903349e-06, 'epoch': 1.88}

 63%|██████▎   | 8352/13317 [1:26:05<50:49,  1.63it/s]
 63%|██████▎   | 8353/13317 [1:26:06<50:46,  1.63it/s]
                                                      
{'loss': 0.181, 'grad_norm': 0.48274508118629456, 'learning_rate': 3.6682817065246454e-06, 'epoch': 1.88}

 63%|██████▎   | 8353/13317 [1:26:06<50:46,  1.63it/s]
 63%|██████▎   | 8354/13317 [1:26:07<50:45,  1.63it/s]
                                                      
{'loss': 0.2209, 'grad_norm': 0.5220543146133423, 'learning_rate': 3.6670184596490876e-06, 'epoch': 1.88}

 63%|██████▎   | 8354/13317 [1:26:07<50:45,  1.63it/s]
 63%|██████▎   | 8355/13317 [1:26:07<50:46,  1.63it/s]
                                                      
{'loss': 0.2015, 'grad_norm': 0.597937822341919, 'learning_rate': 3.6657553043634747e-06, 'epoch': 1.88}

 63%|██████▎   | 8355/13317 [1:26:07<50:46,  1.63it/s]
 63%|██████▎   | 8356/13317 [1:26:08<50:45,  1.63it/s]
                                                      
{'loss': 0.1631, 'grad_norm': 0.4586212635040283, 'learning_rate': 3.6644922407546018e-06, 'epoch': 1.88}

 63%|██████▎   | 8356/13317 [1:26:08<50:45,  1.63it/s]
 63%|██████▎   | 8357/13317 [1:26:08<50:43,  1.63it/s]
                                                      
{'loss': 0.2279, 'grad_norm': 0.582848310470581, 'learning_rate': 3.6632292689092515e-06, 'epoch': 1.88}

 63%|██████▎   | 8357/13317 [1:26:09<50:43,  1.63it/s]
 63%|██████▎   | 8358/13317 [1:26:09<50:41,  1.63it/s]
                                                      
{'loss': 0.1697, 'grad_norm': 0.4728994369506836, 'learning_rate': 3.6619663889142042e-06, 'epoch': 1.88}

 63%|██████▎   | 8358/13317 [1:26:09<50:41,  1.63it/s]
 63%|██████▎   | 8359/13317 [1:26:10<50:41,  1.63it/s]
                                                      
{'loss': 0.2326, 'grad_norm': 0.5796023607254028, 'learning_rate': 3.6607036008562337e-06, 'epoch': 1.88}

 63%|██████▎   | 8359/13317 [1:26:10<50:41,  1.63it/s]
 63%|██████▎   | 8360/13317 [1:26:10<50:43,  1.63it/s]
                                                      
{'loss': 0.1721, 'grad_norm': 0.5369097590446472, 'learning_rate': 3.659440904822107e-06, 'epoch': 1.88}

 63%|██████▎   | 8360/13317 [1:26:10<50:43,  1.63it/s]
 63%|██████▎   | 8361/13317 [1:26:11<50:41,  1.63it/s]
                                                      
{'loss': 0.2014, 'grad_norm': 0.5190542936325073, 'learning_rate': 3.6581783008985817e-06, 'epoch': 1.88}

 63%|██████▎   | 8361/13317 [1:26:11<50:41,  1.63it/s]
 63%|██████▎   | 8362/13317 [1:26:11<50:41,  1.63it/s]
                                                      
{'loss': 0.2024, 'grad_norm': 0.4702089726924896, 'learning_rate': 3.6569157891724166e-06, 'epoch': 1.88}

 63%|██████▎   | 8362/13317 [1:26:12<50:41,  1.63it/s]
 63%|██████▎   | 8363/13317 [1:26:12<50:38,  1.63it/s]
                                                      
{'loss': 0.1803, 'grad_norm': 0.4577096104621887, 'learning_rate': 3.655653369730358e-06, 'epoch': 1.88}

 63%|██████▎   | 8363/13317 [1:26:12<50:38,  1.63it/s]
 63%|██████▎   | 8364/13317 [1:26:13<50:36,  1.63it/s]
                                                      
{'loss': 0.1608, 'grad_norm': 0.4449288845062256, 'learning_rate': 3.6543910426591467e-06, 'epoch': 1.88}

 63%|██████▎   | 8364/13317 [1:26:13<50:36,  1.63it/s]
 63%|██████▎   | 8365/13317 [1:26:13<50:59,  1.62it/s]
                                                      
{'loss': 0.1489, 'grad_norm': 0.461972713470459, 'learning_rate': 3.6531288080455176e-06, 'epoch': 1.88}

 63%|██████▎   | 8365/13317 [1:26:13<50:59,  1.62it/s]
 63%|██████▎   | 8366/13317 [1:26:14<50:51,  1.62it/s]
                                                      
{'loss': 0.1876, 'grad_norm': 0.5131687521934509, 'learning_rate': 3.6518666659761994e-06, 'epoch': 1.88}

 63%|██████▎   | 8366/13317 [1:26:14<50:51,  1.62it/s]
 63%|██████▎   | 8367/13317 [1:26:15<50:45,  1.63it/s]
                                                      
{'loss': 0.1863, 'grad_norm': 0.5131630897521973, 'learning_rate': 3.650604616537919e-06, 'epoch': 1.88}

 63%|██████▎   | 8367/13317 [1:26:15<50:45,  1.63it/s]
 63%|██████▎   | 8368/13317 [1:26:15<50:42,  1.63it/s]
                                                      
{'loss': 0.1941, 'grad_norm': 0.48674696683883667, 'learning_rate': 3.6493426598173853e-06, 'epoch': 1.89}

 63%|██████▎   | 8368/13317 [1:26:15<50:42,  1.63it/s]
 63%|██████▎   | 8369/13317 [1:26:16<50:38,  1.63it/s]
                                                      
{'loss': 0.1873, 'grad_norm': 0.4753018021583557, 'learning_rate': 3.6480807959013135e-06, 'epoch': 1.89}

 63%|██████▎   | 8369/13317 [1:26:16<50:38,  1.63it/s]
 63%|██████▎   | 8370/13317 [1:26:16<50:43,  1.63it/s]
                                                      
{'loss': 0.1864, 'grad_norm': 0.4755779802799225, 'learning_rate': 3.6468190248764063e-06, 'epoch': 1.89}

 63%|██████▎   | 8370/13317 [1:26:17<50:43,  1.63it/s]
 63%|██████▎   | 8371/13317 [1:26:17<50:39,  1.63it/s]
                                                      
{'loss': 0.1665, 'grad_norm': 0.48196202516555786, 'learning_rate': 3.6455573468293587e-06, 'epoch': 1.89}

 63%|██████▎   | 8371/13317 [1:26:17<50:39,  1.63it/s]
 63%|██████▎   | 8372/13317 [1:26:18<50:37,  1.63it/s]
                                                      
{'loss': 0.2622, 'grad_norm': 0.5787290334701538, 'learning_rate': 3.644295761846862e-06, 'epoch': 1.89}

 63%|██████▎   | 8372/13317 [1:26:18<50:37,  1.63it/s]
 63%|██████▎   | 8373/13317 [1:26:18<50:38,  1.63it/s]
                                                      
{'loss': 0.2086, 'grad_norm': 0.540163516998291, 'learning_rate': 3.643034270015603e-06, 'epoch': 1.89}

 63%|██████▎   | 8373/13317 [1:26:18<50:38,  1.63it/s]
 63%|██████▎   | 8374/13317 [1:26:19<50:38,  1.63it/s]
                                                      
{'loss': 0.192, 'grad_norm': 0.5092630982398987, 'learning_rate': 3.6417728714222556e-06, 'epoch': 1.89}

 63%|██████▎   | 8374/13317 [1:26:19<50:38,  1.63it/s]
 63%|██████▎   | 8375/13317 [1:26:19<50:41,  1.62it/s]
                                                      
{'loss': 0.1708, 'grad_norm': 0.47996315360069275, 'learning_rate': 3.6405115661534925e-06, 'epoch': 1.89}

 63%|██████▎   | 8375/13317 [1:26:20<50:41,  1.62it/s]
 63%|██████▎   | 8376/13317 [1:26:20<50:36,  1.63it/s]
                                                      
{'loss': 0.2657, 'grad_norm': 0.9590208530426025, 'learning_rate': 3.63925035429598e-06, 'epoch': 1.89}

 63%|██████▎   | 8376/13317 [1:26:20<50:36,  1.63it/s]
 63%|██████▎   | 8377/13317 [1:26:21<50:36,  1.63it/s]
                                                      
{'loss': 0.1733, 'grad_norm': 0.47701770067214966, 'learning_rate': 3.6379892359363765e-06, 'epoch': 1.89}

 63%|██████▎   | 8377/13317 [1:26:21<50:36,  1.63it/s]
 63%|██████▎   | 8378/13317 [1:26:21<50:35,  1.63it/s]
                                                      
{'loss': 0.1832, 'grad_norm': 0.4501291811466217, 'learning_rate': 3.6367282111613333e-06, 'epoch': 1.89}

 63%|██████▎   | 8378/13317 [1:26:21<50:35,  1.63it/s]
 63%|██████▎   | 8379/13317 [1:26:22<50:34,  1.63it/s]
                                                      
{'loss': 0.2113, 'grad_norm': 0.5558070540428162, 'learning_rate': 3.6354672800574962e-06, 'epoch': 1.89}

 63%|██████▎   | 8379/13317 [1:26:22<50:34,  1.63it/s]
 63%|██████▎   | 8380/13317 [1:26:23<50:37,  1.63it/s]
                                                      
{'loss': 0.1889, 'grad_norm': 0.5136067867279053, 'learning_rate': 3.6342064427115066e-06, 'epoch': 1.89}

 63%|██████▎   | 8380/13317 [1:26:23<50:37,  1.63it/s]
 63%|██████▎   | 8381/13317 [1:26:23<50:36,  1.63it/s]
                                                      
{'loss': 0.262, 'grad_norm': 0.7654765844345093, 'learning_rate': 3.632945699209994e-06, 'epoch': 1.89}

 63%|██████▎   | 8381/13317 [1:26:23<50:36,  1.63it/s]
 63%|██████▎   | 8382/13317 [1:26:24<50:34,  1.63it/s]
                                                      
{'loss': 0.2009, 'grad_norm': 0.5141570568084717, 'learning_rate': 3.6316850496395863e-06, 'epoch': 1.89}

 63%|██████▎   | 8382/13317 [1:26:24<50:34,  1.63it/s]
 63%|██████▎   | 8383/13317 [1:26:24<50:35,  1.63it/s]
                                                      
{'loss': 0.204, 'grad_norm': 0.5209286212921143, 'learning_rate': 3.6304244940869036e-06, 'epoch': 1.89}

 63%|██████▎   | 8383/13317 [1:26:25<50:35,  1.63it/s]
 63%|██████▎   | 8384/13317 [1:26:25<50:32,  1.63it/s]
                                                      
{'loss': 0.1977, 'grad_norm': 0.5250702500343323, 'learning_rate': 3.6291640326385607e-06, 'epoch': 1.89}

 63%|██████▎   | 8384/13317 [1:26:25<50:32,  1.63it/s]
 63%|██████▎   | 8385/13317 [1:26:26<50:33,  1.63it/s]
                                                      
{'loss': 0.1698, 'grad_norm': 0.454780250787735, 'learning_rate': 3.627903665381163e-06, 'epoch': 1.89}

 63%|██████▎   | 8385/13317 [1:26:26<50:33,  1.63it/s]
 63%|██████▎   | 8386/13317 [1:26:26<50:29,  1.63it/s]
                                                      
{'loss': 0.1906, 'grad_norm': 0.6227230429649353, 'learning_rate': 3.6266433924013094e-06, 'epoch': 1.89}

 63%|██████▎   | 8386/13317 [1:26:26<50:29,  1.63it/s]
 63%|██████▎   | 8387/13317 [1:26:27<50:25,  1.63it/s]
                                                      
{'loss': 0.2243, 'grad_norm': 0.6268666982650757, 'learning_rate': 3.6253832137856e-06, 'epoch': 1.89}

 63%|██████▎   | 8387/13317 [1:26:27<50:25,  1.63it/s]
 63%|██████▎   | 8388/13317 [1:26:27<50:28,  1.63it/s]
                                                      
{'loss': 0.1869, 'grad_norm': 0.5368334054946899, 'learning_rate': 3.624123129620615e-06, 'epoch': 1.89}

 63%|██████▎   | 8388/13317 [1:26:28<50:28,  1.63it/s]
 63%|██████▎   | 8389/13317 [1:26:28<50:24,  1.63it/s]
                                                      
{'loss': 0.2506, 'grad_norm': 0.5965041518211365, 'learning_rate': 3.62286313999294e-06, 'epoch': 1.89}

 63%|██████▎   | 8389/13317 [1:26:28<50:24,  1.63it/s]
 63%|██████▎   | 8390/13317 [1:26:29<50:29,  1.63it/s]
                                                      
{'loss': 0.3018, 'grad_norm': 0.5228327512741089, 'learning_rate': 3.6216032449891504e-06, 'epoch': 1.89}

 63%|██████▎   | 8390/13317 [1:26:29<50:29,  1.63it/s]
 63%|██████▎   | 8391/13317 [1:26:29<50:25,  1.63it/s]
                                                      
{'loss': 0.1675, 'grad_norm': 0.5063783526420593, 'learning_rate': 3.620343444695811e-06, 'epoch': 1.89}

 63%|██████▎   | 8391/13317 [1:26:29<50:25,  1.63it/s]
 63%|██████▎   | 8392/13317 [1:26:30<50:23,  1.63it/s]
                                                      
{'loss': 0.1627, 'grad_norm': 0.519229531288147, 'learning_rate': 3.619083739199485e-06, 'epoch': 1.89}

 63%|██████▎   | 8392/13317 [1:26:30<50:23,  1.63it/s]
 63%|██████▎   | 8393/13317 [1:26:31<50:24,  1.63it/s]
                                                      
{'loss': 0.1459, 'grad_norm': 0.45517662167549133, 'learning_rate': 3.617824128586726e-06, 'epoch': 1.89}

 63%|██████▎   | 8393/13317 [1:26:31<50:24,  1.63it/s]
 63%|██████▎   | 8394/13317 [1:26:31<50:23,  1.63it/s]
                                                      
{'loss': 0.2517, 'grad_norm': 0.5930984020233154, 'learning_rate': 3.616564612944088e-06, 'epoch': 1.89}

 63%|██████▎   | 8394/13317 [1:26:31<50:23,  1.63it/s]
 63%|██████▎   | 8395/13317 [1:26:32<50:27,  1.63it/s]
                                                      
{'loss': 0.1771, 'grad_norm': 0.5293441414833069, 'learning_rate': 3.6153051923581058e-06, 'epoch': 1.89}

 63%|██████▎   | 8395/13317 [1:26:32<50:27,  1.63it/s]
 63%|██████▎   | 8396/13317 [1:26:32<50:24,  1.63it/s]
                                                      
{'loss': 0.1973, 'grad_norm': 0.5188011527061462, 'learning_rate': 3.614045866915319e-06, 'epoch': 1.89}

 63%|██████▎   | 8396/13317 [1:26:32<50:24,  1.63it/s]
 63%|██████▎   | 8397/13317 [1:26:33<50:22,  1.63it/s]
                                                      
{'loss': 0.1909, 'grad_norm': 0.5472999215126038, 'learning_rate': 3.612786636702257e-06, 'epoch': 1.89}

 63%|██████▎   | 8397/13317 [1:26:33<50:22,  1.63it/s]
 63%|██████▎   | 8398/13317 [1:26:34<50:20,  1.63it/s]
                                                      
{'loss': 0.2677, 'grad_norm': 0.6801888942718506, 'learning_rate': 3.6115275018054396e-06, 'epoch': 1.89}

 63%|██████▎   | 8398/13317 [1:26:34<50:20,  1.63it/s]
 63%|██████▎   | 8399/13317 [1:26:34<50:18,  1.63it/s]
                                                      
{'loss': 0.1977, 'grad_norm': 0.536766767501831, 'learning_rate': 3.6102684623113847e-06, 'epoch': 1.89}

 63%|██████▎   | 8399/13317 [1:26:34<50:18,  1.63it/s]
 63%|██████▎   | 8400/13317 [1:26:35<50:21,  1.63it/s]
                                                      
{'loss': 0.1743, 'grad_norm': 0.5004127621650696, 'learning_rate': 3.609009518306601e-06, 'epoch': 1.89}

 63%|██████▎   | 8400/13317 [1:26:35<50:21,  1.63it/s]
 63%|██████▎   | 8401/13317 [1:26:35<50:20,  1.63it/s]
                                                      
{'loss': 0.1776, 'grad_norm': 0.5685902833938599, 'learning_rate': 3.607750669877593e-06, 'epoch': 1.89}

 63%|██████▎   | 8401/13317 [1:26:36<50:20,  1.63it/s]
 63%|██████▎   | 8402/13317 [1:26:36<50:20,  1.63it/s]
                                                      
{'loss': 0.2037, 'grad_norm': 0.6049041152000427, 'learning_rate': 3.6064919171108537e-06, 'epoch': 1.89}

 63%|██████▎   | 8402/13317 [1:26:36<50:20,  1.63it/s]
 63%|██████▎   | 8403/13317 [1:26:37<50:20,  1.63it/s]
                                                      
{'loss': 0.1842, 'grad_norm': 0.4845612645149231, 'learning_rate': 3.605233260092874e-06, 'epoch': 1.89}

 63%|██████▎   | 8403/13317 [1:26:37<50:20,  1.63it/s]
 63%|██████▎   | 8404/13317 [1:26:37<50:18,  1.63it/s]
                                                      
{'loss': 0.2186, 'grad_norm': 0.5291534662246704, 'learning_rate': 3.6039746989101395e-06, 'epoch': 1.89}

 63%|██████▎   | 8404/13317 [1:26:37<50:18,  1.63it/s]
 63%|██████▎   | 8405/13317 [1:26:38<50:19,  1.63it/s]
                                                      
{'loss': 0.1927, 'grad_norm': 0.5313776135444641, 'learning_rate': 3.602716233649123e-06, 'epoch': 1.89}

 63%|██████▎   | 8405/13317 [1:26:38<50:19,  1.63it/s]
 63%|██████▎   | 8406/13317 [1:26:39<50:15,  1.63it/s]
                                                      
{'loss': 0.1734, 'grad_norm': 0.47098150849342346, 'learning_rate': 3.601457864396296e-06, 'epoch': 1.89}

 63%|██████▎   | 8406/13317 [1:26:39<50:15,  1.63it/s]
 63%|██████▎   | 8407/13317 [1:26:39<50:13,  1.63it/s]
                                                      
{'loss': 0.2106, 'grad_norm': 0.5508794784545898, 'learning_rate': 3.6001995912381227e-06, 'epoch': 1.89}

 63%|██████▎   | 8407/13317 [1:26:39<50:13,  1.63it/s]
 63%|██████▎   | 8408/13317 [1:26:40<50:14,  1.63it/s]
                                                      
{'loss': 0.2132, 'grad_norm': 0.5966536998748779, 'learning_rate': 3.598941414261057e-06, 'epoch': 1.89}

 63%|██████▎   | 8408/13317 [1:26:40<50:14,  1.63it/s]
 63%|██████▎   | 8409/13317 [1:26:40<50:13,  1.63it/s]
                                                      
{'loss': 0.1895, 'grad_norm': 0.5450761914253235, 'learning_rate': 3.5976833335515517e-06, 'epoch': 1.89}

 63%|██████▎   | 8409/13317 [1:26:40<50:13,  1.63it/s]
 63%|██████▎   | 8410/13317 [1:26:41<50:15,  1.63it/s]
                                                      
{'loss': 0.1936, 'grad_norm': 0.5364528894424438, 'learning_rate': 3.5964253491960487e-06, 'epoch': 1.89}

 63%|██████▎   | 8410/13317 [1:26:41<50:15,  1.63it/s]
 63%|██████▎   | 8411/13317 [1:26:42<50:12,  1.63it/s]
                                                      
{'loss': 0.1761, 'grad_norm': 0.46335938572883606, 'learning_rate': 3.5951674612809874e-06, 'epoch': 1.89}

 63%|██████▎   | 8411/13317 [1:26:42<50:12,  1.63it/s]
 63%|██████▎   | 8412/13317 [1:26:42<50:11,  1.63it/s]
                                                      
{'loss': 0.2251, 'grad_norm': 0.5557557940483093, 'learning_rate': 3.5939096698927946e-06, 'epoch': 1.9}

 63%|██████▎   | 8412/13317 [1:26:42<50:11,  1.63it/s]
 63%|██████▎   | 8413/13317 [1:26:43<50:12,  1.63it/s]
                                                      
{'loss': 0.2127, 'grad_norm': 0.6156031489372253, 'learning_rate': 3.5926519751178946e-06, 'epoch': 1.9}

 63%|██████▎   | 8413/13317 [1:26:43<50:12,  1.63it/s]
 63%|██████▎   | 8414/13317 [1:26:43<50:10,  1.63it/s]
                                                      
{'loss': 0.1761, 'grad_norm': 0.5655094385147095, 'learning_rate': 3.5913943770427085e-06, 'epoch': 1.9}

 63%|██████▎   | 8414/13317 [1:26:44<50:10,  1.63it/s]
 63%|██████▎   | 8415/13317 [1:26:44<50:11,  1.63it/s]
                                                      
{'loss': 0.154, 'grad_norm': 0.482270747423172, 'learning_rate': 3.5901368757536393e-06, 'epoch': 1.9}

 63%|██████▎   | 8415/13317 [1:26:44<50:11,  1.63it/s]
 63%|██████▎   | 8416/13317 [1:26:45<50:08,  1.63it/s]
                                                      
{'loss': 0.1844, 'grad_norm': 0.5088723301887512, 'learning_rate': 3.5888794713370965e-06, 'epoch': 1.9}

 63%|██████▎   | 8416/13317 [1:26:45<50:08,  1.63it/s]
 63%|██████▎   | 8417/13317 [1:26:45<50:06,  1.63it/s]
                                                      
{'loss': 0.1657, 'grad_norm': 0.4332846701145172, 'learning_rate': 3.5876221638794762e-06, 'epoch': 1.9}

 63%|██████▎   | 8417/13317 [1:26:45<50:06,  1.63it/s]
 63%|██████▎   | 8418/13317 [1:26:46<50:09,  1.63it/s]
                                                      
{'loss': 0.216, 'grad_norm': 0.5536457300186157, 'learning_rate': 3.586364953467166e-06, 'epoch': 1.9}

 63%|██████▎   | 8418/13317 [1:26:46<50:09,  1.63it/s]
 63%|██████▎   | 8419/13317 [1:26:46<50:06,  1.63it/s]
                                                      
{'loss': 0.1582, 'grad_norm': 0.456676185131073, 'learning_rate': 3.585107840186552e-06, 'epoch': 1.9}

 63%|██████▎   | 8419/13317 [1:26:47<50:06,  1.63it/s]
 63%|██████▎   | 8420/13317 [1:26:47<50:07,  1.63it/s]
                                                      
{'loss': 0.1728, 'grad_norm': 0.49657687544822693, 'learning_rate': 3.583850824124009e-06, 'epoch': 1.9}

 63%|██████▎   | 8420/13317 [1:26:47<50:07,  1.63it/s]
 63%|██████▎   | 8421/13317 [1:26:48<50:06,  1.63it/s]
                                                      
{'loss': 0.2569, 'grad_norm': 0.6439160108566284, 'learning_rate': 3.582593905365912e-06, 'epoch': 1.9}

 63%|██████▎   | 8421/13317 [1:26:48<50:06,  1.63it/s]
 63%|██████▎   | 8422/13317 [1:26:48<50:05,  1.63it/s]
                                                      
{'loss': 0.2725, 'grad_norm': 0.6368997097015381, 'learning_rate': 3.5813370839986182e-06, 'epoch': 1.9}

 63%|██████▎   | 8422/13317 [1:26:48<50:05,  1.63it/s]
 63%|██████▎   | 8423/13317 [1:26:49<50:07,  1.63it/s]
                                                      
{'loss': 0.2362, 'grad_norm': 0.58073490858078, 'learning_rate': 3.5800803601084894e-06, 'epoch': 1.9}

 63%|██████▎   | 8423/13317 [1:26:49<50:07,  1.63it/s]
 63%|██████▎   | 8424/13317 [1:26:50<50:06,  1.63it/s]
                                                      
{'loss': 0.1943, 'grad_norm': 0.47661393880844116, 'learning_rate': 3.578823733781875e-06, 'epoch': 1.9}

 63%|██████▎   | 8424/13317 [1:26:50<50:06,  1.63it/s]
 63%|██████▎   | 8425/13317 [1:26:50<50:10,  1.63it/s]
                                                      
{'loss': 0.1881, 'grad_norm': 0.5212938189506531, 'learning_rate': 3.577567205105117e-06, 'epoch': 1.9}

 63%|██████▎   | 8425/13317 [1:26:50<50:10,  1.63it/s]
 63%|██████▎   | 8426/13317 [1:26:51<50:06,  1.63it/s]
                                                      
{'loss': 0.1622, 'grad_norm': 0.4846746027469635, 'learning_rate': 3.576310774164552e-06, 'epoch': 1.9}

 63%|██████▎   | 8426/13317 [1:26:51<50:06,  1.63it/s]
 63%|██████▎   | 8427/13317 [1:26:51<50:04,  1.63it/s]
                                                      
{'loss': 0.2041, 'grad_norm': 0.5597932934761047, 'learning_rate': 3.5750544410465103e-06, 'epoch': 1.9}

 63%|██████▎   | 8427/13317 [1:26:52<50:04,  1.63it/s]
 63%|██████▎   | 8428/13317 [1:26:52<50:01,  1.63it/s]
                                                      
{'loss': 0.2257, 'grad_norm': 0.5634179711341858, 'learning_rate': 3.57379820583732e-06, 'epoch': 1.9}

 63%|██████▎   | 8428/13317 [1:26:52<50:01,  1.63it/s]
 63%|██████▎   | 8429/13317 [1:26:53<50:00,  1.63it/s]
                                                      
{'loss': 0.1765, 'grad_norm': 0.47502684593200684, 'learning_rate': 3.5725420686232903e-06, 'epoch': 1.9}

 63%|██████▎   | 8429/13317 [1:26:53<50:00,  1.63it/s]
 63%|██████▎   | 8430/13317 [1:26:53<50:14,  1.62it/s]
                                                      
{'loss': 0.1915, 'grad_norm': 0.560696542263031, 'learning_rate': 3.5712860294907354e-06, 'epoch': 1.9}

 63%|██████▎   | 8430/13317 [1:26:53<50:14,  1.62it/s]
 63%|██████▎   | 8431/13317 [1:26:54<50:08,  1.62it/s]
                                                      
{'loss': 0.2057, 'grad_norm': 0.5686498880386353, 'learning_rate': 3.5700300885259588e-06, 'epoch': 1.9}

 63%|██████▎   | 8431/13317 [1:26:54<50:08,  1.62it/s]
 63%|██████▎   | 8432/13317 [1:26:54<50:03,  1.63it/s]
                                                      
{'loss': 0.1726, 'grad_norm': 0.4582690894603729, 'learning_rate': 3.5687742458152553e-06, 'epoch': 1.9}

 63%|██████▎   | 8432/13317 [1:26:55<50:03,  1.63it/s]
 63%|██████▎   | 8433/13317 [1:26:55<49:59,  1.63it/s]
                                                      
{'loss': 0.1714, 'grad_norm': 0.4939255714416504, 'learning_rate': 3.567518501444915e-06, 'epoch': 1.9}

 63%|██████▎   | 8433/13317 [1:26:55<49:59,  1.63it/s]
 63%|██████▎   | 8434/13317 [1:26:56<49:56,  1.63it/s]
                                                      
{'loss': 0.1851, 'grad_norm': 0.5233125686645508, 'learning_rate': 3.5662628555012217e-06, 'epoch': 1.9}

 63%|██████▎   | 8434/13317 [1:26:56<49:56,  1.63it/s]
 63%|██████▎   | 8435/13317 [1:26:56<49:58,  1.63it/s]
                                                      
{'loss': 0.1874, 'grad_norm': 0.5621899962425232, 'learning_rate': 3.565007308070449e-06, 'epoch': 1.9}

 63%|██████▎   | 8435/13317 [1:26:56<49:58,  1.63it/s]
 63%|██████▎   | 8436/13317 [1:26:57<49:55,  1.63it/s]
                                                      
{'loss': 0.1495, 'grad_norm': 0.4732416272163391, 'learning_rate': 3.5637518592388693e-06, 'epoch': 1.9}

 63%|██████▎   | 8436/13317 [1:26:57<49:55,  1.63it/s]
 63%|██████▎   | 8437/13317 [1:26:58<49:53,  1.63it/s]
                                                      
{'loss': 0.2653, 'grad_norm': 0.6082133650779724, 'learning_rate': 3.5624965090927443e-06, 'epoch': 1.9}

 63%|██████▎   | 8437/13317 [1:26:58<49:53,  1.63it/s]
 63%|██████▎   | 8438/13317 [1:26:58<49:52,  1.63it/s]
                                                      
{'loss': 0.2085, 'grad_norm': 0.5538050532341003, 'learning_rate': 3.5612412577183305e-06, 'epoch': 1.9}

 63%|██████▎   | 8438/13317 [1:26:58<49:52,  1.63it/s]
 63%|██████▎   | 8439/13317 [1:26:59<49:53,  1.63it/s]
                                                      
{'loss': 0.1893, 'grad_norm': 0.5415672659873962, 'learning_rate': 3.5599861052018747e-06, 'epoch': 1.9}

 63%|██████▎   | 8439/13317 [1:26:59<49:53,  1.63it/s]
 63%|██████▎   | 8440/13317 [1:26:59<49:55,  1.63it/s]
                                                      
{'loss': 0.2243, 'grad_norm': 0.5716651678085327, 'learning_rate': 3.5587310516296203e-06, 'epoch': 1.9}

 63%|██████▎   | 8440/13317 [1:27:00<49:55,  1.63it/s]
 63%|██████▎   | 8441/13317 [1:27:00<49:53,  1.63it/s]
                                                      
{'loss': 0.137, 'grad_norm': 0.3680700659751892, 'learning_rate': 3.5574760970878054e-06, 'epoch': 1.9}

 63%|██████▎   | 8441/13317 [1:27:00<49:53,  1.63it/s]
 63%|██████▎   | 8442/13317 [1:27:01<49:50,  1.63it/s]
                                                      
{'loss': 0.2134, 'grad_norm': 0.5709421038627625, 'learning_rate': 3.5562212416626536e-06, 'epoch': 1.9}

 63%|██████▎   | 8442/13317 [1:27:01<49:50,  1.63it/s]
 63%|██████▎   | 8443/13317 [1:27:01<49:50,  1.63it/s]
                                                      
{'loss': 0.1955, 'grad_norm': 0.5205080509185791, 'learning_rate': 3.5549664854403904e-06, 'epoch': 1.9}

 63%|██████▎   | 8443/13317 [1:27:01<49:50,  1.63it/s]
 63%|██████▎   | 8444/13317 [1:27:02<49:49,  1.63it/s]
                                                      
{'loss': 0.1976, 'grad_norm': 0.5408107042312622, 'learning_rate': 3.55371182850723e-06, 'epoch': 1.9}

 63%|██████▎   | 8444/13317 [1:27:02<49:49,  1.63it/s]
 63%|██████▎   | 8445/13317 [1:27:02<49:50,  1.63it/s]
                                                      
{'loss': 0.1821, 'grad_norm': 0.5135701894760132, 'learning_rate': 3.5524572709493817e-06, 'epoch': 1.9}

 63%|██████▎   | 8445/13317 [1:27:03<49:50,  1.63it/s]
 63%|██████▎   | 8446/13317 [1:27:03<49:49,  1.63it/s]
                                                      
{'loss': 0.1595, 'grad_norm': 0.5176504254341125, 'learning_rate': 3.551202812853044e-06, 'epoch': 1.9}

 63%|██████▎   | 8446/13317 [1:27:03<49:49,  1.63it/s]
 63%|██████▎   | 8447/13317 [1:27:04<49:46,  1.63it/s]
                                                      
{'loss': 0.1419, 'grad_norm': 0.41459518671035767, 'learning_rate': 3.5499484543044125e-06, 'epoch': 1.9}

 63%|██████▎   | 8447/13317 [1:27:04<49:46,  1.63it/s]
 63%|██████▎   | 8448/13317 [1:27:04<49:45,  1.63it/s]
                                                      
{'loss': 0.248, 'grad_norm': 0.6023019552230835, 'learning_rate': 3.548694195389679e-06, 'epoch': 1.9}

 63%|██████▎   | 8448/13317 [1:27:04<49:45,  1.63it/s]
 63%|██████▎   | 8449/13317 [1:27:05<49:46,  1.63it/s]
                                                      
{'loss': 0.1757, 'grad_norm': 0.48698297142982483, 'learning_rate': 3.547440036195018e-06, 'epoch': 1.9}

 63%|██████▎   | 8449/13317 [1:27:05<49:46,  1.63it/s]
 63%|██████▎   | 8450/13317 [1:27:06<49:49,  1.63it/s]
                                                      
{'loss': 0.2562, 'grad_norm': 0.6641130447387695, 'learning_rate': 3.5461859768066074e-06, 'epoch': 1.9}

 63%|██████▎   | 8450/13317 [1:27:06<49:49,  1.63it/s]
 63%|██████▎   | 8451/13317 [1:27:06<49:47,  1.63it/s]
                                                      
{'loss': 0.1928, 'grad_norm': 0.5666017532348633, 'learning_rate': 3.544932017310614e-06, 'epoch': 1.9}

 63%|██████▎   | 8451/13317 [1:27:06<49:47,  1.63it/s]
 63%|██████▎   | 8452/13317 [1:27:07<49:47,  1.63it/s]
                                                      
{'loss': 0.1822, 'grad_norm': 0.4807245135307312, 'learning_rate': 3.5436781577931967e-06, 'epoch': 1.9}

 63%|██████▎   | 8452/13317 [1:27:07<49:47,  1.63it/s]
 63%|██████▎   | 8453/13317 [1:27:07<49:44,  1.63it/s]
                                                      
{'loss': 0.1967, 'grad_norm': 0.5154634118080139, 'learning_rate': 3.5424243983405097e-06, 'epoch': 1.9}

 63%|██████▎   | 8453/13317 [1:27:07<49:44,  1.63it/s]
 63%|██████▎   | 8454/13317 [1:27:08<49:44,  1.63it/s]
                                                      
{'loss': 0.1695, 'grad_norm': 0.5205509066581726, 'learning_rate': 3.5411707390386983e-06, 'epoch': 1.9}

 63%|██████▎   | 8454/13317 [1:27:08<49:44,  1.63it/s]
 63%|██████▎   | 8455/13317 [1:27:09<49:46,  1.63it/s]
                                                      
{'loss': 0.1718, 'grad_norm': 0.472313791513443, 'learning_rate': 3.5399171799739074e-06, 'epoch': 1.9}

 63%|██████▎   | 8455/13317 [1:27:09<49:46,  1.63it/s]
 63%|██████▎   | 8456/13317 [1:27:09<49:44,  1.63it/s]
                                                      
{'loss': 0.2064, 'grad_norm': 0.6115954518318176, 'learning_rate': 3.538663721232262e-06, 'epoch': 1.9}

 63%|██████▎   | 8456/13317 [1:27:09<49:44,  1.63it/s]
 64%|██████▎   | 8457/13317 [1:27:10<49:42,  1.63it/s]
                                                      
{'loss': 0.19, 'grad_norm': 0.47663721442222595, 'learning_rate': 3.5374103628998936e-06, 'epoch': 1.91}

 64%|██████▎   | 8457/13317 [1:27:10<49:42,  1.63it/s]
 64%|██████▎   | 8458/13317 [1:27:10<49:41,  1.63it/s]
                                                      
{'loss': 0.2144, 'grad_norm': 0.548392653465271, 'learning_rate': 3.5361571050629196e-06, 'epoch': 1.91}

 64%|██████▎   | 8458/13317 [1:27:11<49:41,  1.63it/s]
 64%|██████▎   | 8459/13317 [1:27:11<49:39,  1.63it/s]
                                                      
{'loss': 0.1676, 'grad_norm': 0.41755378246307373, 'learning_rate': 3.534903947807451e-06, 'epoch': 1.91}

 64%|██████▎   | 8459/13317 [1:27:11<49:39,  1.63it/s]
 64%|██████▎   | 8460/13317 [1:27:12<49:43,  1.63it/s]
                                                      
{'loss': 0.1669, 'grad_norm': 0.4651932418346405, 'learning_rate': 3.5336508912195946e-06, 'epoch': 1.91}

 64%|██████▎   | 8460/13317 [1:27:12<49:43,  1.63it/s]
 64%|██████▎   | 8461/13317 [1:27:12<49:43,  1.63it/s]
                                                      
{'loss': 0.2066, 'grad_norm': 0.509249210357666, 'learning_rate': 3.532397935385446e-06, 'epoch': 1.91}

 64%|██████▎   | 8461/13317 [1:27:12<49:43,  1.63it/s]
 64%|██████▎   | 8462/13317 [1:27:13<49:43,  1.63it/s]
                                                      
{'loss': 0.1813, 'grad_norm': 0.5066456198692322, 'learning_rate': 3.5311450803911005e-06, 'epoch': 1.91}

 64%|██████▎   | 8462/13317 [1:27:13<49:43,  1.63it/s]
 64%|██████▎   | 8463/13317 [1:27:14<49:42,  1.63it/s]
                                                      
{'loss': 0.2027, 'grad_norm': 0.5451805591583252, 'learning_rate': 3.5298923263226405e-06, 'epoch': 1.91}

 64%|██████▎   | 8463/13317 [1:27:14<49:42,  1.63it/s]
 64%|██████▎   | 8464/13317 [1:27:14<49:42,  1.63it/s]
                                                      
{'loss': 0.1454, 'grad_norm': 0.4309399127960205, 'learning_rate': 3.528639673266142e-06, 'epoch': 1.91}

 64%|██████▎   | 8464/13317 [1:27:14<49:42,  1.63it/s]
 64%|██████▎   | 8465/13317 [1:27:15<49:46,  1.62it/s]
                                                      
{'loss': 0.1792, 'grad_norm': 0.45540717244148254, 'learning_rate': 3.5273871213076793e-06, 'epoch': 1.91}

 64%|██████▎   | 8465/13317 [1:27:15<49:46,  1.62it/s]
 64%|██████▎   | 8466/13317 [1:27:15<49:44,  1.63it/s]
                                                      
{'loss': 0.2048, 'grad_norm': 0.5983158349990845, 'learning_rate': 3.526134670533312e-06, 'epoch': 1.91}

 64%|██████▎   | 8466/13317 [1:27:15<49:44,  1.63it/s]
 64%|██████▎   | 8467/13317 [1:27:16<49:45,  1.62it/s]
                                                      
{'loss': 0.1815, 'grad_norm': 0.48484230041503906, 'learning_rate': 3.5248823210290973e-06, 'epoch': 1.91}

 64%|██████▎   | 8467/13317 [1:27:16<49:45,  1.62it/s]
 64%|██████▎   | 8468/13317 [1:27:17<49:43,  1.63it/s]
                                                      
{'loss': 0.2137, 'grad_norm': 0.5760316848754883, 'learning_rate': 3.523630072881089e-06, 'epoch': 1.91}

 64%|██████▎   | 8468/13317 [1:27:17<49:43,  1.63it/s]
 64%|██████▎   | 8469/13317 [1:27:17<49:39,  1.63it/s]
                                                      
{'loss': 0.1924, 'grad_norm': 0.4886530041694641, 'learning_rate': 3.5223779261753245e-06, 'epoch': 1.91}

 64%|██████▎   | 8469/13317 [1:27:17<49:39,  1.63it/s]
 64%|██████▎   | 8470/13317 [1:27:18<49:39,  1.63it/s]
                                                      
{'loss': 0.1829, 'grad_norm': 0.538425087928772, 'learning_rate': 3.5211258809978416e-06, 'epoch': 1.91}

 64%|██████▎   | 8470/13317 [1:27:18<49:39,  1.63it/s]
 64%|██████▎   | 8471/13317 [1:27:18<49:36,  1.63it/s]
                                                      
{'loss': 0.2157, 'grad_norm': 0.5521208643913269, 'learning_rate': 3.5198739374346695e-06, 'epoch': 1.91}

 64%|██████▎   | 8471/13317 [1:27:19<49:36,  1.63it/s]
 64%|██████▎   | 8472/13317 [1:27:19<49:37,  1.63it/s]
                                                      
{'loss': 0.21, 'grad_norm': 0.4757653474807739, 'learning_rate': 3.518622095571831e-06, 'epoch': 1.91}

 64%|██████▎   | 8472/13317 [1:27:19<49:37,  1.63it/s]
 64%|██████▎   | 8473/13317 [1:27:20<49:37,  1.63it/s]
                                                      
{'loss': 0.1512, 'grad_norm': 0.4423864483833313, 'learning_rate': 3.5173703554953386e-06, 'epoch': 1.91}

 64%|██████▎   | 8473/13317 [1:27:20<49:37,  1.63it/s]
 64%|██████▎   | 8474/13317 [1:27:20<49:35,  1.63it/s]
                                                      
{'loss': 0.1815, 'grad_norm': 0.4778220057487488, 'learning_rate': 3.5161187172912003e-06, 'epoch': 1.91}

 64%|██████▎   | 8474/13317 [1:27:20<49:35,  1.63it/s]
 64%|██████▎   | 8475/13317 [1:27:21<49:35,  1.63it/s]
                                                      
{'loss': 0.1552, 'grad_norm': 0.4840396046638489, 'learning_rate': 3.5148671810454203e-06, 'epoch': 1.91}

 64%|██████▎   | 8475/13317 [1:27:21<49:35,  1.63it/s]
 64%|██████▎   | 8476/13317 [1:27:22<49:32,  1.63it/s]
                                                      
{'loss': 0.2287, 'grad_norm': 0.6530333757400513, 'learning_rate': 3.513615746843987e-06, 'epoch': 1.91}

 64%|██████▎   | 8476/13317 [1:27:22<49:32,  1.63it/s]
 64%|██████▎   | 8477/13317 [1:27:22<49:29,  1.63it/s]
                                                      
{'loss': 0.2032, 'grad_norm': 0.5555511713027954, 'learning_rate': 3.5123644147728908e-06, 'epoch': 1.91}

 64%|██████▎   | 8477/13317 [1:27:22<49:29,  1.63it/s]
 64%|██████▎   | 8478/13317 [1:27:23<49:29,  1.63it/s]
                                                      
{'loss': 0.1967, 'grad_norm': 0.5281544923782349, 'learning_rate': 3.5111131849181123e-06, 'epoch': 1.91}

 64%|██████▎   | 8478/13317 [1:27:23<49:29,  1.63it/s]
 64%|██████▎   | 8479/13317 [1:27:23<49:29,  1.63it/s]
                                                      
{'loss': 0.2233, 'grad_norm': 0.5651992559432983, 'learning_rate': 3.5098620573656213e-06, 'epoch': 1.91}

 64%|██████▎   | 8479/13317 [1:27:23<49:29,  1.63it/s]
 64%|██████▎   | 8480/13317 [1:27:24<49:29,  1.63it/s]
                                                      
{'loss': 0.2565, 'grad_norm': 0.6842526197433472, 'learning_rate': 3.5086110322013846e-06, 'epoch': 1.91}

 64%|██████▎   | 8480/13317 [1:27:24<49:29,  1.63it/s]
 64%|██████▎   | 8481/13317 [1:27:25<49:31,  1.63it/s]
                                                      
{'loss': 0.1881, 'grad_norm': 0.5696654915809631, 'learning_rate': 3.50736010951136e-06, 'epoch': 1.91}

 64%|██████▎   | 8481/13317 [1:27:25<49:31,  1.63it/s]
 64%|██████▎   | 8482/13317 [1:27:25<49:30,  1.63it/s]
                                                      
{'loss': 0.1925, 'grad_norm': 0.5187650918960571, 'learning_rate': 3.506109289381503e-06, 'epoch': 1.91}

 64%|██████▎   | 8482/13317 [1:27:25<49:30,  1.63it/s]
 64%|██████▎   | 8483/13317 [1:27:26<49:29,  1.63it/s]
                                                      
{'loss': 0.2169, 'grad_norm': 0.581960916519165, 'learning_rate': 3.5048585718977523e-06, 'epoch': 1.91}

 64%|██████▎   | 8483/13317 [1:27:26<49:29,  1.63it/s]
 64%|██████▎   | 8484/13317 [1:27:26<49:26,  1.63it/s]
                                                      
{'loss': 0.2467, 'grad_norm': 0.5408003926277161, 'learning_rate': 3.5036079571460503e-06, 'epoch': 1.91}

 64%|██████▎   | 8484/13317 [1:27:27<49:26,  1.63it/s]
 64%|██████▎   | 8485/13317 [1:27:27<49:26,  1.63it/s]
                                                      
{'loss': 0.2767, 'grad_norm': 0.5904765725135803, 'learning_rate': 3.5023574452123264e-06, 'epoch': 1.91}

 64%|██████▎   | 8485/13317 [1:27:27<49:26,  1.63it/s]
 64%|██████▎   | 8486/13317 [1:27:28<49:26,  1.63it/s]
                                                      
{'loss': 0.2094, 'grad_norm': 0.5607338547706604, 'learning_rate': 3.5011070361825016e-06, 'epoch': 1.91}

 64%|██████▎   | 8486/13317 [1:27:28<49:26,  1.63it/s]
 64%|██████▎   | 8487/13317 [1:27:28<49:24,  1.63it/s]
                                                      
{'loss': 0.1746, 'grad_norm': 0.5271170735359192, 'learning_rate': 3.499856730142494e-06, 'epoch': 1.91}

 64%|██████▎   | 8487/13317 [1:27:28<49:24,  1.63it/s]
 64%|██████▎   | 8488/13317 [1:27:29<49:22,  1.63it/s]
                                                      
{'loss': 0.2028, 'grad_norm': 0.5523853302001953, 'learning_rate': 3.498606527178213e-06, 'epoch': 1.91}

 64%|██████▎   | 8488/13317 [1:27:29<49:22,  1.63it/s]
 64%|██████▎   | 8489/13317 [1:27:29<49:21,  1.63it/s]
                                                      
{'loss': 0.2579, 'grad_norm': 0.6718398928642273, 'learning_rate': 3.4973564273755623e-06, 'epoch': 1.91}

 64%|██████▎   | 8489/13317 [1:27:30<49:21,  1.63it/s]
 64%|██████▍   | 8490/13317 [1:27:30<49:24,  1.63it/s]
                                                      
{'loss': 0.1686, 'grad_norm': 0.4984041452407837, 'learning_rate': 3.496106430820434e-06, 'epoch': 1.91}

 64%|██████▍   | 8490/13317 [1:27:30<49:24,  1.63it/s]
 64%|██████▍   | 8491/13317 [1:27:31<49:23,  1.63it/s]
                                                      
{'loss': 0.1514, 'grad_norm': 0.4830670654773712, 'learning_rate': 3.494856537598717e-06, 'epoch': 1.91}

 64%|██████▍   | 8491/13317 [1:27:31<49:23,  1.63it/s]
 64%|██████▍   | 8492/13317 [1:27:31<49:23,  1.63it/s]
                                                      
{'loss': 0.1687, 'grad_norm': 0.43391215801239014, 'learning_rate': 3.4936067477962943e-06, 'epoch': 1.91}

 64%|██████▍   | 8492/13317 [1:27:31<49:23,  1.63it/s]
 64%|██████▍   | 8493/13317 [1:27:32<49:19,  1.63it/s]
                                                      
{'loss': 0.2693, 'grad_norm': 0.7359750270843506, 'learning_rate': 3.4923570614990364e-06, 'epoch': 1.91}

 64%|██████▍   | 8493/13317 [1:27:32<49:19,  1.63it/s]
 64%|██████▍   | 8494/13317 [1:27:33<49:19,  1.63it/s]
                                                      
{'loss': 0.2019, 'grad_norm': 0.5557070374488831, 'learning_rate': 3.4911074787928107e-06, 'epoch': 1.91}

 64%|██████▍   | 8494/13317 [1:27:33<49:19,  1.63it/s]
 64%|██████▍   | 8495/13317 [1:27:33<49:21,  1.63it/s]
                                                      
{'loss': 0.2335, 'grad_norm': 0.6694797277450562, 'learning_rate': 3.489857999763481e-06, 'epoch': 1.91}

 64%|██████▍   | 8495/13317 [1:27:33<49:21,  1.63it/s]
 64%|██████▍   | 8496/13317 [1:27:34<49:19,  1.63it/s]
                                                      
{'loss': 0.2334, 'grad_norm': 0.5498466491699219, 'learning_rate': 3.4886086244968927e-06, 'epoch': 1.91}

 64%|██████▍   | 8496/13317 [1:27:34<49:19,  1.63it/s]
 64%|██████▍   | 8497/13317 [1:27:34<49:19,  1.63it/s]
                                                      
{'loss': 0.2343, 'grad_norm': 0.5760383009910583, 'learning_rate': 3.487359353078896e-06, 'epoch': 1.91}

 64%|██████▍   | 8497/13317 [1:27:35<49:19,  1.63it/s]
 64%|██████▍   | 8498/13317 [1:27:35<49:18,  1.63it/s]
                                                      
{'loss': 0.1384, 'grad_norm': 0.47449129819869995, 'learning_rate': 3.4861101855953277e-06, 'epoch': 1.91}

 64%|██████▍   | 8498/13317 [1:27:35<49:18,  1.63it/s]
 64%|██████▍   | 8499/13317 [1:27:36<49:16,  1.63it/s]
                                                      
{'loss': 0.1622, 'grad_norm': 0.45091351866722107, 'learning_rate': 3.4848611221320208e-06, 'epoch': 1.91}

 64%|██████▍   | 8499/13317 [1:27:36<49:16,  1.63it/s]
 64%|██████▍   | 8500/13317 [1:27:36<49:20,  1.63it/s]
                                                      
{'loss': 0.1682, 'grad_norm': 0.4704548716545105, 'learning_rate': 3.4836121627747953e-06, 'epoch': 1.91}

 64%|██████▍   | 8500/13317 [1:27:36<49:20,  1.63it/s]

  0%|          | 0/13 [00:00<?, ?it/s]

 15%|█▌        | 2/13 [00:00<00:00, 13.37it/s]

 31%|███       | 4/13 [00:00<00:01,  8.40it/s]

 38%|███▊      | 5/13 [00:00<00:01,  7.80it/s]

 46%|████▌     | 6/13 [00:00<00:00,  7.43it/s]

 54%|█████▍    | 7/13 [00:00<00:00,  7.18it/s]

 62%|██████▏   | 8/13 [00:01<00:00,  7.02it/s]

 69%|██████▉   | 9/13 [00:01<00:00,  6.91it/s]

 77%|███████▋  | 10/13 [00:01<00:00,  6.84it/s]

 85%|████████▍ | 11/13 [00:01<00:00,  6.78it/s]

 92%|█████████▏| 12/13 [00:01<00:00,  6.75it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]
                                                      

                                               
{'eval_loss': 0.5293134450912476, 'eval_runtime': 1.952, 'eval_samples_per_second': 102.458, 'eval_steps_per_second': 6.66, 'epoch': 1.91}

 64%|██████▍   | 8500/13317 [1:27:38<49:20,  1.63it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]

                                               
 64%|██████▍   | 8501/13317 [1:27:39<1:36:28,  1.20s/it]
                                                        
{'loss': 0.2314, 'grad_norm': 0.5928805470466614, 'learning_rate': 3.4823633076094693e-06, 'epoch': 1.92}

 64%|██████▍   | 8501/13317 [1:27:39<1:36:28,  1.20s/it]
 64%|██████▍   | 8502/13317 [1:27:39<1:22:16,  1.03s/it]
                                                        
{'loss': 0.221, 'grad_norm': 0.6511198282241821, 'learning_rate': 3.481114556721856e-06, 'epoch': 1.92}

 64%|██████▍   | 8502/13317 [1:27:40<1:22:16,  1.03s/it]
 64%|██████▍   | 8503/13317 [1:27:40<1:12:20,  1.11it/s]
                                                        
{'loss': 0.1822, 'grad_norm': 0.5082912445068359, 'learning_rate': 3.4798659101977523e-06, 'epoch': 1.92}

 64%|██████▍   | 8503/13317 [1:27:40<1:12:20,  1.11it/s]
 64%|██████▍   | 8504/13317 [1:27:41<1:05:23,  1.23it/s]
                                                        
{'loss': 0.2476, 'grad_norm': 0.6086586117744446, 'learning_rate': 3.478617368122956e-06, 'epoch': 1.92}

 64%|██████▍   | 8504/13317 [1:27:41<1:05:23,  1.23it/s]
 64%|██████▍   | 8505/13317 [1:27:41<1:00:31,  1.33it/s]
                                                        
{'loss': 0.2145, 'grad_norm': 0.6002055406570435, 'learning_rate': 3.4773689305832564e-06, 'epoch': 1.92}

 64%|██████▍   | 8505/13317 [1:27:41<1:00:31,  1.33it/s]
 64%|██████▍   | 8506/13317 [1:27:42<57:06,  1.40it/s]  
                                                      
{'loss': 0.1689, 'grad_norm': 0.4666120409965515, 'learning_rate': 3.476120597664434e-06, 'epoch': 1.92}

 64%|██████▍   | 8506/13317 [1:27:42<57:06,  1.40it/s]
 64%|██████▍   | 8507/13317 [1:27:42<54:46,  1.46it/s]
                                                      
{'loss': 0.2322, 'grad_norm': 0.6248008608818054, 'learning_rate': 3.47487236945226e-06, 'epoch': 1.92}

 64%|██████▍   | 8507/13317 [1:27:43<54:46,  1.46it/s]
 64%|██████▍   | 8508/13317 [1:27:43<53:06,  1.51it/s]
                                                      
{'loss': 0.217, 'grad_norm': 0.6126794815063477, 'learning_rate': 3.473624246032502e-06, 'epoch': 1.92}

 64%|██████▍   | 8508/13317 [1:27:43<53:06,  1.51it/s]
 64%|██████▍   | 8509/13317 [1:27:44<51:54,  1.54it/s]
                                                      
{'loss': 0.2298, 'grad_norm': 0.6645832657814026, 'learning_rate': 3.472376227490921e-06, 'epoch': 1.92}

 64%|██████▍   | 8509/13317 [1:27:44<51:54,  1.54it/s]
 64%|██████▍   | 8510/13317 [1:27:44<51:03,  1.57it/s]
                                                      
{'loss': 0.1783, 'grad_norm': 0.5138168931007385, 'learning_rate': 3.4711283139132677e-06, 'epoch': 1.92}

 64%|██████▍   | 8510/13317 [1:27:44<51:03,  1.57it/s]
 64%|██████▍   | 8511/13317 [1:27:45<50:29,  1.59it/s]
                                                      
{'loss': 0.2131, 'grad_norm': 0.5483157634735107, 'learning_rate': 3.4698805053852873e-06, 'epoch': 1.92}

 64%|██████▍   | 8511/13317 [1:27:45<50:29,  1.59it/s]
 64%|██████▍   | 8512/13317 [1:27:46<50:03,  1.60it/s]
                                                      
{'loss': 0.161, 'grad_norm': 0.5023987889289856, 'learning_rate': 3.4686328019927184e-06, 'epoch': 1.92}

 64%|██████▍   | 8512/13317 [1:27:46<50:03,  1.60it/s]
 64%|██████▍   | 8513/13317 [1:27:46<49:46,  1.61it/s]
                                                      
{'loss': 0.2299, 'grad_norm': 0.6704382300376892, 'learning_rate': 3.467385203821289e-06, 'epoch': 1.92}

 64%|██████▍   | 8513/13317 [1:27:46<49:46,  1.61it/s]
 64%|██████▍   | 8514/13317 [1:27:47<49:34,  1.61it/s]
                                                      
{'loss': 0.1958, 'grad_norm': 0.4911274313926697, 'learning_rate': 3.466137710956723e-06, 'epoch': 1.92}

 64%|██████▍   | 8514/13317 [1:27:47<49:34,  1.61it/s]
 64%|██████▍   | 8515/13317 [1:27:47<49:25,  1.62it/s]
                                                      
{'loss': 0.2032, 'grad_norm': 0.5601648688316345, 'learning_rate': 3.4648903234847377e-06, 'epoch': 1.92}

 64%|██████▍   | 8515/13317 [1:27:48<49:25,  1.62it/s]
 64%|██████▍   | 8516/13317 [1:27:48<49:18,  1.62it/s]
                                                      
{'loss': 0.1871, 'grad_norm': 0.5494166612625122, 'learning_rate': 3.4636430414910423e-06, 'epoch': 1.92}

 64%|██████▍   | 8516/13317 [1:27:48<49:18,  1.62it/s]
 64%|██████▍   | 8517/13317 [1:27:49<49:13,  1.62it/s]
                                                      
{'loss': 0.2338, 'grad_norm': 0.6456691026687622, 'learning_rate': 3.462395865061336e-06, 'epoch': 1.92}

 64%|██████▍   | 8517/13317 [1:27:49<49:13,  1.62it/s]
 64%|██████▍   | 8518/13317 [1:27:49<49:10,  1.63it/s]
                                                      
{'loss': 0.1806, 'grad_norm': 0.5411462187767029, 'learning_rate': 3.4611487942813136e-06, 'epoch': 1.92}

 64%|██████▍   | 8518/13317 [1:27:49<49:10,  1.63it/s]
 64%|██████▍   | 8519/13317 [1:27:50<49:08,  1.63it/s]
                                                      
{'loss': 0.1717, 'grad_norm': 0.5216940641403198, 'learning_rate': 3.4599018292366647e-06, 'epoch': 1.92}

 64%|██████▍   | 8519/13317 [1:27:50<49:08,  1.63it/s]
 64%|██████▍   | 8520/13317 [1:27:50<49:06,  1.63it/s]
                                                      
{'loss': 0.1988, 'grad_norm': 0.5162557363510132, 'learning_rate': 3.4586549700130654e-06, 'epoch': 1.92}

 64%|██████▍   | 8520/13317 [1:27:51<49:06,  1.63it/s]
 64%|██████▍   | 8521/13317 [1:27:51<49:04,  1.63it/s]
                                                      
{'loss': 0.2937, 'grad_norm': 0.5803240537643433, 'learning_rate': 3.45740821669619e-06, 'epoch': 1.92}

 64%|██████▍   | 8521/13317 [1:27:51<49:04,  1.63it/s]
 64%|██████▍   | 8522/13317 [1:27:52<49:01,  1.63it/s]
                                                      
{'loss': 0.2325, 'grad_norm': 0.5668349266052246, 'learning_rate': 3.456161569371704e-06, 'epoch': 1.92}

 64%|██████▍   | 8522/13317 [1:27:52<49:01,  1.63it/s]
 64%|██████▍   | 8523/13317 [1:27:52<49:00,  1.63it/s]
                                                      
{'loss': 0.1656, 'grad_norm': 0.4430341422557831, 'learning_rate': 3.4549150281252635e-06, 'epoch': 1.92}

 64%|██████▍   | 8523/13317 [1:27:52<49:00,  1.63it/s]
 64%|██████▍   | 8524/13317 [1:27:53<49:00,  1.63it/s]
                                                      
{'loss': 0.1849, 'grad_norm': 0.5538650155067444, 'learning_rate': 3.4536685930425206e-06, 'epoch': 1.92}

 64%|██████▍   | 8524/13317 [1:27:53<49:00,  1.63it/s]
 64%|██████▍   | 8525/13317 [1:27:54<48:58,  1.63it/s]
                                                      
{'loss': 0.208, 'grad_norm': 0.5885974764823914, 'learning_rate': 3.452422264209118e-06, 'epoch': 1.92}

 64%|██████▍   | 8525/13317 [1:27:54<48:58,  1.63it/s]
 64%|██████▍   | 8526/13317 [1:27:54<49:00,  1.63it/s]
                                                      
{'loss': 0.2188, 'grad_norm': 0.45257145166397095, 'learning_rate': 3.451176041710693e-06, 'epoch': 1.92}

 64%|██████▍   | 8526/13317 [1:27:54<49:00,  1.63it/s]
 64%|██████▍   | 8527/13317 [1:27:55<48:59,  1.63it/s]
                                                      
{'loss': 0.2275, 'grad_norm': 0.494666188955307, 'learning_rate': 3.4499299256328723e-06, 'epoch': 1.92}

 64%|██████▍   | 8527/13317 [1:27:55<48:59,  1.63it/s]
 64%|██████▍   | 8528/13317 [1:27:55<48:58,  1.63it/s]
                                                      
{'loss': 0.2112, 'grad_norm': 0.5558033585548401, 'learning_rate': 3.4486839160612767e-06, 'epoch': 1.92}

 64%|██████▍   | 8528/13317 [1:27:56<48:58,  1.63it/s]
 64%|██████▍   | 8529/13317 [1:27:56<48:57,  1.63it/s]
                                                      
{'loss': 0.1928, 'grad_norm': 0.5077123641967773, 'learning_rate': 3.4474380130815256e-06, 'epoch': 1.92}

 64%|██████▍   | 8529/13317 [1:27:56<48:57,  1.63it/s]
 64%|██████▍   | 8530/13317 [1:27:57<48:56,  1.63it/s]
                                                      
{'loss': 0.1606, 'grad_norm': 0.465447336435318, 'learning_rate': 3.4461922167792172e-06, 'epoch': 1.92}

 64%|██████▍   | 8530/13317 [1:27:57<48:56,  1.63it/s]
 64%|██████▍   | 8531/13317 [1:27:57<48:56,  1.63it/s]
                                                      
{'loss': 0.2457, 'grad_norm': 0.5435838103294373, 'learning_rate': 3.4449465272399574e-06, 'epoch': 1.92}

 64%|██████▍   | 8531/13317 [1:27:57<48:56,  1.63it/s]
 64%|██████▍   | 8532/13317 [1:27:58<48:56,  1.63it/s]
                                                      
{'loss': 0.2063, 'grad_norm': 0.601591169834137, 'learning_rate': 3.4437009445493362e-06, 'epoch': 1.92}

 64%|██████▍   | 8532/13317 [1:27:58<48:56,  1.63it/s]
 64%|██████▍   | 8533/13317 [1:27:58<48:55,  1.63it/s]
                                                      
{'loss': 0.1641, 'grad_norm': 0.4918394386768341, 'learning_rate': 3.44245546879294e-06, 'epoch': 1.92}

 64%|██████▍   | 8533/13317 [1:27:59<48:55,  1.63it/s]
 64%|██████▍   | 8534/13317 [1:27:59<48:53,  1.63it/s]
                                                      
{'loss': 0.1247, 'grad_norm': 0.36344316601753235, 'learning_rate': 3.4412101000563438e-06, 'epoch': 1.92}

 64%|██████▍   | 8534/13317 [1:27:59<48:53,  1.63it/s]
 64%|██████▍   | 8535/13317 [1:28:00<48:54,  1.63it/s]
                                                      
{'loss': 0.2001, 'grad_norm': 0.4942173659801483, 'learning_rate': 3.439964838425116e-06, 'epoch': 1.92}

 64%|██████▍   | 8535/13317 [1:28:00<48:54,  1.63it/s]
 64%|██████▍   | 8536/13317 [1:28:00<48:52,  1.63it/s]
                                                      
{'loss': 0.1696, 'grad_norm': 0.46518123149871826, 'learning_rate': 3.4387196839848246e-06, 'epoch': 1.92}

 64%|██████▍   | 8536/13317 [1:28:00<48:52,  1.63it/s]
 64%|██████▍   | 8537/13317 [1:28:01<48:51,  1.63it/s]
                                                      
{'loss': 0.1724, 'grad_norm': 0.5233288407325745, 'learning_rate': 3.4374746368210214e-06, 'epoch': 1.92}

 64%|██████▍   | 8537/13317 [1:28:01<48:51,  1.63it/s]
 64%|██████▍   | 8538/13317 [1:28:02<48:51,  1.63it/s]
                                                      
{'loss': 0.1873, 'grad_norm': 0.5050017833709717, 'learning_rate': 3.436229697019254e-06, 'epoch': 1.92}

 64%|██████▍   | 8538/13317 [1:28:02<48:51,  1.63it/s]
 64%|██████▍   | 8539/13317 [1:28:02<48:52,  1.63it/s]
                                                      
{'loss': 0.1787, 'grad_norm': 0.5252310633659363, 'learning_rate': 3.4349848646650664e-06, 'epoch': 1.92}

 64%|██████▍   | 8539/13317 [1:28:02<48:52,  1.63it/s]
 64%|██████▍   | 8540/13317 [1:28:03<48:51,  1.63it/s]
                                                      
{'loss': 0.2747, 'grad_norm': 0.6753919124603271, 'learning_rate': 3.4337401398439873e-06, 'epoch': 1.92}

 64%|██████▍   | 8540/13317 [1:28:03<48:51,  1.63it/s]
 64%|██████▍   | 8541/13317 [1:28:03<48:50,  1.63it/s]
                                                      
{'loss': 0.2223, 'grad_norm': 0.6277807950973511, 'learning_rate': 3.432495522641543e-06, 'epoch': 1.92}

 64%|██████▍   | 8541/13317 [1:28:03<48:50,  1.63it/s]
 64%|██████▍   | 8542/13317 [1:28:04<48:51,  1.63it/s]
                                                      
{'loss': 0.1785, 'grad_norm': 0.5349459648132324, 'learning_rate': 3.4312510131432546e-06, 'epoch': 1.92}

 64%|██████▍   | 8542/13317 [1:28:04<48:51,  1.63it/s]
 64%|██████▍   | 8543/13317 [1:28:05<48:48,  1.63it/s]
                                                      
{'loss': 0.1944, 'grad_norm': 0.5429025292396545, 'learning_rate': 3.430006611434633e-06, 'epoch': 1.92}

 64%|██████▍   | 8543/13317 [1:28:05<48:48,  1.63it/s]
 64%|██████▍   | 8544/13317 [1:28:05<48:47,  1.63it/s]
                                                      
{'loss': 0.1835, 'grad_norm': 0.49556204676628113, 'learning_rate': 3.4287623176011796e-06, 'epoch': 1.92}

 64%|██████▍   | 8544/13317 [1:28:05<48:47,  1.63it/s]
 64%|██████▍   | 8545/13317 [1:28:06<48:46,  1.63it/s]
                                                      
{'loss': 0.1812, 'grad_norm': 0.5296453833580017, 'learning_rate': 3.427518131728391e-06, 'epoch': 1.92}

 64%|██████▍   | 8545/13317 [1:28:06<48:46,  1.63it/s]
 64%|██████▍   | 8546/13317 [1:28:06<48:45,  1.63it/s]
                                                      
{'loss': 0.1945, 'grad_norm': 0.49256765842437744, 'learning_rate': 3.426274053901758e-06, 'epoch': 1.93}

 64%|██████▍   | 8546/13317 [1:28:07<48:45,  1.63it/s]
 64%|██████▍   | 8547/13317 [1:28:07<48:47,  1.63it/s]
                                                      
{'loss': 0.1726, 'grad_norm': 0.4963546097278595, 'learning_rate': 3.425030084206759e-06, 'epoch': 1.93}

 64%|██████▍   | 8547/13317 [1:28:07<48:47,  1.63it/s]
 64%|██████▍   | 8548/13317 [1:28:08<48:47,  1.63it/s]
                                                      
{'loss': 0.1397, 'grad_norm': 0.40384772419929504, 'learning_rate': 3.423786222728869e-06, 'epoch': 1.93}

 64%|██████▍   | 8548/13317 [1:28:08<48:47,  1.63it/s]
 64%|██████▍   | 8549/13317 [1:28:08<48:47,  1.63it/s]
                                                      
{'loss': 0.1823, 'grad_norm': 0.4675900936126709, 'learning_rate': 3.4225424695535547e-06, 'epoch': 1.93}

 64%|██████▍   | 8549/13317 [1:28:08<48:47,  1.63it/s]
 64%|██████▍   | 8550/13317 [1:28:09<48:47,  1.63it/s]
                                                      
{'loss': 0.1163, 'grad_norm': 0.4125412702560425, 'learning_rate': 3.4212988247662765e-06, 'epoch': 1.93}

 64%|██████▍   | 8550/13317 [1:28:09<48:47,  1.63it/s]
 64%|██████▍   | 8551/13317 [1:28:09<48:47,  1.63it/s]
                                                      
{'loss': 0.2089, 'grad_norm': 0.504612147808075, 'learning_rate': 3.420055288452484e-06, 'epoch': 1.93}

 64%|██████▍   | 8551/13317 [1:28:10<48:47,  1.63it/s]
 64%|██████▍   | 8552/13317 [1:28:10<48:42,  1.63it/s]
                                                      
{'loss': 0.2034, 'grad_norm': 0.5277624726295471, 'learning_rate': 3.418811860697622e-06, 'epoch': 1.93}

 64%|██████▍   | 8552/13317 [1:28:10<48:42,  1.63it/s]
 64%|██████▍   | 8553/13317 [1:28:11<48:43,  1.63it/s]
                                                      
{'loss': 0.2218, 'grad_norm': 0.6204598546028137, 'learning_rate': 3.4175685415871275e-06, 'epoch': 1.93}

 64%|██████▍   | 8553/13317 [1:28:11<48:43,  1.63it/s]
 64%|██████▍   | 8554/13317 [1:28:11<48:44,  1.63it/s]
                                                      
{'loss': 0.2262, 'grad_norm': 0.6245377063751221, 'learning_rate': 3.416325331206428e-06, 'epoch': 1.93}

 64%|██████▍   | 8554/13317 [1:28:11<48:44,  1.63it/s]
 64%|██████▍   | 8555/13317 [1:28:12<48:42,  1.63it/s]
                                                      
{'loss': 0.2155, 'grad_norm': 0.6428613662719727, 'learning_rate': 3.415082229640946e-06, 'epoch': 1.93}

 64%|██████▍   | 8555/13317 [1:28:12<48:42,  1.63it/s]
 64%|██████▍   | 8556/13317 [1:28:13<48:42,  1.63it/s]
                                                      
{'loss': 0.228, 'grad_norm': 0.5535551905632019, 'learning_rate': 3.4138392369760986e-06, 'epoch': 1.93}

 64%|██████▍   | 8556/13317 [1:28:13<48:42,  1.63it/s]
 64%|██████▍   | 8557/13317 [1:28:13<48:41,  1.63it/s]
                                                      
{'loss': 0.1266, 'grad_norm': 0.40257930755615234, 'learning_rate': 3.4125963532972878e-06, 'epoch': 1.93}

 64%|██████▍   | 8557/13317 [1:28:13<48:41,  1.63it/s]
 64%|██████▍   | 8558/13317 [1:28:14<48:40,  1.63it/s]
                                                      
{'loss': 0.2339, 'grad_norm': 0.5938694477081299, 'learning_rate': 3.411353578689916e-06, 'epoch': 1.93}

 64%|██████▍   | 8558/13317 [1:28:14<48:40,  1.63it/s]
 64%|██████▍   | 8559/13317 [1:28:14<48:38,  1.63it/s]
                                                      
{'loss': 0.2466, 'grad_norm': 0.5777245163917542, 'learning_rate': 3.410110913239374e-06, 'epoch': 1.93}

 64%|██████▍   | 8559/13317 [1:28:15<48:38,  1.63it/s]
 64%|██████▍   | 8560/13317 [1:28:15<48:37,  1.63it/s]
                                                      
{'loss': 0.2393, 'grad_norm': 0.5848665833473206, 'learning_rate': 3.4088683570310475e-06, 'epoch': 1.93}

 64%|██████▍   | 8560/13317 [1:28:15<48:37,  1.63it/s]
 64%|██████▍   | 8561/13317 [1:28:16<48:36,  1.63it/s]
                                                      
{'loss': 0.2305, 'grad_norm': 0.6313157677650452, 'learning_rate': 3.407625910150312e-06, 'epoch': 1.93}

 64%|██████▍   | 8561/13317 [1:28:16<48:36,  1.63it/s]
 64%|██████▍   | 8562/13317 [1:28:16<48:34,  1.63it/s]
                                                      
{'loss': 0.1918, 'grad_norm': 0.6143839955329895, 'learning_rate': 3.4063835726825343e-06, 'epoch': 1.93}

 64%|██████▍   | 8562/13317 [1:28:16<48:34,  1.63it/s]
 64%|██████▍   | 8563/13317 [1:28:17<48:35,  1.63it/s]
                                                      
{'loss': 0.2288, 'grad_norm': 0.6133716106414795, 'learning_rate': 3.405141344713081e-06, 'epoch': 1.93}

 64%|██████▍   | 8563/13317 [1:28:17<48:35,  1.63it/s]
 64%|██████▍   | 8564/13317 [1:28:17<48:36,  1.63it/s]
                                                      
{'loss': 0.2579, 'grad_norm': 0.5954828858375549, 'learning_rate': 3.403899226327303e-06, 'epoch': 1.93}

 64%|██████▍   | 8564/13317 [1:28:18<48:36,  1.63it/s]
 64%|██████▍   | 8565/13317 [1:28:18<48:35,  1.63it/s]
                                                      
{'loss': 0.1512, 'grad_norm': 0.4233104884624481, 'learning_rate': 3.4026572176105483e-06, 'epoch': 1.93}

 64%|██████▍   | 8565/13317 [1:28:18<48:35,  1.63it/s]
 64%|██████▍   | 8566/13317 [1:28:19<48:36,  1.63it/s]
                                                      
{'loss': 0.169, 'grad_norm': 0.5131866335868835, 'learning_rate': 3.4014153186481547e-06, 'epoch': 1.93}

 64%|██████▍   | 8566/13317 [1:28:19<48:36,  1.63it/s]
 64%|██████▍   | 8567/13317 [1:28:19<48:36,  1.63it/s]
                                                      
{'loss': 0.1681, 'grad_norm': 0.48837631940841675, 'learning_rate': 3.4001735295254574e-06, 'epoch': 1.93}

 64%|██████▍   | 8567/13317 [1:28:19<48:36,  1.63it/s]
 64%|██████▍   | 8568/13317 [1:28:20<48:36,  1.63it/s]
                                                      
{'loss': 0.2161, 'grad_norm': 0.5932177901268005, 'learning_rate': 3.3989318503277745e-06, 'epoch': 1.93}

 64%|██████▍   | 8568/13317 [1:28:20<48:36,  1.63it/s]
 64%|██████▍   | 8569/13317 [1:28:21<48:36,  1.63it/s]
                                                      
{'loss': 0.1858, 'grad_norm': 0.5127747058868408, 'learning_rate': 3.3976902811404278e-06, 'epoch': 1.93}

 64%|██████▍   | 8569/13317 [1:28:21<48:36,  1.63it/s]
 64%|██████▍   | 8570/13317 [1:28:21<48:36,  1.63it/s]
                                                      
{'loss': 0.198, 'grad_norm': 0.47824469208717346, 'learning_rate': 3.3964488220487252e-06, 'epoch': 1.93}

 64%|██████▍   | 8570/13317 [1:28:21<48:36,  1.63it/s]
 64%|██████▍   | 8571/13317 [1:28:22<48:35,  1.63it/s]
                                                      
{'loss': 0.2036, 'grad_norm': 0.5219146013259888, 'learning_rate': 3.395207473137966e-06, 'epoch': 1.93}

 64%|██████▍   | 8571/13317 [1:28:22<48:35,  1.63it/s]
 64%|██████▍   | 8572/13317 [1:28:22<48:36,  1.63it/s]
                                                      
{'loss': 0.1941, 'grad_norm': 0.5009335875511169, 'learning_rate': 3.3939662344934447e-06, 'epoch': 1.93}

 64%|██████▍   | 8572/13317 [1:28:23<48:36,  1.63it/s]
 64%|██████▍   | 8573/13317 [1:28:23<48:34,  1.63it/s]
                                                      
{'loss': 0.1653, 'grad_norm': 0.4468216896057129, 'learning_rate': 3.3927251062004494e-06, 'epoch': 1.93}

 64%|██████▍   | 8573/13317 [1:28:23<48:34,  1.63it/s]
 64%|██████▍   | 8574/13317 [1:28:24<48:31,  1.63it/s]
                                                      
{'loss': 0.2182, 'grad_norm': 0.5827550888061523, 'learning_rate': 3.391484088344257e-06, 'epoch': 1.93}

 64%|██████▍   | 8574/13317 [1:28:24<48:31,  1.63it/s]
 64%|██████▍   | 8575/13317 [1:28:24<48:29,  1.63it/s]
                                                      
{'loss': 0.2006, 'grad_norm': 0.49431318044662476, 'learning_rate': 3.3902431810101373e-06, 'epoch': 1.93}

 64%|██████▍   | 8575/13317 [1:28:24<48:29,  1.63it/s]
 64%|██████▍   | 8576/13317 [1:28:25<48:33,  1.63it/s]
                                                      
{'loss': 0.1705, 'grad_norm': 0.4562270939350128, 'learning_rate': 3.3890023842833565e-06, 'epoch': 1.93}

 64%|██████▍   | 8576/13317 [1:28:25<48:33,  1.63it/s]
 64%|██████▍   | 8577/13317 [1:28:25<48:30,  1.63it/s]
                                                      
{'loss': 0.2609, 'grad_norm': 0.5725427865982056, 'learning_rate': 3.38776169824917e-06, 'epoch': 1.93}

 64%|██████▍   | 8577/13317 [1:28:26<48:30,  1.63it/s]
 64%|██████▍   | 8578/13317 [1:28:26<48:29,  1.63it/s]
                                                      
{'loss': 0.2174, 'grad_norm': 0.5950865745544434, 'learning_rate': 3.3865211229928254e-06, 'epoch': 1.93}

 64%|██████▍   | 8578/13317 [1:28:26<48:29,  1.63it/s]
 64%|██████▍   | 8579/13317 [1:28:27<48:31,  1.63it/s]
                                                      
{'loss': 0.18, 'grad_norm': 0.46504461765289307, 'learning_rate': 3.385280658599563e-06, 'epoch': 1.93}

 64%|██████▍   | 8579/13317 [1:28:27<48:31,  1.63it/s]
 64%|██████▍   | 8580/13317 [1:28:27<48:28,  1.63it/s]
                                                      
{'loss': 0.1957, 'grad_norm': 0.46299612522125244, 'learning_rate': 3.384040305154617e-06, 'epoch': 1.93}

 64%|██████▍   | 8580/13317 [1:28:27<48:28,  1.63it/s]
 64%|██████▍   | 8581/13317 [1:28:28<48:35,  1.62it/s]
                                                      
{'loss': 0.1824, 'grad_norm': 0.4724891483783722, 'learning_rate': 3.3828000627432117e-06, 'epoch': 1.93}

 64%|██████▍   | 8581/13317 [1:28:28<48:35,  1.62it/s]
 64%|██████▍   | 8582/13317 [1:28:29<48:33,  1.63it/s]
                                                      
{'loss': 0.1902, 'grad_norm': 0.4564947783946991, 'learning_rate': 3.381559931450564e-06, 'epoch': 1.93}

 64%|██████▍   | 8582/13317 [1:28:29<48:33,  1.63it/s]
 64%|██████▍   | 8583/13317 [1:28:29<48:30,  1.63it/s]
                                                      
{'loss': 0.197, 'grad_norm': 0.5470547676086426, 'learning_rate': 3.380319911361888e-06, 'epoch': 1.93}

 64%|██████▍   | 8583/13317 [1:28:29<48:30,  1.63it/s]
 64%|██████▍   | 8584/13317 [1:28:30<48:28,  1.63it/s]
                                                      
{'loss': 0.1971, 'grad_norm': 0.5616493225097656, 'learning_rate': 3.379080002562381e-06, 'epoch': 1.93}

 64%|██████▍   | 8584/13317 [1:28:30<48:28,  1.63it/s]
 64%|██████▍   | 8585/13317 [1:28:30<48:27,  1.63it/s]
                                                      
{'loss': 0.2908, 'grad_norm': 0.6082634925842285, 'learning_rate': 3.377840205137241e-06, 'epoch': 1.93}

 64%|██████▍   | 8585/13317 [1:28:30<48:27,  1.63it/s]
 64%|██████▍   | 8586/13317 [1:28:31<48:24,  1.63it/s]
                                                      
{'loss': 0.2314, 'grad_norm': 0.5812414288520813, 'learning_rate': 3.376600519171655e-06, 'epoch': 1.93}

 64%|██████▍   | 8586/13317 [1:28:31<48:24,  1.63it/s]
 64%|██████▍   | 8587/13317 [1:28:32<48:23,  1.63it/s]
                                                      
{'loss': 0.1987, 'grad_norm': 0.4836063086986542, 'learning_rate': 3.375360944750803e-06, 'epoch': 1.93}

 64%|██████▍   | 8587/13317 [1:28:32<48:23,  1.63it/s]
 64%|██████▍   | 8588/13317 [1:28:32<48:26,  1.63it/s]
                                                      
{'loss': 0.2006, 'grad_norm': 0.48237648606300354, 'learning_rate': 3.374121481959854e-06, 'epoch': 1.93}

 64%|██████▍   | 8588/13317 [1:28:32<48:26,  1.63it/s]
 64%|██████▍   | 8589/13317 [1:28:33<48:25,  1.63it/s]
                                                      
{'loss': 0.2683, 'grad_norm': 0.6331433653831482, 'learning_rate': 3.3728821308839756e-06, 'epoch': 1.93}

 64%|██████▍   | 8589/13317 [1:28:33<48:25,  1.63it/s]
 65%|██████▍   | 8590/13317 [1:28:33<48:24,  1.63it/s]
                                                      
{'loss': 0.1797, 'grad_norm': 0.4653623700141907, 'learning_rate': 3.3716428916083233e-06, 'epoch': 1.94}

 65%|██████▍   | 8590/13317 [1:28:34<48:24,  1.63it/s]
 65%|██████▍   | 8591/13317 [1:28:34<48:24,  1.63it/s]
                                                      
{'loss': 0.2026, 'grad_norm': 0.5488123297691345, 'learning_rate': 3.3704037642180454e-06, 'epoch': 1.94}

 65%|██████▍   | 8591/13317 [1:28:34<48:24,  1.63it/s]
 65%|██████▍   | 8592/13317 [1:28:35<48:21,  1.63it/s]
                                                      
{'loss': 0.2403, 'grad_norm': 0.7504948377609253, 'learning_rate': 3.3691647487982827e-06, 'epoch': 1.94}

 65%|██████▍   | 8592/13317 [1:28:35<48:21,  1.63it/s]
 65%|██████▍   | 8593/13317 [1:28:35<48:26,  1.63it/s]
                                                      
{'loss': 0.1728, 'grad_norm': 0.4604358971118927, 'learning_rate': 3.367925845434169e-06, 'epoch': 1.94}

 65%|██████▍   | 8593/13317 [1:28:35<48:26,  1.63it/s]
 65%|██████▍   | 8594/13317 [1:28:36<48:24,  1.63it/s]
                                                      
{'loss': 0.4769, 'grad_norm': 0.6966660618782043, 'learning_rate': 3.366687054210833e-06, 'epoch': 1.94}

 65%|██████▍   | 8594/13317 [1:28:36<48:24,  1.63it/s]
 65%|██████▍   | 8595/13317 [1:28:37<48:25,  1.63it/s]
                                                      
{'loss': 0.1743, 'grad_norm': 0.5108683705329895, 'learning_rate': 3.365448375213387e-06, 'epoch': 1.94}

 65%|██████▍   | 8595/13317 [1:28:37<48:25,  1.63it/s]
 65%|██████▍   | 8596/13317 [1:28:37<48:23,  1.63it/s]
                                                      
{'loss': 0.1852, 'grad_norm': 0.5781875252723694, 'learning_rate': 3.364209808526946e-06, 'epoch': 1.94}

 65%|██████▍   | 8596/13317 [1:28:37<48:23,  1.63it/s]
 65%|██████▍   | 8597/13317 [1:28:38<48:22,  1.63it/s]
                                                      
{'loss': 0.2072, 'grad_norm': 0.7386845946311951, 'learning_rate': 3.3629713542366118e-06, 'epoch': 1.94}

 65%|██████▍   | 8597/13317 [1:28:38<48:22,  1.63it/s]
 65%|██████▍   | 8598/13317 [1:28:38<48:22,  1.63it/s]
                                                      
{'loss': 0.2289, 'grad_norm': 0.5804474353790283, 'learning_rate': 3.361733012427478e-06, 'epoch': 1.94}

 65%|██████▍   | 8598/13317 [1:28:38<48:22,  1.63it/s]
 65%|██████▍   | 8599/13317 [1:28:39<48:20,  1.63it/s]
                                                      
{'loss': 0.1594, 'grad_norm': 0.4549613296985626, 'learning_rate': 3.3604947831846325e-06, 'epoch': 1.94}

 65%|██████▍   | 8599/13317 [1:28:39<48:20,  1.63it/s]
 65%|██████▍   | 8600/13317 [1:28:40<48:20,  1.63it/s]
                                                      
{'loss': 0.1814, 'grad_norm': 0.507834792137146, 'learning_rate': 3.3592566665931555e-06, 'epoch': 1.94}

 65%|██████▍   | 8600/13317 [1:28:40<48:20,  1.63it/s]
 65%|██████▍   | 8601/13317 [1:28:40<49:40,  1.58it/s]
                                                      
{'loss': 0.1931, 'grad_norm': 0.48596885800361633, 'learning_rate': 3.358018662738117e-06, 'epoch': 1.94}

 65%|██████▍   | 8601/13317 [1:28:40<49:40,  1.58it/s]
 65%|██████▍   | 8602/13317 [1:28:41<49:13,  1.60it/s]
                                                      
{'loss': 0.1848, 'grad_norm': 0.46913832426071167, 'learning_rate': 3.3567807717045807e-06, 'epoch': 1.94}

 65%|██████▍   | 8602/13317 [1:28:41<49:13,  1.60it/s]
 65%|██████▍   | 8603/13317 [1:28:41<48:59,  1.60it/s]
                                                      
{'loss': 0.1515, 'grad_norm': 0.42290470004081726, 'learning_rate': 3.355542993577605e-06, 'epoch': 1.94}

 65%|██████▍   | 8603/13317 [1:28:42<48:59,  1.60it/s]
 65%|██████▍   | 8604/13317 [1:28:42<48:43,  1.61it/s]
                                                      
{'loss': 0.1916, 'grad_norm': 0.507904589176178, 'learning_rate': 3.354305328442239e-06, 'epoch': 1.94}

 65%|██████▍   | 8604/13317 [1:28:42<48:43,  1.61it/s]
 65%|██████▍   | 8605/13317 [1:28:43<48:37,  1.62it/s]
                                                      
{'loss': 0.1583, 'grad_norm': 0.4506308138370514, 'learning_rate': 3.3530677763835206e-06, 'epoch': 1.94}

 65%|██████▍   | 8605/13317 [1:28:43<48:37,  1.62it/s]
 65%|██████▍   | 8606/13317 [1:28:43<48:28,  1.62it/s]
                                                      
{'loss': 0.2121, 'grad_norm': 0.5401188135147095, 'learning_rate': 3.351830337486484e-06, 'epoch': 1.94}

 65%|██████▍   | 8606/13317 [1:28:43<48:28,  1.62it/s]
 65%|██████▍   | 8607/13317 [1:28:44<48:22,  1.62it/s]
                                                      
{'loss': 0.1535, 'grad_norm': 0.4561203718185425, 'learning_rate': 3.3505930118361552e-06, 'epoch': 1.94}

 65%|██████▍   | 8607/13317 [1:28:44<48:22,  1.62it/s]
 65%|██████▍   | 8608/13317 [1:28:45<48:18,  1.62it/s]
                                                      
{'loss': 0.1701, 'grad_norm': 0.45570096373558044, 'learning_rate': 3.34935579951755e-06, 'epoch': 1.94}

 65%|██████▍   | 8608/13317 [1:28:45<48:18,  1.62it/s]
 65%|██████▍   | 8609/13317 [1:28:45<48:14,  1.63it/s]
                                                      
{'loss': 0.2275, 'grad_norm': 0.5592678785324097, 'learning_rate': 3.348118700615677e-06, 'epoch': 1.94}

 65%|██████▍   | 8609/13317 [1:28:45<48:14,  1.63it/s]
 65%|██████▍   | 8610/13317 [1:28:46<48:17,  1.62it/s]
                                                      
{'loss': 0.223, 'grad_norm': 0.6050535440444946, 'learning_rate': 3.346881715215542e-06, 'epoch': 1.94}

 65%|██████▍   | 8610/13317 [1:28:46<48:17,  1.62it/s]
 65%|██████▍   | 8611/13317 [1:28:46<48:13,  1.63it/s]
                                                      
{'loss': 0.1845, 'grad_norm': 0.4651741683483124, 'learning_rate': 3.345644843402136e-06, 'epoch': 1.94}

 65%|██████▍   | 8611/13317 [1:28:47<48:13,  1.63it/s]
 65%|██████▍   | 8612/13317 [1:28:47<48:11,  1.63it/s]
                                                      
{'loss': 0.1918, 'grad_norm': 0.4948730766773224, 'learning_rate': 3.3444080852604462e-06, 'epoch': 1.94}

 65%|██████▍   | 8612/13317 [1:28:47<48:11,  1.63it/s]
 65%|██████▍   | 8613/13317 [1:28:48<48:10,  1.63it/s]
                                                      
{'loss': 0.2074, 'grad_norm': 0.554137110710144, 'learning_rate': 3.3431714408754502e-06, 'epoch': 1.94}

 65%|██████▍   | 8613/13317 [1:28:48<48:10,  1.63it/s]
 65%|██████▍   | 8614/13317 [1:28:48<48:07,  1.63it/s]
                                                      
{'loss': 0.2073, 'grad_norm': 0.5244306921958923, 'learning_rate': 3.34193491033212e-06, 'epoch': 1.94}

 65%|██████▍   | 8614/13317 [1:28:48<48:07,  1.63it/s]
 65%|██████▍   | 8615/13317 [1:28:49<48:07,  1.63it/s]
                                                      
{'loss': 0.1804, 'grad_norm': 0.4825620949268341, 'learning_rate': 3.3406984937154146e-06, 'epoch': 1.94}

 65%|██████▍   | 8615/13317 [1:28:49<48:07,  1.63it/s]
 65%|██████▍   | 8616/13317 [1:28:49<48:06,  1.63it/s]
                                                      
{'loss': 0.2065, 'grad_norm': 0.4854905903339386, 'learning_rate': 3.339462191110293e-06, 'epoch': 1.94}

 65%|██████▍   | 8616/13317 [1:28:50<48:06,  1.63it/s]
 65%|██████▍   | 8617/13317 [1:28:50<48:06,  1.63it/s]
                                                      
{'loss': 0.1538, 'grad_norm': 0.46457818150520325, 'learning_rate': 3.3382260026017027e-06, 'epoch': 1.94}

 65%|██████▍   | 8617/13317 [1:28:50<48:06,  1.63it/s]
 65%|██████▍   | 8618/13317 [1:28:51<48:03,  1.63it/s]
                                                      
{'loss': 0.1646, 'grad_norm': 0.48022887110710144, 'learning_rate': 3.33698992827458e-06, 'epoch': 1.94}

 65%|██████▍   | 8618/13317 [1:28:51<48:03,  1.63it/s]
 65%|██████▍   | 8619/13317 [1:28:51<48:03,  1.63it/s]
                                                      
{'loss': 0.2144, 'grad_norm': 0.5130760073661804, 'learning_rate': 3.335753968213857e-06, 'epoch': 1.94}

 65%|██████▍   | 8619/13317 [1:28:51<48:03,  1.63it/s]
 65%|██████▍   | 8620/13317 [1:28:52<48:06,  1.63it/s]
                                                      
{'loss': 0.2463, 'grad_norm': 0.6572377681732178, 'learning_rate': 3.3345181225044563e-06, 'epoch': 1.94}

 65%|██████▍   | 8620/13317 [1:28:52<48:06,  1.63it/s]
 65%|██████▍   | 8621/13317 [1:28:53<48:05,  1.63it/s]
                                                      
{'loss': 0.1841, 'grad_norm': 0.5376429557800293, 'learning_rate': 3.3332823912312985e-06, 'epoch': 1.94}

 65%|██████▍   | 8621/13317 [1:28:53<48:05,  1.63it/s]
 65%|██████▍   | 8622/13317 [1:28:53<48:03,  1.63it/s]
                                                      
{'loss': 0.1521, 'grad_norm': 0.4348663091659546, 'learning_rate': 3.332046774479285e-06, 'epoch': 1.94}

 65%|██████▍   | 8622/13317 [1:28:53<48:03,  1.63it/s]
 65%|██████▍   | 8623/13317 [1:28:54<48:02,  1.63it/s]
                                                      
{'loss': 0.2128, 'grad_norm': 0.5429885387420654, 'learning_rate': 3.3308112723333195e-06, 'epoch': 1.94}

 65%|██████▍   | 8623/13317 [1:28:54<48:02,  1.63it/s]
 65%|██████▍   | 8624/13317 [1:28:54<48:00,  1.63it/s]
                                                      
{'loss': 0.1681, 'grad_norm': 0.5021315217018127, 'learning_rate': 3.3295758848782944e-06, 'epoch': 1.94}

 65%|██████▍   | 8624/13317 [1:28:55<48:00,  1.63it/s]
 65%|██████▍   | 8625/13317 [1:28:55<48:00,  1.63it/s]
                                                      
{'loss': 0.1315, 'grad_norm': 0.4150716960430145, 'learning_rate': 3.328340612199091e-06, 'epoch': 1.94}

 65%|██████▍   | 8625/13317 [1:28:55<48:00,  1.63it/s]
 65%|██████▍   | 8626/13317 [1:28:56<47:59,  1.63it/s]
                                                      
{'loss': 0.145, 'grad_norm': 0.46151405572891235, 'learning_rate': 3.327105454380589e-06, 'epoch': 1.94}

 65%|██████▍   | 8626/13317 [1:28:56<47:59,  1.63it/s]
 65%|██████▍   | 8627/13317 [1:28:56<47:57,  1.63it/s]
                                                      
{'loss': 0.1765, 'grad_norm': 0.47868913412094116, 'learning_rate': 3.3258704115076545e-06, 'epoch': 1.94}

 65%|██████▍   | 8627/13317 [1:28:56<47:57,  1.63it/s]
 65%|██████▍   | 8628/13317 [1:28:57<47:57,  1.63it/s]
                                                      
{'loss': 0.1705, 'grad_norm': 0.4884660243988037, 'learning_rate': 3.3246354836651483e-06, 'epoch': 1.94}

 65%|██████▍   | 8628/13317 [1:28:57<47:57,  1.63it/s]
 65%|██████▍   | 8629/13317 [1:28:57<47:55,  1.63it/s]
                                                      
{'loss': 0.1513, 'grad_norm': 0.4694434106349945, 'learning_rate': 3.323400670937923e-06, 'epoch': 1.94}

 65%|██████▍   | 8629/13317 [1:28:58<47:55,  1.63it/s]
 65%|██████▍   | 8630/13317 [1:28:58<47:59,  1.63it/s]
                                                      
{'loss': 0.131, 'grad_norm': 0.38679972290992737, 'learning_rate': 3.322165973410824e-06, 'epoch': 1.94}

 65%|██████▍   | 8630/13317 [1:28:58<47:59,  1.63it/s]
 65%|██████▍   | 8631/13317 [1:28:59<47:57,  1.63it/s]
                                                      
{'loss': 0.176, 'grad_norm': 0.5460016131401062, 'learning_rate': 3.320931391168689e-06, 'epoch': 1.94}

 65%|██████▍   | 8631/13317 [1:28:59<47:57,  1.63it/s]
 65%|██████▍   | 8632/13317 [1:28:59<47:54,  1.63it/s]
                                                      
{'loss': 0.1603, 'grad_norm': 0.4448426365852356, 'learning_rate': 3.319696924296345e-06, 'epoch': 1.94}

 65%|██████▍   | 8632/13317 [1:28:59<47:54,  1.63it/s]
 65%|██████▍   | 8633/13317 [1:29:00<47:54,  1.63it/s]
                                                      
{'loss': 0.2625, 'grad_norm': 0.6370447278022766, 'learning_rate': 3.3184625728786136e-06, 'epoch': 1.94}

 65%|██████▍   | 8633/13317 [1:29:00<47:54,  1.63it/s]
 65%|██████▍   | 8634/13317 [1:29:01<47:56,  1.63it/s]
                                                      
{'loss': 0.2148, 'grad_norm': 0.7564237713813782, 'learning_rate': 3.3172283370003083e-06, 'epoch': 1.95}

 65%|██████▍   | 8634/13317 [1:29:01<47:56,  1.63it/s]
 65%|██████▍   | 8635/13317 [1:29:01<48:02,  1.62it/s]
                                                      
{'loss': 0.2429, 'grad_norm': 0.729333221912384, 'learning_rate': 3.3159942167462334e-06, 'epoch': 1.95}

 65%|██████▍   | 8635/13317 [1:29:01<48:02,  1.62it/s]
 65%|██████▍   | 8636/13317 [1:29:02<48:01,  1.62it/s]
                                                      
{'loss': 0.1563, 'grad_norm': 0.4765336513519287, 'learning_rate': 3.3147602122011844e-06, 'epoch': 1.95}

 65%|██████▍   | 8636/13317 [1:29:02<48:01,  1.62it/s]
 65%|██████▍   | 8637/13317 [1:29:02<48:02,  1.62it/s]
                                                      
{'loss': 0.1581, 'grad_norm': 0.4629310965538025, 'learning_rate': 3.3135263234499545e-06, 'epoch': 1.95}

 65%|██████▍   | 8637/13317 [1:29:03<48:02,  1.62it/s]
 65%|██████▍   | 8638/13317 [1:29:03<47:58,  1.63it/s]
                                                      
{'loss': 0.1731, 'grad_norm': 0.5355927348136902, 'learning_rate': 3.3122925505773227e-06, 'epoch': 1.95}

 65%|██████▍   | 8638/13317 [1:29:03<47:58,  1.63it/s]
 65%|██████▍   | 8639/13317 [1:29:04<47:55,  1.63it/s]
                                                      
{'loss': 0.255, 'grad_norm': 0.6620864272117615, 'learning_rate': 3.3110588936680623e-06, 'epoch': 1.95}

 65%|██████▍   | 8639/13317 [1:29:04<47:55,  1.63it/s]
 65%|██████▍   | 8640/13317 [1:29:04<47:58,  1.62it/s]
                                                      
{'loss': 0.1489, 'grad_norm': 0.4714154303073883, 'learning_rate': 3.309825352806939e-06, 'epoch': 1.95}

 65%|██████▍   | 8640/13317 [1:29:04<47:58,  1.62it/s]
 65%|██████▍   | 8641/13317 [1:29:05<47:55,  1.63it/s]
                                                      
{'loss': 0.1606, 'grad_norm': 0.5431163311004639, 'learning_rate': 3.30859192807871e-06, 'epoch': 1.95}

 65%|██████▍   | 8641/13317 [1:29:05<47:55,  1.63it/s]
 65%|██████▍   | 8642/13317 [1:29:05<47:52,  1.63it/s]
                                                      
{'loss': 0.1776, 'grad_norm': 0.5202245712280273, 'learning_rate': 3.307358619568123e-06, 'epoch': 1.95}

 65%|██████▍   | 8642/13317 [1:29:06<47:52,  1.63it/s]
 65%|██████▍   | 8643/13317 [1:29:06<47:49,  1.63it/s]
                                                      
{'loss': 0.1246, 'grad_norm': 0.38266900181770325, 'learning_rate': 3.3061254273599212e-06, 'epoch': 1.95}

 65%|██████▍   | 8643/13317 [1:29:06<47:49,  1.63it/s]
 65%|██████▍   | 8644/13317 [1:29:07<47:47,  1.63it/s]
                                                      
{'loss': 0.1494, 'grad_norm': 0.45803317427635193, 'learning_rate': 3.3048923515388385e-06, 'epoch': 1.95}

 65%|██████▍   | 8644/13317 [1:29:07<47:47,  1.63it/s]
 65%|██████▍   | 8645/13317 [1:29:07<48:05,  1.62it/s]
                                                      
{'loss': 0.1786, 'grad_norm': 0.5938267111778259, 'learning_rate': 3.3036593921895993e-06, 'epoch': 1.95}

 65%|██████▍   | 8645/13317 [1:29:07<48:05,  1.62it/s]
 65%|██████▍   | 8646/13317 [1:29:08<47:59,  1.62it/s]
                                                      
{'loss': 0.1772, 'grad_norm': 0.5242406129837036, 'learning_rate': 3.3024265493969205e-06, 'epoch': 1.95}

 65%|██████▍   | 8646/13317 [1:29:08<47:59,  1.62it/s]
 65%|██████▍   | 8647/13317 [1:29:09<47:55,  1.62it/s]
                                                      
{'loss': 0.194, 'grad_norm': 0.5348848700523376, 'learning_rate': 3.3011938232455108e-06, 'epoch': 1.95}

 65%|██████▍   | 8647/13317 [1:29:09<47:55,  1.62it/s]
 65%|██████▍   | 8648/13317 [1:29:09<47:51,  1.63it/s]
                                                      
{'loss': 0.1598, 'grad_norm': 0.41434788703918457, 'learning_rate': 3.299961213820076e-06, 'epoch': 1.95}

 65%|██████▍   | 8648/13317 [1:29:09<47:51,  1.63it/s]
 65%|██████▍   | 8649/13317 [1:29:10<47:49,  1.63it/s]
                                                      
{'loss': 0.1711, 'grad_norm': 0.5000318884849548, 'learning_rate': 3.2987287212053028e-06, 'epoch': 1.95}

 65%|██████▍   | 8649/13317 [1:29:10<47:49,  1.63it/s]
 65%|██████▍   | 8650/13317 [1:29:10<47:52,  1.62it/s]
                                                      
{'loss': 0.1747, 'grad_norm': 0.4690450429916382, 'learning_rate': 3.297496345485881e-06, 'epoch': 1.95}

 65%|██████▍   | 8650/13317 [1:29:10<47:52,  1.62it/s]
 65%|██████▍   | 8651/13317 [1:29:11<47:48,  1.63it/s]
                                                      
{'loss': 0.1764, 'grad_norm': 0.5260430574417114, 'learning_rate': 3.2962640867464872e-06, 'epoch': 1.95}

 65%|██████▍   | 8651/13317 [1:29:11<47:48,  1.63it/s]
 65%|██████▍   | 8652/13317 [1:29:12<47:46,  1.63it/s]
                                                      
{'loss': 0.1831, 'grad_norm': 0.5455065965652466, 'learning_rate': 3.2950319450717894e-06, 'epoch': 1.95}

 65%|██████▍   | 8652/13317 [1:29:12<47:46,  1.63it/s]
 65%|██████▍   | 8653/13317 [1:29:12<47:49,  1.63it/s]
                                                      
{'loss': 0.1621, 'grad_norm': 0.45390060544013977, 'learning_rate': 3.29379992054645e-06, 'epoch': 1.95}

 65%|██████▍   | 8653/13317 [1:29:12<47:49,  1.63it/s]
 65%|██████▍   | 8654/13317 [1:29:13<47:44,  1.63it/s]
                                                      
{'loss': 0.2003, 'grad_norm': 0.5784905552864075, 'learning_rate': 3.2925680132551195e-06, 'epoch': 1.95}

 65%|██████▍   | 8654/13317 [1:29:13<47:44,  1.63it/s]
 65%|██████▍   | 8655/13317 [1:29:13<48:02,  1.62it/s]
                                                      
{'loss': 0.2462, 'grad_norm': 0.5990914106369019, 'learning_rate': 3.291336223282449e-06, 'epoch': 1.95}

 65%|██████▍   | 8655/13317 [1:29:14<48:02,  1.62it/s]
 65%|██████▍   | 8656/13317 [1:29:14<47:55,  1.62it/s]
                                                      
{'loss': 0.1799, 'grad_norm': 0.4792625308036804, 'learning_rate': 3.290104550713068e-06, 'epoch': 1.95}

 65%|██████▍   | 8656/13317 [1:29:14<47:55,  1.62it/s]
 65%|██████▌   | 8657/13317 [1:29:15<47:48,  1.62it/s]
                                                      
{'loss': 0.2382, 'grad_norm': 0.5872822999954224, 'learning_rate': 3.288872995631611e-06, 'epoch': 1.95}

 65%|██████▌   | 8657/13317 [1:29:15<47:48,  1.62it/s]
 65%|██████▌   | 8658/13317 [1:29:15<47:50,  1.62it/s]
                                                      
{'loss': 0.1755, 'grad_norm': 0.4667012691497803, 'learning_rate': 3.2876415581226972e-06, 'epoch': 1.95}

 65%|██████▌   | 8658/13317 [1:29:15<47:50,  1.62it/s]
 65%|██████▌   | 8659/13317 [1:29:16<47:46,  1.63it/s]
                                                      
{'loss': 0.1736, 'grad_norm': 0.5654725432395935, 'learning_rate': 3.286410238270938e-06, 'epoch': 1.95}

 65%|██████▌   | 8659/13317 [1:29:16<47:46,  1.63it/s]
 65%|██████▌   | 8660/13317 [1:29:17<47:56,  1.62it/s]
                                                      
{'loss': 0.2, 'grad_norm': 0.4899995028972626, 'learning_rate': 3.2851790361609393e-06, 'epoch': 1.95}

 65%|██████▌   | 8660/13317 [1:29:17<47:56,  1.62it/s]
 65%|██████▌   | 8661/13317 [1:29:17<47:50,  1.62it/s]
                                                      
{'loss': 0.2173, 'grad_norm': 0.5010257363319397, 'learning_rate': 3.2839479518772987e-06, 'epoch': 1.95}

 65%|██████▌   | 8661/13317 [1:29:17<47:50,  1.62it/s]
 65%|██████▌   | 8662/13317 [1:29:18<47:45,  1.62it/s]
                                                      
{'loss': 0.2248, 'grad_norm': 0.5864254832267761, 'learning_rate': 3.2827169855046025e-06, 'epoch': 1.95}

 65%|██████▌   | 8662/13317 [1:29:18<47:45,  1.62it/s]
 65%|██████▌   | 8663/13317 [1:29:18<47:44,  1.62it/s]
                                                      
{'loss': 0.1412, 'grad_norm': 0.3972235918045044, 'learning_rate': 3.281486137127431e-06, 'epoch': 1.95}

 65%|██████▌   | 8663/13317 [1:29:19<47:44,  1.62it/s]
 65%|██████▌   | 8664/13317 [1:29:19<47:41,  1.63it/s]
                                                      
{'loss': 0.2297, 'grad_norm': 0.5926198959350586, 'learning_rate': 3.2802554068303595e-06, 'epoch': 1.95}

 65%|██████▌   | 8664/13317 [1:29:19<47:41,  1.63it/s]
 65%|██████▌   | 8665/13317 [1:29:20<47:40,  1.63it/s]
                                                      
{'loss': 0.2708, 'grad_norm': 0.5784163475036621, 'learning_rate': 3.2790247946979516e-06, 'epoch': 1.95}

 65%|██████▌   | 8665/13317 [1:29:20<47:40,  1.63it/s]
 65%|██████▌   | 8666/13317 [1:29:20<47:41,  1.63it/s]
                                                      
{'loss': 0.1783, 'grad_norm': 0.46878480911254883, 'learning_rate': 3.2777943008147604e-06, 'epoch': 1.95}

 65%|██████▌   | 8666/13317 [1:29:20<47:41,  1.63it/s]
 65%|██████▌   | 8667/13317 [1:29:21<47:38,  1.63it/s]
                                                      
{'loss': 0.165, 'grad_norm': 0.4825177788734436, 'learning_rate': 3.2765639252653348e-06, 'epoch': 1.95}

 65%|██████▌   | 8667/13317 [1:29:21<47:38,  1.63it/s]
 65%|██████▌   | 8668/13317 [1:29:21<47:39,  1.63it/s]
                                                      
{'loss': 0.1839, 'grad_norm': 0.5164515972137451, 'learning_rate': 3.2753336681342195e-06, 'epoch': 1.95}

 65%|██████▌   | 8668/13317 [1:29:22<47:39,  1.63it/s]
 65%|██████▌   | 8669/13317 [1:29:22<47:42,  1.62it/s]
                                                      
{'loss': 0.1592, 'grad_norm': 0.4582112729549408, 'learning_rate': 3.274103529505939e-06, 'epoch': 1.95}

 65%|██████▌   | 8669/13317 [1:29:22<47:42,  1.62it/s]
 65%|██████▌   | 8670/13317 [1:29:23<47:43,  1.62it/s]
                                                      
{'loss': 0.1646, 'grad_norm': 0.485221266746521, 'learning_rate': 3.2728735094650206e-06, 'epoch': 1.95}

 65%|██████▌   | 8670/13317 [1:29:23<47:43,  1.62it/s]
 65%|██████▌   | 8671/13317 [1:29:23<47:41,  1.62it/s]
                                                      
{'loss': 0.1732, 'grad_norm': 0.5628772377967834, 'learning_rate': 3.271643608095979e-06, 'epoch': 1.95}

 65%|██████▌   | 8671/13317 [1:29:23<47:41,  1.62it/s]
 65%|██████▌   | 8672/13317 [1:29:24<47:38,  1.63it/s]
                                                      
{'loss': 0.2096, 'grad_norm': 0.5233174562454224, 'learning_rate': 3.2704138254833235e-06, 'epoch': 1.95}

 65%|██████▌   | 8672/13317 [1:29:24<47:38,  1.63it/s]
 65%|██████▌   | 8673/13317 [1:29:25<47:37,  1.63it/s]
                                                      
{'loss': 0.2024, 'grad_norm': 0.5185497403144836, 'learning_rate': 3.269184161711549e-06, 'epoch': 1.95}

 65%|██████▌   | 8673/13317 [1:29:25<47:37,  1.63it/s]
 65%|██████▌   | 8674/13317 [1:29:25<47:34,  1.63it/s]
                                                      
{'loss': 0.2187, 'grad_norm': 0.5420189499855042, 'learning_rate': 3.2679546168651493e-06, 'epoch': 1.95}

 65%|██████▌   | 8674/13317 [1:29:25<47:34,  1.63it/s]
 65%|██████▌   | 8675/13317 [1:29:26<47:35,  1.63it/s]
                                                      
{'loss': 0.1545, 'grad_norm': 0.4326602518558502, 'learning_rate': 3.266725191028609e-06, 'epoch': 1.95}

 65%|██████▌   | 8675/13317 [1:29:26<47:35,  1.63it/s]
 65%|██████▌   | 8676/13317 [1:29:26<47:34,  1.63it/s]
                                                      
{'loss': 0.1782, 'grad_norm': 0.5689724087715149, 'learning_rate': 3.265495884286397e-06, 'epoch': 1.95}

 65%|██████▌   | 8676/13317 [1:29:27<47:34,  1.63it/s]
 65%|██████▌   | 8677/13317 [1:29:27<47:33,  1.63it/s]
                                                      
{'loss': 0.2505, 'grad_norm': 0.5403818488121033, 'learning_rate': 3.2642666967229847e-06, 'epoch': 1.95}

 65%|██████▌   | 8677/13317 [1:29:27<47:33,  1.63it/s]
 65%|██████▌   | 8678/13317 [1:29:28<47:33,  1.63it/s]
                                                      
{'loss': 0.2518, 'grad_norm': 0.6472939848899841, 'learning_rate': 3.263037628422829e-06, 'epoch': 1.95}

 65%|██████▌   | 8678/13317 [1:29:28<47:33,  1.63it/s]
 65%|██████▌   | 8679/13317 [1:29:28<47:32,  1.63it/s]
                                                      
{'loss': 0.2672, 'grad_norm': 0.7488575577735901, 'learning_rate': 3.2618086794703787e-06, 'epoch': 1.96}

 65%|██████▌   | 8679/13317 [1:29:28<47:32,  1.63it/s]
 65%|██████▌   | 8680/13317 [1:29:29<47:33,  1.63it/s]
                                                      
{'loss': 0.1605, 'grad_norm': 0.47007474303245544, 'learning_rate': 3.2605798499500764e-06, 'epoch': 1.96}

 65%|██████▌   | 8680/13317 [1:29:29<47:33,  1.63it/s]
 65%|██████▌   | 8681/13317 [1:29:29<47:29,  1.63it/s]
                                                      
{'loss': 0.3092, 'grad_norm': 0.609458863735199, 'learning_rate': 3.2593511399463547e-06, 'epoch': 1.96}

 65%|██████▌   | 8681/13317 [1:29:30<47:29,  1.63it/s]
 65%|██████▌   | 8682/13317 [1:29:30<47:28,  1.63it/s]
                                                      
{'loss': 0.1549, 'grad_norm': 0.41854822635650635, 'learning_rate': 3.258122549543643e-06, 'epoch': 1.96}

 65%|██████▌   | 8682/13317 [1:29:30<47:28,  1.63it/s]
 65%|██████▌   | 8683/13317 [1:29:31<47:29,  1.63it/s]
                                                      
{'loss': 0.1767, 'grad_norm': 0.5060117244720459, 'learning_rate': 3.256894078826352e-06, 'epoch': 1.96}

 65%|██████▌   | 8683/13317 [1:29:31<47:29,  1.63it/s]
 65%|██████▌   | 8684/13317 [1:29:31<47:26,  1.63it/s]
                                                      
{'loss': 0.192, 'grad_norm': 0.573695957660675, 'learning_rate': 3.255665727878895e-06, 'epoch': 1.96}

 65%|██████▌   | 8684/13317 [1:29:31<47:26,  1.63it/s]
 65%|██████▌   | 8685/13317 [1:29:32<47:37,  1.62it/s]
                                                      
{'loss': 0.1731, 'grad_norm': 0.5467793345451355, 'learning_rate': 3.254437496785674e-06, 'epoch': 1.96}

 65%|██████▌   | 8685/13317 [1:29:32<47:37,  1.62it/s]
 65%|██████▌   | 8686/13317 [1:29:33<47:33,  1.62it/s]
                                                      
{'loss': 0.2219, 'grad_norm': 0.5839321613311768, 'learning_rate': 3.253209385631077e-06, 'epoch': 1.96}

 65%|██████▌   | 8686/13317 [1:29:33<47:33,  1.62it/s]
 65%|██████▌   | 8687/13317 [1:29:33<47:29,  1.63it/s]
                                                      
{'loss': 0.1767, 'grad_norm': 0.456856369972229, 'learning_rate': 3.251981394499491e-06, 'epoch': 1.96}

 65%|██████▌   | 8687/13317 [1:29:33<47:29,  1.63it/s]
 65%|██████▌   | 8688/13317 [1:29:34<47:25,  1.63it/s]
                                                      
{'loss': 0.1799, 'grad_norm': 0.457760214805603, 'learning_rate': 3.250753523475293e-06, 'epoch': 1.96}

 65%|██████▌   | 8688/13317 [1:29:34<47:25,  1.63it/s]
 65%|██████▌   | 8689/13317 [1:29:34<47:25,  1.63it/s]
                                                      
{'loss': 0.1955, 'grad_norm': 0.5110090374946594, 'learning_rate': 3.249525772642846e-06, 'epoch': 1.96}

 65%|██████▌   | 8689/13317 [1:29:35<47:25,  1.63it/s]
 65%|██████▌   | 8690/13317 [1:29:35<47:28,  1.62it/s]
                                                      
{'loss': 0.2108, 'grad_norm': 0.5929892063140869, 'learning_rate': 3.2482981420865144e-06, 'epoch': 1.96}

 65%|██████▌   | 8690/13317 [1:29:35<47:28,  1.62it/s]
 65%|██████▌   | 8691/13317 [1:29:36<47:25,  1.63it/s]
                                                      
{'loss': 0.1762, 'grad_norm': 0.5054202675819397, 'learning_rate': 3.2470706318906485e-06, 'epoch': 1.96}

 65%|██████▌   | 8691/13317 [1:29:36<47:25,  1.63it/s]
 65%|██████▌   | 8692/13317 [1:29:36<47:24,  1.63it/s]
                                                      
{'loss': 0.1851, 'grad_norm': 0.5476803779602051, 'learning_rate': 3.245843242139591e-06, 'epoch': 1.96}

 65%|██████▌   | 8692/13317 [1:29:36<47:24,  1.63it/s]
 65%|██████▌   | 8693/13317 [1:29:37<47:22,  1.63it/s]
                                                      
{'loss': 0.1744, 'grad_norm': 0.4985075891017914, 'learning_rate': 3.244615972917675e-06, 'epoch': 1.96}

 65%|██████▌   | 8693/13317 [1:29:37<47:22,  1.63it/s]
 65%|██████▌   | 8694/13317 [1:29:37<47:20,  1.63it/s]
                                                      
{'loss': 0.2181, 'grad_norm': 0.5379402041435242, 'learning_rate': 3.2433888243092266e-06, 'epoch': 1.96}

 65%|██████▌   | 8694/13317 [1:29:38<47:20,  1.63it/s]
 65%|██████▌   | 8695/13317 [1:29:38<47:23,  1.63it/s]
                                                      
{'loss': 0.267, 'grad_norm': 0.5313089489936829, 'learning_rate': 3.2421617963985684e-06, 'epoch': 1.96}

 65%|██████▌   | 8695/13317 [1:29:38<47:23,  1.63it/s]
 65%|██████▌   | 8696/13317 [1:29:39<47:20,  1.63it/s]
                                                      
{'loss': 0.2223, 'grad_norm': 0.5523727536201477, 'learning_rate': 3.2409348892700045e-06, 'epoch': 1.96}

 65%|██████▌   | 8696/13317 [1:29:39<47:20,  1.63it/s]
 65%|██████▌   | 8697/13317 [1:29:39<47:18,  1.63it/s]
                                                      
{'loss': 0.2117, 'grad_norm': 0.6149114966392517, 'learning_rate': 3.2397081030078392e-06, 'epoch': 1.96}

 65%|██████▌   | 8697/13317 [1:29:39<47:18,  1.63it/s]
 65%|██████▌   | 8698/13317 [1:29:40<47:16,  1.63it/s]
                                                      
{'loss': 0.2355, 'grad_norm': 0.6728450059890747, 'learning_rate': 3.2384814376963664e-06, 'epoch': 1.96}

 65%|██████▌   | 8698/13317 [1:29:40<47:16,  1.63it/s]
 65%|██████▌   | 8699/13317 [1:29:41<47:15,  1.63it/s]
                                                      
{'loss': 0.1679, 'grad_norm': 0.4773491322994232, 'learning_rate': 3.2372548934198705e-06, 'epoch': 1.96}

 65%|██████▌   | 8699/13317 [1:29:41<47:15,  1.63it/s]
 65%|██████▌   | 8700/13317 [1:29:41<47:18,  1.63it/s]
                                                      
{'loss': 0.1844, 'grad_norm': 0.48122304677963257, 'learning_rate': 3.236028470262627e-06, 'epoch': 1.96}

 65%|██████▌   | 8700/13317 [1:29:41<47:18,  1.63it/s]
 65%|██████▌   | 8701/13317 [1:29:42<47:17,  1.63it/s]
                                                      
{'loss': 0.208, 'grad_norm': 0.5781407356262207, 'learning_rate': 3.2348021683089038e-06, 'epoch': 1.96}

 65%|██████▌   | 8701/13317 [1:29:42<47:17,  1.63it/s]
 65%|██████▌   | 8702/13317 [1:29:42<47:14,  1.63it/s]
                                                      
{'loss': 0.1885, 'grad_norm': 0.482606440782547, 'learning_rate': 3.233575987642965e-06, 'epoch': 1.96}

 65%|██████▌   | 8702/13317 [1:29:42<47:14,  1.63it/s]
 65%|██████▌   | 8703/13317 [1:29:43<47:11,  1.63it/s]
                                                      
{'loss': 0.2371, 'grad_norm': 0.640363335609436, 'learning_rate': 3.2323499283490557e-06, 'epoch': 1.96}

 65%|██████▌   | 8703/13317 [1:29:43<47:11,  1.63it/s]
 65%|██████▌   | 8704/13317 [1:29:44<47:10,  1.63it/s]
                                                      
{'loss': 0.1868, 'grad_norm': 0.5436619520187378, 'learning_rate': 3.2311239905114235e-06, 'epoch': 1.96}

 65%|██████▌   | 8704/13317 [1:29:44<47:10,  1.63it/s]
 65%|██████▌   | 8705/13317 [1:29:44<47:24,  1.62it/s]
                                                      
{'loss': 0.1777, 'grad_norm': 0.4731261432170868, 'learning_rate': 3.2298981742143033e-06, 'epoch': 1.96}

 65%|██████▌   | 8705/13317 [1:29:44<47:24,  1.62it/s]
 65%|██████▌   | 8706/13317 [1:29:45<47:19,  1.62it/s]
                                                      
{'loss': 0.2154, 'grad_norm': 0.4709511995315552, 'learning_rate': 3.22867247954192e-06, 'epoch': 1.96}

 65%|██████▌   | 8706/13317 [1:29:45<47:19,  1.62it/s]
 65%|██████▌   | 8707/13317 [1:29:45<47:17,  1.62it/s]
                                                      
{'loss': 0.175, 'grad_norm': 0.4311782419681549, 'learning_rate': 3.227446906578492e-06, 'epoch': 1.96}

 65%|██████▌   | 8707/13317 [1:29:46<47:17,  1.62it/s]
 65%|██████▌   | 8708/13317 [1:29:46<47:12,  1.63it/s]
                                                      
{'loss': 0.2103, 'grad_norm': 0.5015023946762085, 'learning_rate': 3.226221455408229e-06, 'epoch': 1.96}

 65%|██████▌   | 8708/13317 [1:29:46<47:12,  1.63it/s]
 65%|██████▌   | 8709/13317 [1:29:47<47:10,  1.63it/s]
                                                      
{'loss': 0.1615, 'grad_norm': 0.5027396082878113, 'learning_rate': 3.224996126115336e-06, 'epoch': 1.96}

 65%|██████▌   | 8709/13317 [1:29:47<47:10,  1.63it/s]
 65%|██████▌   | 8710/13317 [1:29:47<47:14,  1.63it/s]
                                                      
{'loss': 0.1835, 'grad_norm': 0.5641205906867981, 'learning_rate': 3.223770918784e-06, 'epoch': 1.96}

 65%|██████▌   | 8710/13317 [1:29:47<47:14,  1.63it/s]
 65%|██████▌   | 8711/13317 [1:29:48<47:10,  1.63it/s]
                                                      
{'loss': 0.1682, 'grad_norm': 0.4441681206226349, 'learning_rate': 3.22254583349841e-06, 'epoch': 1.96}

 65%|██████▌   | 8711/13317 [1:29:48<47:10,  1.63it/s]
 65%|██████▌   | 8712/13317 [1:29:49<47:08,  1.63it/s]
                                                      
{'loss': 0.184, 'grad_norm': 0.496718168258667, 'learning_rate': 3.2213208703427422e-06, 'epoch': 1.96}

 65%|██████▌   | 8712/13317 [1:29:49<47:08,  1.63it/s]
 65%|██████▌   | 8713/13317 [1:29:49<47:13,  1.62it/s]
                                                      
{'loss': 0.2006, 'grad_norm': 0.5746641755104065, 'learning_rate': 3.220096029401163e-06, 'epoch': 1.96}

 65%|██████▌   | 8713/13317 [1:29:49<47:13,  1.62it/s]
 65%|██████▌   | 8714/13317 [1:29:50<47:11,  1.63it/s]
                                                      
{'loss': 0.1793, 'grad_norm': 0.43857452273368835, 'learning_rate': 3.2188713107578316e-06, 'epoch': 1.96}

 65%|██████▌   | 8714/13317 [1:29:50<47:11,  1.63it/s]
 65%|██████▌   | 8715/13317 [1:29:50<47:15,  1.62it/s]
                                                      
{'loss': 0.1708, 'grad_norm': 0.4468333423137665, 'learning_rate': 3.2176467144968993e-06, 'epoch': 1.96}

 65%|██████▌   | 8715/13317 [1:29:50<47:15,  1.62it/s]
 65%|██████▌   | 8716/13317 [1:29:51<47:12,  1.62it/s]
                                                      
{'loss': 0.1982, 'grad_norm': 0.520114541053772, 'learning_rate': 3.2164222407025126e-06, 'epoch': 1.96}

 65%|██████▌   | 8716/13317 [1:29:51<47:12,  1.62it/s]
 65%|██████▌   | 8717/13317 [1:29:52<47:09,  1.63it/s]
                                                      
{'loss': 0.2006, 'grad_norm': 0.545775294303894, 'learning_rate': 3.215197889458801e-06, 'epoch': 1.96}

 65%|██████▌   | 8717/13317 [1:29:52<47:09,  1.63it/s]
 65%|██████▌   | 8718/13317 [1:29:52<47:08,  1.63it/s]
                                                      
{'loss': 0.1909, 'grad_norm': 0.5151386857032776, 'learning_rate': 3.213973660849892e-06, 'epoch': 1.96}

 65%|██████▌   | 8718/13317 [1:29:52<47:08,  1.63it/s]
 65%|██████▌   | 8719/13317 [1:29:53<47:07,  1.63it/s]
                                                      
{'loss': 0.2188, 'grad_norm': 0.5400554537773132, 'learning_rate': 3.212749554959904e-06, 'epoch': 1.96}

 65%|██████▌   | 8719/13317 [1:29:53<47:07,  1.63it/s]
 65%|██████▌   | 8720/13317 [1:29:53<47:09,  1.62it/s]
                                                      
{'loss': 0.3072, 'grad_norm': 0.6978657841682434, 'learning_rate': 3.2115255718729456e-06, 'epoch': 1.96}

 65%|██████▌   | 8720/13317 [1:29:54<47:09,  1.62it/s]
 65%|██████▌   | 8721/13317 [1:29:54<47:07,  1.63it/s]
                                                      
{'loss': 0.1892, 'grad_norm': 0.5040059089660645, 'learning_rate': 3.210301711673115e-06, 'epoch': 1.96}

 65%|██████▌   | 8721/13317 [1:29:54<47:07,  1.63it/s]
 65%|██████▌   | 8722/13317 [1:29:55<47:06,  1.63it/s]
                                                      
{'loss': 0.1672, 'grad_norm': 0.4618811011314392, 'learning_rate': 3.2090779744445096e-06, 'epoch': 1.96}

 65%|██████▌   | 8722/13317 [1:29:55<47:06,  1.63it/s]
 66%|██████▌   | 8723/13317 [1:29:55<47:06,  1.63it/s]
                                                      
{'loss': 0.1977, 'grad_norm': 0.5149081349372864, 'learning_rate': 3.2078543602712062e-06, 'epoch': 1.97}

 66%|██████▌   | 8723/13317 [1:29:55<47:06,  1.63it/s]
 66%|██████▌   | 8724/13317 [1:29:56<47:02,  1.63it/s]
                                                      
{'loss': 0.2093, 'grad_norm': 0.634545087814331, 'learning_rate': 3.206630869237286e-06, 'epoch': 1.97}

 66%|██████▌   | 8724/13317 [1:29:56<47:02,  1.63it/s]
 66%|██████▌   | 8725/13317 [1:29:57<47:05,  1.62it/s]
                                                      
{'loss': 0.1909, 'grad_norm': 0.499769389629364, 'learning_rate': 3.2054075014268134e-06, 'epoch': 1.97}

 66%|██████▌   | 8725/13317 [1:29:57<47:05,  1.62it/s]
 66%|██████▌   | 8726/13317 [1:29:57<47:02,  1.63it/s]
                                                      
{'loss': 0.2348, 'grad_norm': 0.5149517059326172, 'learning_rate': 3.204184256923848e-06, 'epoch': 1.97}

 66%|██████▌   | 8726/13317 [1:29:57<47:02,  1.63it/s]
 66%|██████▌   | 8727/13317 [1:29:58<47:00,  1.63it/s]
                                                      
{'loss': 0.1778, 'grad_norm': 0.5304781198501587, 'learning_rate': 3.202961135812437e-06, 'epoch': 1.97}

 66%|██████▌   | 8727/13317 [1:29:58<47:00,  1.63it/s]
 66%|██████▌   | 8728/13317 [1:29:58<47:03,  1.63it/s]
                                                      
{'loss': 0.1697, 'grad_norm': 0.46895551681518555, 'learning_rate': 3.2017381381766223e-06, 'epoch': 1.97}

 66%|██████▌   | 8728/13317 [1:29:58<47:03,  1.63it/s]
 66%|██████▌   | 8729/13317 [1:29:59<47:01,  1.63it/s]
                                                      
{'loss': 0.1613, 'grad_norm': 0.48829686641693115, 'learning_rate': 3.200515264100441e-06, 'epoch': 1.97}

 66%|██████▌   | 8729/13317 [1:29:59<47:01,  1.63it/s]
 66%|██████▌   | 8730/13317 [1:30:00<47:02,  1.63it/s]
                                                      
{'loss': 0.1612, 'grad_norm': 0.5261015892028809, 'learning_rate': 3.1992925136679116e-06, 'epoch': 1.97}

 66%|██████▌   | 8730/13317 [1:30:00<47:02,  1.63it/s]
 66%|██████▌   | 8731/13317 [1:30:00<46:59,  1.63it/s]
                                                      
{'loss': 0.169, 'grad_norm': 0.47959113121032715, 'learning_rate': 3.1980698869630533e-06, 'epoch': 1.97}

 66%|██████▌   | 8731/13317 [1:30:00<46:59,  1.63it/s]
 66%|██████▌   | 8732/13317 [1:30:01<46:55,  1.63it/s]
                                                      
{'loss': 0.1616, 'grad_norm': 0.43524274230003357, 'learning_rate': 3.1968473840698734e-06, 'epoch': 1.97}

 66%|██████▌   | 8732/13317 [1:30:01<46:55,  1.63it/s]
 66%|██████▌   | 8733/13317 [1:30:01<47:09,  1.62it/s]
                                                      
{'loss': 0.1611, 'grad_norm': 0.4312155842781067, 'learning_rate': 3.1956250050723704e-06, 'epoch': 1.97}

 66%|██████▌   | 8733/13317 [1:30:02<47:09,  1.62it/s]
 66%|██████▌   | 8734/13317 [1:30:02<47:03,  1.62it/s]
                                                      
{'loss': 0.1817, 'grad_norm': 0.5678026676177979, 'learning_rate': 3.1944027500545344e-06, 'epoch': 1.97}

 66%|██████▌   | 8734/13317 [1:30:02<47:03,  1.62it/s]
 66%|██████▌   | 8735/13317 [1:30:03<47:03,  1.62it/s]
                                                      
{'loss': 0.2044, 'grad_norm': 0.5345420241355896, 'learning_rate': 3.193180619100346e-06, 'epoch': 1.97}

 66%|██████▌   | 8735/13317 [1:30:03<47:03,  1.62it/s]
 66%|██████▌   | 8736/13317 [1:30:03<46:59,  1.62it/s]
                                                      
{'loss': 0.1937, 'grad_norm': 0.4634542465209961, 'learning_rate': 3.1919586122937835e-06, 'epoch': 1.97}

 66%|██████▌   | 8736/13317 [1:30:03<46:59,  1.62it/s]
 66%|██████▌   | 8737/13317 [1:30:04<46:55,  1.63it/s]
                                                      
{'loss': 0.1785, 'grad_norm': 0.5087742209434509, 'learning_rate': 3.190736729718805e-06, 'epoch': 1.97}

 66%|██████▌   | 8737/13317 [1:30:04<46:55,  1.63it/s]
 66%|██████▌   | 8738/13317 [1:30:05<47:06,  1.62it/s]
                                                      
{'loss': 0.2001, 'grad_norm': 0.5281535387039185, 'learning_rate': 3.1895149714593704e-06, 'epoch': 1.97}

 66%|██████▌   | 8738/13317 [1:30:05<47:06,  1.62it/s]
 66%|██████▌   | 8739/13317 [1:30:05<47:00,  1.62it/s]
                                                      
{'loss': 0.2199, 'grad_norm': 0.5999683737754822, 'learning_rate': 3.1882933375994286e-06, 'epoch': 1.97}

 66%|██████▌   | 8739/13317 [1:30:05<47:00,  1.62it/s]
 66%|██████▌   | 8740/13317 [1:30:06<47:11,  1.62it/s]
                                                      
{'loss': 0.2092, 'grad_norm': 0.6278722286224365, 'learning_rate': 3.1870718282229156e-06, 'epoch': 1.97}

 66%|██████▌   | 8740/13317 [1:30:06<47:11,  1.62it/s]
 66%|██████▌   | 8741/13317 [1:30:06<47:04,  1.62it/s]
                                                      
{'loss': 0.1932, 'grad_norm': 0.566177487373352, 'learning_rate': 3.185850443413763e-06, 'epoch': 1.97}

 66%|██████▌   | 8741/13317 [1:30:06<47:04,  1.62it/s]
 66%|██████▌   | 8742/13317 [1:30:07<46:58,  1.62it/s]
                                                      
{'loss': 0.2301, 'grad_norm': 0.6130174398422241, 'learning_rate': 3.1846291832558946e-06, 'epoch': 1.97}

 66%|██████▌   | 8742/13317 [1:30:07<46:58,  1.62it/s]
 66%|██████▌   | 8743/13317 [1:30:08<46:58,  1.62it/s]
                                                      
{'loss': 0.1637, 'grad_norm': 0.5069092512130737, 'learning_rate': 3.1834080478332237e-06, 'epoch': 1.97}

 66%|██████▌   | 8743/13317 [1:30:08<46:58,  1.62it/s]
 66%|██████▌   | 8744/13317 [1:30:08<46:52,  1.63it/s]
                                                      
{'loss': 0.1911, 'grad_norm': 0.5015242695808411, 'learning_rate': 3.182187037229653e-06, 'epoch': 1.97}

 66%|██████▌   | 8744/13317 [1:30:08<46:52,  1.63it/s]
 66%|██████▌   | 8745/13317 [1:30:09<46:55,  1.62it/s]
                                                      
{'loss': 0.1785, 'grad_norm': 0.5362311005592346, 'learning_rate': 3.1809661515290803e-06, 'epoch': 1.97}

 66%|██████▌   | 8745/13317 [1:30:09<46:55,  1.62it/s]
 66%|██████▌   | 8746/13317 [1:30:09<46:51,  1.63it/s]
                                                      
{'loss': 0.1674, 'grad_norm': 0.4835430085659027, 'learning_rate': 3.179745390815394e-06, 'epoch': 1.97}

 66%|██████▌   | 8746/13317 [1:30:10<46:51,  1.63it/s]
 66%|██████▌   | 8747/13317 [1:30:10<46:48,  1.63it/s]
                                                      
{'loss': 0.2332, 'grad_norm': 0.5828952789306641, 'learning_rate': 3.178524755172472e-06, 'epoch': 1.97}

 66%|██████▌   | 8747/13317 [1:30:10<46:48,  1.63it/s]
 66%|██████▌   | 8748/13317 [1:30:11<46:46,  1.63it/s]
                                                      
{'loss': 0.2488, 'grad_norm': 0.6304410696029663, 'learning_rate': 3.177304244684184e-06, 'epoch': 1.97}

 66%|██████▌   | 8748/13317 [1:30:11<46:46,  1.63it/s]
 66%|██████▌   | 8749/13317 [1:30:11<46:45,  1.63it/s]
                                                      
{'loss': 0.2105, 'grad_norm': 0.533015787601471, 'learning_rate': 3.1760838594343966e-06, 'epoch': 1.97}

 66%|██████▌   | 8749/13317 [1:30:11<46:45,  1.63it/s]
 66%|██████▌   | 8750/13317 [1:30:12<46:49,  1.63it/s]
                                                      
{'loss': 0.1969, 'grad_norm': 0.5303695797920227, 'learning_rate': 3.1748635995069578e-06, 'epoch': 1.97}

 66%|██████▌   | 8750/13317 [1:30:12<46:49,  1.63it/s]
 66%|██████▌   | 8751/13317 [1:30:13<46:46,  1.63it/s]
                                                      
{'loss': 0.1932, 'grad_norm': 0.515116274356842, 'learning_rate': 3.173643464985714e-06, 'epoch': 1.97}

 66%|██████▌   | 8751/13317 [1:30:13<46:46,  1.63it/s]
 66%|██████▌   | 8752/13317 [1:30:13<46:42,  1.63it/s]
                                                      
{'loss': 0.1981, 'grad_norm': 0.5520653128623962, 'learning_rate': 3.172423455954503e-06, 'epoch': 1.97}

 66%|██████▌   | 8752/13317 [1:30:13<46:42,  1.63it/s]
 66%|██████▌   | 8753/13317 [1:30:14<46:42,  1.63it/s]
                                                      
{'loss': 0.282, 'grad_norm': 0.6667347550392151, 'learning_rate': 3.1712035724971516e-06, 'epoch': 1.97}

 66%|██████▌   | 8753/13317 [1:30:14<46:42,  1.63it/s]
 66%|██████▌   | 8754/13317 [1:30:14<46:41,  1.63it/s]
                                                      
{'loss': 0.1711, 'grad_norm': 0.7753921747207642, 'learning_rate': 3.1699838146974784e-06, 'epoch': 1.97}

 66%|██████▌   | 8754/13317 [1:30:14<46:41,  1.63it/s]
 66%|██████▌   | 8755/13317 [1:30:15<46:46,  1.63it/s]
                                                      
{'loss': 0.1907, 'grad_norm': 0.5442038178443909, 'learning_rate': 3.1687641826392913e-06, 'epoch': 1.97}

 66%|██████▌   | 8755/13317 [1:30:15<46:46,  1.63it/s]
 66%|██████▌   | 8756/13317 [1:30:16<46:43,  1.63it/s]
                                                      
{'loss': 0.1644, 'grad_norm': 0.47116023302078247, 'learning_rate': 3.1675446764063984e-06, 'epoch': 1.97}

 66%|██████▌   | 8756/13317 [1:30:16<46:43,  1.63it/s]
 66%|██████▌   | 8757/13317 [1:30:16<46:41,  1.63it/s]
                                                      
{'loss': 0.1965, 'grad_norm': 0.4801520109176636, 'learning_rate': 3.1663252960825852e-06, 'epoch': 1.97}

 66%|██████▌   | 8757/13317 [1:30:16<46:41,  1.63it/s]
 66%|██████▌   | 8758/13317 [1:30:17<46:39,  1.63it/s]
                                                      
{'loss': 0.2149, 'grad_norm': 0.597922146320343, 'learning_rate': 3.16510604175164e-06, 'epoch': 1.97}

 66%|██████▌   | 8758/13317 [1:30:17<46:39,  1.63it/s]
 66%|██████▌   | 8759/13317 [1:30:17<46:36,  1.63it/s]
                                                      
{'loss': 0.1807, 'grad_norm': 0.47496771812438965, 'learning_rate': 3.163886913497338e-06, 'epoch': 1.97}

 66%|██████▌   | 8759/13317 [1:30:18<46:36,  1.63it/s]
 66%|██████▌   | 8760/13317 [1:30:18<46:50,  1.62it/s]
                                                      
{'loss': 0.2724, 'grad_norm': 0.5897224545478821, 'learning_rate': 3.1626679114034477e-06, 'epoch': 1.97}

 66%|██████▌   | 8760/13317 [1:30:18<46:50,  1.62it/s]
 66%|██████▌   | 8761/13317 [1:30:19<46:45,  1.62it/s]
                                                      
{'loss': 0.1886, 'grad_norm': 0.5221827626228333, 'learning_rate': 3.161449035553724e-06, 'epoch': 1.97}

 66%|██████▌   | 8761/13317 [1:30:19<46:45,  1.62it/s]
 66%|██████▌   | 8762/13317 [1:30:19<46:42,  1.63it/s]
                                                      
{'loss': 0.1833, 'grad_norm': 0.5011544823646545, 'learning_rate': 3.1602302860319185e-06, 'epoch': 1.97}

 66%|██████▌   | 8762/13317 [1:30:19<46:42,  1.63it/s]
 66%|██████▌   | 8763/13317 [1:30:20<46:40,  1.63it/s]
                                                      
{'loss': 0.1672, 'grad_norm': 0.49152448773384094, 'learning_rate': 3.1590116629217738e-06, 'epoch': 1.97}

 66%|██████▌   | 8763/13317 [1:30:20<46:40,  1.63it/s]
 66%|██████▌   | 8764/13317 [1:30:21<46:38,  1.63it/s]
                                                      
{'loss': 0.1782, 'grad_norm': 0.48580634593963623, 'learning_rate': 3.1577931663070194e-06, 'epoch': 1.97}

 66%|██████▌   | 8764/13317 [1:30:21<46:38,  1.63it/s]
 66%|██████▌   | 8765/13317 [1:30:21<46:52,  1.62it/s]
                                                      
{'loss': 0.2516, 'grad_norm': 0.6015028953552246, 'learning_rate': 3.15657479627138e-06, 'epoch': 1.97}

 66%|██████▌   | 8765/13317 [1:30:21<46:52,  1.62it/s]
 66%|██████▌   | 8766/13317 [1:30:22<47:05,  1.61it/s]
                                                      
{'loss': 0.1601, 'grad_norm': 0.4673313498497009, 'learning_rate': 3.1553565528985718e-06, 'epoch': 1.97}

 66%|██████▌   | 8766/13317 [1:30:22<47:05,  1.61it/s]
 66%|██████▌   | 8767/13317 [1:30:22<48:35,  1.56it/s]
                                                      
{'loss': 0.2628, 'grad_norm': 0.6824458241462708, 'learning_rate': 3.1541384362722983e-06, 'epoch': 1.97}

 66%|██████▌   | 8767/13317 [1:30:23<48:35,  1.56it/s]
 66%|██████▌   | 8768/13317 [1:30:23<50:06,  1.51it/s]
                                                      
{'loss': 0.165, 'grad_norm': 0.48491206765174866, 'learning_rate': 3.152920446476258e-06, 'epoch': 1.98}

 66%|██████▌   | 8768/13317 [1:30:23<50:06,  1.51it/s]
 66%|██████▌   | 8769/13317 [1:30:24<49:56,  1.52it/s]
                                                      
{'loss': 0.3117, 'grad_norm': 0.563646674156189, 'learning_rate': 3.151702583594141e-06, 'epoch': 1.98}

 66%|██████▌   | 8769/13317 [1:30:24<49:56,  1.52it/s]
 66%|██████▌   | 8770/13317 [1:30:24<49:22,  1.53it/s]
                                                      
{'loss': 0.2124, 'grad_norm': 0.6091720461845398, 'learning_rate': 3.1504848477096276e-06, 'epoch': 1.98}

 66%|██████▌   | 8770/13317 [1:30:25<49:22,  1.53it/s]
 66%|██████▌   | 8771/13317 [1:30:25<48:30,  1.56it/s]
                                                      
{'loss': 0.256, 'grad_norm': 0.6265940070152283, 'learning_rate': 3.149267238906387e-06, 'epoch': 1.98}

 66%|██████▌   | 8771/13317 [1:30:25<48:30,  1.56it/s]
 66%|██████▌   | 8772/13317 [1:30:26<47:54,  1.58it/s]
                                                      
{'loss': 0.2168, 'grad_norm': 0.5743520855903625, 'learning_rate': 3.1480497572680835e-06, 'epoch': 1.98}

 66%|██████▌   | 8772/13317 [1:30:26<47:54,  1.58it/s]
 66%|██████▌   | 8773/13317 [1:30:26<47:31,  1.59it/s]
                                                      
{'loss': 0.1655, 'grad_norm': 0.4338301420211792, 'learning_rate': 3.1468324028783716e-06, 'epoch': 1.98}

 66%|██████▌   | 8773/13317 [1:30:26<47:31,  1.59it/s]
 66%|██████▌   | 8774/13317 [1:30:27<47:10,  1.60it/s]
                                                      
{'loss': 0.1811, 'grad_norm': 0.5024557709693909, 'learning_rate': 3.1456151758208936e-06, 'epoch': 1.98}

 66%|██████▌   | 8774/13317 [1:30:27<47:10,  1.60it/s]
 66%|██████▌   | 8775/13317 [1:30:28<46:56,  1.61it/s]
                                                      
{'loss': 0.1765, 'grad_norm': 0.458493173122406, 'learning_rate': 3.144398076179287e-06, 'epoch': 1.98}

 66%|██████▌   | 8775/13317 [1:30:28<46:56,  1.61it/s]
 66%|██████▌   | 8776/13317 [1:30:28<46:46,  1.62it/s]
                                                      
{'loss': 0.1834, 'grad_norm': 0.499682754278183, 'learning_rate': 3.143181104037181e-06, 'epoch': 1.98}

 66%|██████▌   | 8776/13317 [1:30:28<46:46,  1.62it/s]
 66%|██████▌   | 8777/13317 [1:30:29<46:39,  1.62it/s]
                                                      
{'loss': 0.1442, 'grad_norm': 0.4453289210796356, 'learning_rate': 3.141964259478194e-06, 'epoch': 1.98}

 66%|██████▌   | 8777/13317 [1:30:29<46:39,  1.62it/s]
 66%|██████▌   | 8778/13317 [1:30:29<46:38,  1.62it/s]
                                                      
{'loss': 0.1651, 'grad_norm': 0.44988009333610535, 'learning_rate': 3.1407475425859348e-06, 'epoch': 1.98}

 66%|██████▌   | 8778/13317 [1:30:29<46:38,  1.62it/s]
 66%|██████▌   | 8779/13317 [1:30:30<46:35,  1.62it/s]
                                                      
{'loss': 0.149, 'grad_norm': 0.43774887919425964, 'learning_rate': 3.1395309534440055e-06, 'epoch': 1.98}

 66%|██████▌   | 8779/13317 [1:30:30<46:35,  1.62it/s]
 66%|██████▌   | 8780/13317 [1:30:31<46:30,  1.63it/s]
                                                      
{'loss': 0.2763, 'grad_norm': 0.6629745364189148, 'learning_rate': 3.1383144921359998e-06, 'epoch': 1.98}

 66%|██████▌   | 8780/13317 [1:30:31<46:30,  1.63it/s]
 66%|██████▌   | 8781/13317 [1:30:31<46:26,  1.63it/s]
                                                      
{'loss': 0.2518, 'grad_norm': 0.6247119903564453, 'learning_rate': 3.1370981587454986e-06, 'epoch': 1.98}

 66%|██████▌   | 8781/13317 [1:30:31<46:26,  1.63it/s]
 66%|██████▌   | 8782/13317 [1:30:32<46:23,  1.63it/s]
                                                      
{'loss': 0.213, 'grad_norm': 0.5507243275642395, 'learning_rate': 3.1358819533560787e-06, 'epoch': 1.98}

 66%|██████▌   | 8782/13317 [1:30:32<46:23,  1.63it/s]
 66%|██████▌   | 8783/13317 [1:30:32<46:28,  1.63it/s]
                                                      
{'loss': 0.1803, 'grad_norm': 0.6185203194618225, 'learning_rate': 3.1346658760513083e-06, 'epoch': 1.98}

 66%|██████▌   | 8783/13317 [1:30:33<46:28,  1.63it/s]
 66%|██████▌   | 8784/13317 [1:30:33<46:25,  1.63it/s]
                                                      
{'loss': 0.1596, 'grad_norm': 0.5038921236991882, 'learning_rate': 3.13344992691474e-06, 'epoch': 1.98}

 66%|██████▌   | 8784/13317 [1:30:33<46:25,  1.63it/s]
 66%|██████▌   | 8785/13317 [1:30:34<46:35,  1.62it/s]
                                                      
{'loss': 0.1744, 'grad_norm': 0.47501182556152344, 'learning_rate': 3.1322341060299266e-06, 'epoch': 1.98}

 66%|██████▌   | 8785/13317 [1:30:34<46:35,  1.62it/s]
 66%|██████▌   | 8786/13317 [1:30:34<46:40,  1.62it/s]
                                                      
{'loss': 0.1615, 'grad_norm': 0.4685271978378296, 'learning_rate': 3.1310184134804057e-06, 'epoch': 1.98}

 66%|██████▌   | 8786/13317 [1:30:34<46:40,  1.62it/s]
 66%|██████▌   | 8787/13317 [1:30:35<46:33,  1.62it/s]
                                                      
{'loss': 0.1813, 'grad_norm': 0.5186488032341003, 'learning_rate': 3.12980284934971e-06, 'epoch': 1.98}

 66%|██████▌   | 8787/13317 [1:30:35<46:33,  1.62it/s]
 66%|██████▌   | 8788/13317 [1:30:36<46:34,  1.62it/s]
                                                      
{'loss': 0.1847, 'grad_norm': 0.48152244091033936, 'learning_rate': 3.12858741372136e-06, 'epoch': 1.98}

 66%|██████▌   | 8788/13317 [1:30:36<46:34,  1.62it/s]
 66%|██████▌   | 8789/13317 [1:30:36<46:28,  1.62it/s]
                                                      
{'loss': 0.1978, 'grad_norm': 0.5575301051139832, 'learning_rate': 3.1273721066788683e-06, 'epoch': 1.98}

 66%|██████▌   | 8789/13317 [1:30:36<46:28,  1.62it/s]
 66%|██████▌   | 8790/13317 [1:30:37<46:24,  1.63it/s]
                                                      
{'loss': 0.1692, 'grad_norm': 0.49287542700767517, 'learning_rate': 3.1261569283057423e-06, 'epoch': 1.98}

 66%|██████▌   | 8790/13317 [1:30:37<46:24,  1.63it/s]
 66%|██████▌   | 8791/13317 [1:30:37<46:26,  1.62it/s]
                                                      
{'loss': 0.1777, 'grad_norm': 0.528741180896759, 'learning_rate': 3.1249418786854746e-06, 'epoch': 1.98}

 66%|██████▌   | 8791/13317 [1:30:37<46:26,  1.62it/s]
 66%|██████▌   | 8792/13317 [1:30:38<46:22,  1.63it/s]
                                                      
{'loss': 0.2276, 'grad_norm': 0.6329333782196045, 'learning_rate': 3.123726957901554e-06, 'epoch': 1.98}

 66%|██████▌   | 8792/13317 [1:30:38<46:22,  1.63it/s]
 66%|██████▌   | 8793/13317 [1:30:39<46:23,  1.63it/s]
                                                      
{'loss': 0.1453, 'grad_norm': 0.444606751203537, 'learning_rate': 3.122512166037458e-06, 'epoch': 1.98}

 66%|██████▌   | 8793/13317 [1:30:39<46:23,  1.63it/s]
 66%|██████▌   | 8794/13317 [1:30:39<46:19,  1.63it/s]
                                                      
{'loss': 0.1926, 'grad_norm': 0.5204915404319763, 'learning_rate': 3.121297503176655e-06, 'epoch': 1.98}

 66%|██████▌   | 8794/13317 [1:30:39<46:19,  1.63it/s]
 66%|██████▌   | 8795/13317 [1:30:40<46:18,  1.63it/s]
                                                      
{'loss': 0.1691, 'grad_norm': 0.5085502862930298, 'learning_rate': 3.120082969402605e-06, 'epoch': 1.98}

 66%|██████▌   | 8795/13317 [1:30:40<46:18,  1.63it/s]
 66%|██████▌   | 8796/13317 [1:30:40<46:17,  1.63it/s]
                                                      
{'loss': 0.2267, 'grad_norm': 0.6122068762779236, 'learning_rate': 3.1188685647987595e-06, 'epoch': 1.98}

 66%|██████▌   | 8796/13317 [1:30:41<46:17,  1.63it/s]
 66%|██████▌   | 8797/13317 [1:30:41<46:17,  1.63it/s]
                                                      
{'loss': 0.1635, 'grad_norm': 0.4600483477115631, 'learning_rate': 3.1176542894485627e-06, 'epoch': 1.98}

 66%|██████▌   | 8797/13317 [1:30:41<46:17,  1.63it/s]
 66%|██████▌   | 8798/13317 [1:30:42<46:14,  1.63it/s]
                                                      
{'loss': 0.1745, 'grad_norm': 0.5021507143974304, 'learning_rate': 3.1164401434354465e-06, 'epoch': 1.98}

 66%|██████▌   | 8798/13317 [1:30:42<46:14,  1.63it/s]
 66%|██████▌   | 8799/13317 [1:30:42<46:14,  1.63it/s]
                                                      
{'loss': 0.2183, 'grad_norm': 0.5568603873252869, 'learning_rate': 3.115226126842835e-06, 'epoch': 1.98}

 66%|██████▌   | 8799/13317 [1:30:42<46:14,  1.63it/s]
 66%|██████▌   | 8800/13317 [1:30:43<46:12,  1.63it/s]
                                                      
{'loss': 0.2019, 'grad_norm': 0.5542581081390381, 'learning_rate': 3.1140122397541467e-06, 'epoch': 1.98}

 66%|██████▌   | 8800/13317 [1:30:43<46:12,  1.63it/s]
 66%|██████▌   | 8801/13317 [1:30:43<46:15,  1.63it/s]
                                                      
{'loss': 0.2008, 'grad_norm': 0.5287497639656067, 'learning_rate': 3.112798482252786e-06, 'epoch': 1.98}

 66%|██████▌   | 8801/13317 [1:30:44<46:15,  1.63it/s]
 66%|██████▌   | 8802/13317 [1:30:44<46:12,  1.63it/s]
                                                      
{'loss': 0.2253, 'grad_norm': 0.5952925682067871, 'learning_rate': 3.1115848544221493e-06, 'epoch': 1.98}

 66%|██████▌   | 8802/13317 [1:30:44<46:12,  1.63it/s]
 66%|██████▌   | 8803/13317 [1:30:45<46:09,  1.63it/s]
                                                      
{'loss': 0.2289, 'grad_norm': 0.6020736694335938, 'learning_rate': 3.1103713563456294e-06, 'epoch': 1.98}

 66%|██████▌   | 8803/13317 [1:30:45<46:09,  1.63it/s]
 66%|██████▌   | 8804/13317 [1:30:45<46:08,  1.63it/s]
                                                      
{'loss': 0.1604, 'grad_norm': 0.46905943751335144, 'learning_rate': 3.1091579881066066e-06, 'epoch': 1.98}

 66%|██████▌   | 8804/13317 [1:30:45<46:08,  1.63it/s]
 66%|██████▌   | 8805/13317 [1:30:46<46:08,  1.63it/s]
                                                      
{'loss': 0.1941, 'grad_norm': 0.5357271432876587, 'learning_rate': 3.107944749788449e-06, 'epoch': 1.98}

 66%|██████▌   | 8805/13317 [1:30:46<46:08,  1.63it/s]
 66%|██████▌   | 8806/13317 [1:30:47<46:10,  1.63it/s]
                                                      
{'loss': 0.1403, 'grad_norm': 0.4126317501068115, 'learning_rate': 3.1067316414745207e-06, 'epoch': 1.98}

 66%|██████▌   | 8806/13317 [1:30:47<46:10,  1.63it/s]
 66%|██████▌   | 8807/13317 [1:30:47<46:08,  1.63it/s]
                                                      
{'loss': 0.2074, 'grad_norm': 0.5426793694496155, 'learning_rate': 3.1055186632481758e-06, 'epoch': 1.98}

 66%|██████▌   | 8807/13317 [1:30:47<46:08,  1.63it/s]
 66%|██████▌   | 8808/13317 [1:30:48<46:08,  1.63it/s]
                                                      
{'loss': 0.1789, 'grad_norm': 0.4961150586605072, 'learning_rate': 3.1043058151927573e-06, 'epoch': 1.98}

 66%|██████▌   | 8808/13317 [1:30:48<46:08,  1.63it/s]
 66%|██████▌   | 8809/13317 [1:30:48<46:07,  1.63it/s]
                                                      
{'loss': 0.1989, 'grad_norm': 0.6140630841255188, 'learning_rate': 3.1030930973915997e-06, 'epoch': 1.98}

 66%|██████▌   | 8809/13317 [1:30:49<46:07,  1.63it/s]
 66%|██████▌   | 8810/13317 [1:30:49<48:29,  1.55it/s]
                                                      
{'loss': 0.1502, 'grad_norm': 0.43174582719802856, 'learning_rate': 3.1018805099280345e-06, 'epoch': 1.98}

 66%|██████▌   | 8810/13317 [1:30:49<48:29,  1.55it/s]
 66%|██████▌   | 8811/13317 [1:30:50<47:48,  1.57it/s]
                                                      
{'loss': 0.1742, 'grad_norm': 0.48628830909729004, 'learning_rate': 3.1006680528853726e-06, 'epoch': 1.98}

 66%|██████▌   | 8811/13317 [1:30:50<47:48,  1.57it/s]
 66%|██████▌   | 8812/13317 [1:30:50<47:18,  1.59it/s]
                                                      
{'loss': 0.252, 'grad_norm': 0.5708914995193481, 'learning_rate': 3.0994557263469267e-06, 'epoch': 1.99}

 66%|██████▌   | 8812/13317 [1:30:50<47:18,  1.59it/s]
 66%|██████▌   | 8813/13317 [1:30:51<46:59,  1.60it/s]
                                                      
{'loss': 0.1694, 'grad_norm': 0.5069010257720947, 'learning_rate': 3.0982435303959955e-06, 'epoch': 1.99}

 66%|██████▌   | 8813/13317 [1:30:51<46:59,  1.60it/s]
 66%|██████▌   | 8814/13317 [1:30:52<46:41,  1.61it/s]
                                                      
{'loss': 0.1977, 'grad_norm': 0.5909059643745422, 'learning_rate': 3.097031465115872e-06, 'epoch': 1.99}

 66%|██████▌   | 8814/13317 [1:30:52<46:41,  1.61it/s]
 66%|██████▌   | 8815/13317 [1:30:52<46:29,  1.61it/s]
                                                      
{'loss': 0.1801, 'grad_norm': 0.5548432469367981, 'learning_rate': 3.0958195305898335e-06, 'epoch': 1.99}

 66%|██████▌   | 8815/13317 [1:30:52<46:29,  1.61it/s]
 66%|██████▌   | 8816/13317 [1:30:53<48:46,  1.54it/s]
                                                      
{'loss': 0.2442, 'grad_norm': 0.7108258008956909, 'learning_rate': 3.094607726901155e-06, 'epoch': 1.99}

 66%|██████▌   | 8816/13317 [1:30:53<48:46,  1.54it/s]
 66%|██████▌   | 8817/13317 [1:30:54<47:56,  1.56it/s]
                                                      
{'loss': 0.1894, 'grad_norm': 0.5074585676193237, 'learning_rate': 3.093396054133102e-06, 'epoch': 1.99}

 66%|██████▌   | 8817/13317 [1:30:54<47:56,  1.56it/s]
 66%|██████▌   | 8818/13317 [1:30:54<47:24,  1.58it/s]
                                                      
{'loss': 0.2426, 'grad_norm': 0.6349966526031494, 'learning_rate': 3.0921845123689264e-06, 'epoch': 1.99}

 66%|██████▌   | 8818/13317 [1:30:54<47:24,  1.58it/s]
 66%|██████▌   | 8819/13317 [1:30:55<47:14,  1.59it/s]
                                                      
{'loss': 0.1703, 'grad_norm': 0.49747881293296814, 'learning_rate': 3.090973101691875e-06, 'epoch': 1.99}

 66%|██████▌   | 8819/13317 [1:30:55<47:14,  1.59it/s]
 66%|██████▌   | 8820/13317 [1:30:55<46:51,  1.60it/s]
                                                      
{'loss': 0.1852, 'grad_norm': 0.4949232339859009, 'learning_rate': 3.0897618221851833e-06, 'epoch': 1.99}

 66%|██████▌   | 8820/13317 [1:30:56<46:51,  1.60it/s]
 66%|██████▌   | 8821/13317 [1:30:56<46:41,  1.61it/s]
                                                      
{'loss': 0.2003, 'grad_norm': 0.4735756516456604, 'learning_rate': 3.088550673932084e-06, 'epoch': 1.99}

 66%|██████▌   | 8821/13317 [1:30:56<46:41,  1.61it/s]
 66%|██████▌   | 8822/13317 [1:30:57<46:27,  1.61it/s]
                                                      
{'loss': 0.151, 'grad_norm': 0.4534057676792145, 'learning_rate': 3.087339657015789e-06, 'epoch': 1.99}

 66%|██████▌   | 8822/13317 [1:30:57<46:27,  1.61it/s]
 66%|██████▋   | 8823/13317 [1:30:57<46:21,  1.62it/s]
                                                      
{'loss': 0.1934, 'grad_norm': 0.5669113397598267, 'learning_rate': 3.0861287715195123e-06, 'epoch': 1.99}

 66%|██████▋   | 8823/13317 [1:30:57<46:21,  1.62it/s]
 66%|██████▋   | 8824/13317 [1:30:58<46:16,  1.62it/s]
                                                      
{'loss': 0.1633, 'grad_norm': 0.4926477074623108, 'learning_rate': 3.0849180175264538e-06, 'epoch': 1.99}

 66%|██████▋   | 8824/13317 [1:30:58<46:16,  1.62it/s]
 66%|██████▋   | 8825/13317 [1:30:58<46:09,  1.62it/s]
                                                      
{'loss': 0.1911, 'grad_norm': 0.5619789958000183, 'learning_rate': 3.083707395119804e-06, 'epoch': 1.99}

 66%|██████▋   | 8825/13317 [1:30:59<46:09,  1.62it/s]
 66%|██████▋   | 8826/13317 [1:30:59<46:07,  1.62it/s]
                                                      
{'loss': 0.1799, 'grad_norm': 0.5332507491111755, 'learning_rate': 3.082496904382746e-06, 'epoch': 1.99}

 66%|██████▋   | 8826/13317 [1:30:59<46:07,  1.62it/s]
 66%|██████▋   | 8827/13317 [1:31:00<46:02,  1.63it/s]
                                                      
{'loss': 0.2729, 'grad_norm': 0.6107766628265381, 'learning_rate': 3.081286545398455e-06, 'epoch': 1.99}

 66%|██████▋   | 8827/13317 [1:31:00<46:02,  1.63it/s]
 66%|██████▋   | 8828/13317 [1:31:00<46:00,  1.63it/s]
                                                      
{'loss': 0.153, 'grad_norm': 0.46370628476142883, 'learning_rate': 3.0800763182500926e-06, 'epoch': 1.99}

 66%|██████▋   | 8828/13317 [1:31:00<46:00,  1.63it/s]
 66%|██████▋   | 8829/13317 [1:31:01<45:56,  1.63it/s]
                                                      
{'loss': 0.1573, 'grad_norm': 0.4871472716331482, 'learning_rate': 3.078866223020815e-06, 'epoch': 1.99}

 66%|██████▋   | 8829/13317 [1:31:01<45:56,  1.63it/s]
 66%|██████▋   | 8830/13317 [1:31:02<45:53,  1.63it/s]
                                                      
{'loss': 0.1576, 'grad_norm': 0.5101860761642456, 'learning_rate': 3.07765625979377e-06, 'epoch': 1.99}

 66%|██████▋   | 8830/13317 [1:31:02<45:53,  1.63it/s]
 66%|██████▋   | 8831/13317 [1:31:02<45:57,  1.63it/s]
                                                      
{'loss': 0.1758, 'grad_norm': 0.4841567873954773, 'learning_rate': 3.0764464286520944e-06, 'epoch': 1.99}

 66%|██████▋   | 8831/13317 [1:31:02<45:57,  1.63it/s]
 66%|██████▋   | 8832/13317 [1:31:03<45:55,  1.63it/s]
                                                      
{'loss': 0.2149, 'grad_norm': 0.5121442675590515, 'learning_rate': 3.075236729678916e-06, 'epoch': 1.99}

 66%|██████▋   | 8832/13317 [1:31:03<45:55,  1.63it/s]
 66%|██████▋   | 8833/13317 [1:31:03<45:52,  1.63it/s]
                                                      
{'loss': 0.1815, 'grad_norm': 0.46190333366394043, 'learning_rate': 3.0740271629573537e-06, 'epoch': 1.99}

 66%|██████▋   | 8833/13317 [1:31:04<45:52,  1.63it/s]
 66%|██████▋   | 8834/13317 [1:31:04<45:51,  1.63it/s]
                                                      
{'loss': 0.1831, 'grad_norm': 0.5321147441864014, 'learning_rate': 3.0728177285705186e-06, 'epoch': 1.99}

 66%|██████▋   | 8834/13317 [1:31:04<45:51,  1.63it/s]
 66%|██████▋   | 8835/13317 [1:31:05<45:50,  1.63it/s]
                                                      
{'loss': 0.1894, 'grad_norm': 0.5135007500648499, 'learning_rate': 3.07160842660151e-06, 'epoch': 1.99}

 66%|██████▋   | 8835/13317 [1:31:05<45:50,  1.63it/s]
 66%|██████▋   | 8836/13317 [1:31:05<45:49,  1.63it/s]
                                                      
{'loss': 0.2113, 'grad_norm': 0.5288596153259277, 'learning_rate': 3.0703992571334196e-06, 'epoch': 1.99}

 66%|██████▋   | 8836/13317 [1:31:05<45:49,  1.63it/s]
 66%|██████▋   | 8837/13317 [1:31:06<45:47,  1.63it/s]
                                                      
{'loss': 0.1794, 'grad_norm': 0.5220291018486023, 'learning_rate': 3.069190220249333e-06, 'epoch': 1.99}

 66%|██████▋   | 8837/13317 [1:31:06<45:47,  1.63it/s]
 66%|██████▋   | 8838/13317 [1:31:06<45:47,  1.63it/s]
                                                      
{'loss': 0.1316, 'grad_norm': 0.4009779691696167, 'learning_rate': 3.067981316032322e-06, 'epoch': 1.99}

 66%|██████▋   | 8838/13317 [1:31:07<45:47,  1.63it/s]
 66%|██████▋   | 8839/13317 [1:31:07<45:46,  1.63it/s]
                                                      
{'loss': 0.1486, 'grad_norm': 0.40539515018463135, 'learning_rate': 3.0667725445654508e-06, 'epoch': 1.99}

 66%|██████▋   | 8839/13317 [1:31:07<45:46,  1.63it/s]
 66%|██████▋   | 8840/13317 [1:31:08<45:46,  1.63it/s]
                                                      
{'loss': 0.1903, 'grad_norm': 0.5748844146728516, 'learning_rate': 3.0655639059317745e-06, 'epoch': 1.99}

 66%|██████▋   | 8840/13317 [1:31:08<45:46,  1.63it/s]
 66%|██████▋   | 8841/13317 [1:31:08<45:46,  1.63it/s]
                                                      
{'loss': 0.1893, 'grad_norm': 0.4726221263408661, 'learning_rate': 3.0643554002143407e-06, 'epoch': 1.99}

 66%|██████▋   | 8841/13317 [1:31:08<45:46,  1.63it/s]
 66%|██████▋   | 8842/13317 [1:31:09<45:46,  1.63it/s]
                                                      
{'loss': 0.1904, 'grad_norm': 0.5574299693107605, 'learning_rate': 3.0631470274961842e-06, 'epoch': 1.99}

 66%|██████▋   | 8842/13317 [1:31:09<45:46,  1.63it/s]
 66%|██████▋   | 8843/13317 [1:31:10<45:44,  1.63it/s]
                                                      
{'loss': 0.2074, 'grad_norm': 0.5733981132507324, 'learning_rate': 3.0619387878603357e-06, 'epoch': 1.99}

 66%|██████▋   | 8843/13317 [1:31:10<45:44,  1.63it/s]
 66%|██████▋   | 8844/13317 [1:31:10<45:43,  1.63it/s]
                                                      
{'loss': 0.2175, 'grad_norm': 0.6051886081695557, 'learning_rate': 3.0607306813898143e-06, 'epoch': 1.99}

 66%|██████▋   | 8844/13317 [1:31:10<45:43,  1.63it/s]
 66%|██████▋   | 8845/13317 [1:31:11<45:43,  1.63it/s]
                                                      
{'loss': 0.2001, 'grad_norm': 0.44511234760284424, 'learning_rate': 3.0595227081676273e-06, 'epoch': 1.99}

 66%|██████▋   | 8845/13317 [1:31:11<45:43,  1.63it/s]
 66%|██████▋   | 8846/13317 [1:31:11<45:42,  1.63it/s]
                                                      
{'loss': 0.1864, 'grad_norm': 0.49022120237350464, 'learning_rate': 3.0583148682767762e-06, 'epoch': 1.99}

 66%|██████▋   | 8846/13317 [1:31:11<45:42,  1.63it/s]
 66%|██████▋   | 8847/13317 [1:31:12<45:42,  1.63it/s]
                                                      
{'loss': 0.1574, 'grad_norm': 0.4815666377544403, 'learning_rate': 3.0571071618002517e-06, 'epoch': 1.99}

 66%|██████▋   | 8847/13317 [1:31:12<45:42,  1.63it/s]
 66%|██████▋   | 8848/13317 [1:31:13<45:42,  1.63it/s]
                                                      
{'loss': 0.2315, 'grad_norm': 0.6173742413520813, 'learning_rate': 3.0558995888210397e-06, 'epoch': 1.99}

 66%|██████▋   | 8848/13317 [1:31:13<45:42,  1.63it/s]
 66%|██████▋   | 8849/13317 [1:31:13<45:42,  1.63it/s]
                                                      
{'loss': 0.1744, 'grad_norm': 0.46765515208244324, 'learning_rate': 3.054692149422107e-06, 'epoch': 1.99}

 66%|██████▋   | 8849/13317 [1:31:13<45:42,  1.63it/s]
 66%|██████▋   | 8850/13317 [1:31:14<45:42,  1.63it/s]
                                                      
{'loss': 0.1419, 'grad_norm': 0.404977411031723, 'learning_rate': 3.053484843686423e-06, 'epoch': 1.99}

 66%|██████▋   | 8850/13317 [1:31:14<45:42,  1.63it/s]
 66%|██████▋   | 8851/13317 [1:31:14<45:40,  1.63it/s]
                                                      
{'loss': 0.2001, 'grad_norm': 0.4967178404331207, 'learning_rate': 3.0522776716969404e-06, 'epoch': 1.99}

 66%|██████▋   | 8851/13317 [1:31:15<45:40,  1.63it/s]
 66%|██████▋   | 8852/13317 [1:31:15<45:39,  1.63it/s]
                                                      
{'loss': 0.1927, 'grad_norm': 0.5499921441078186, 'learning_rate': 3.0510706335366034e-06, 'epoch': 1.99}

 66%|██████▋   | 8852/13317 [1:31:15<45:39,  1.63it/s]
 66%|██████▋   | 8853/13317 [1:31:16<45:38,  1.63it/s]
                                                      
{'loss': 0.2217, 'grad_norm': 0.5761085152626038, 'learning_rate': 3.0498637292883506e-06, 'epoch': 1.99}

 66%|██████▋   | 8853/13317 [1:31:16<45:38,  1.63it/s]
 66%|██████▋   | 8854/13317 [1:31:16<45:38,  1.63it/s]
                                                      
{'loss': 0.1585, 'grad_norm': 0.438773512840271, 'learning_rate': 3.0486569590351078e-06, 'epoch': 1.99}

 66%|██████▋   | 8854/13317 [1:31:16<45:38,  1.63it/s]
 66%|██████▋   | 8855/13317 [1:31:17<45:36,  1.63it/s]
                                                      
{'loss': 0.1931, 'grad_norm': 0.5607253909111023, 'learning_rate': 3.047450322859793e-06, 'epoch': 1.99}

 66%|██████▋   | 8855/13317 [1:31:17<45:36,  1.63it/s]
 67%|██████▋   | 8856/13317 [1:31:17<45:36,  1.63it/s]
                                                      
{'loss': 0.1784, 'grad_norm': 0.5425475835800171, 'learning_rate': 3.0462438208453125e-06, 'epoch': 2.0}

 67%|██████▋   | 8856/13317 [1:31:18<45:36,  1.63it/s]
 67%|██████▋   | 8857/13317 [1:31:18<45:35,  1.63it/s]
                                                      
{'loss': 0.1437, 'grad_norm': 0.38906899094581604, 'learning_rate': 3.045037453074571e-06, 'epoch': 2.0}

 67%|██████▋   | 8857/13317 [1:31:18<45:35,  1.63it/s]
 67%|██████▋   | 8858/13317 [1:31:19<45:33,  1.63it/s]
                                                      
{'loss': 0.2278, 'grad_norm': 0.6357388496398926, 'learning_rate': 3.0438312196304558e-06, 'epoch': 2.0}

 67%|██████▋   | 8858/13317 [1:31:19<45:33,  1.63it/s]
 67%|██████▋   | 8859/13317 [1:31:19<45:33,  1.63it/s]
                                                      
{'loss': 0.1759, 'grad_norm': 0.46541544795036316, 'learning_rate': 3.0426251205958474e-06, 'epoch': 2.0}

 67%|██████▋   | 8859/13317 [1:31:19<45:33,  1.63it/s]
 67%|██████▋   | 8860/13317 [1:31:20<45:32,  1.63it/s]
                                                      
{'loss': 0.1638, 'grad_norm': 0.5240520238876343, 'learning_rate': 3.0414191560536183e-06, 'epoch': 2.0}

 67%|██████▋   | 8860/13317 [1:31:20<45:32,  1.63it/s]
 67%|██████▋   | 8861/13317 [1:31:21<45:34,  1.63it/s]
                                                      
{'loss': 0.1954, 'grad_norm': 0.5494428277015686, 'learning_rate': 3.040213326086632e-06, 'epoch': 2.0}

 67%|██████▋   | 8861/13317 [1:31:21<45:34,  1.63it/s]
 67%|██████▋   | 8862/13317 [1:31:21<45:33,  1.63it/s]
                                                      
{'loss': 0.1835, 'grad_norm': 0.4915767312049866, 'learning_rate': 3.0390076307777393e-06, 'epoch': 2.0}

 67%|██████▋   | 8862/13317 [1:31:21<45:33,  1.63it/s]
 67%|██████▋   | 8863/13317 [1:31:22<45:33,  1.63it/s]
                                                      
{'loss': 0.1816, 'grad_norm': 0.46008753776550293, 'learning_rate': 3.0378020702097847e-06, 'epoch': 2.0}

 67%|██████▋   | 8863/13317 [1:31:22<45:33,  1.63it/s]
 67%|██████▋   | 8864/13317 [1:31:22<45:31,  1.63it/s]
                                                      
{'loss': 0.1448, 'grad_norm': 0.4309210181236267, 'learning_rate': 3.036596644465606e-06, 'epoch': 2.0}

 67%|██████▋   | 8864/13317 [1:31:23<45:31,  1.63it/s]
 67%|██████▋   | 8865/13317 [1:31:23<45:31,  1.63it/s]
                                                      
{'loss': 0.1693, 'grad_norm': 0.4568261206150055, 'learning_rate': 3.0353913536280267e-06, 'epoch': 2.0}

 67%|██████▋   | 8865/13317 [1:31:23<45:31,  1.63it/s]
 67%|██████▋   | 8866/13317 [1:31:24<45:30,  1.63it/s]
                                                      
{'loss': 0.2358, 'grad_norm': 0.5993092060089111, 'learning_rate': 3.034186197779862e-06, 'epoch': 2.0}

 67%|██████▋   | 8866/13317 [1:31:24<45:30,  1.63it/s]
 67%|██████▋   | 8867/13317 [1:31:24<45:29,  1.63it/s]
                                                      
{'loss': 0.2189, 'grad_norm': 0.6480884552001953, 'learning_rate': 3.0329811770039203e-06, 'epoch': 2.0}

 67%|██████▋   | 8867/13317 [1:31:24<45:29,  1.63it/s]
 67%|██████▋   | 8868/13317 [1:31:25<45:30,  1.63it/s]
                                                      
{'loss': 0.1602, 'grad_norm': 0.46584638953208923, 'learning_rate': 3.0317762913829995e-06, 'epoch': 2.0}

 67%|██████▋   | 8868/13317 [1:31:25<45:30,  1.63it/s]
 67%|██████▋   | 8869/13317 [1:31:25<45:29,  1.63it/s]
                                                      
{'loss': 0.1807, 'grad_norm': 0.4698997139930725, 'learning_rate': 3.030571540999886e-06, 'epoch': 2.0}

 67%|██████▋   | 8869/13317 [1:31:26<45:29,  1.63it/s]
 67%|██████▋   | 8870/13317 [1:31:26<45:28,  1.63it/s]
                                                      
{'loss': 0.217, 'grad_norm': 0.596944272518158, 'learning_rate': 3.0293669259373605e-06, 'epoch': 2.0}

 67%|██████▋   | 8870/13317 [1:31:26<45:28,  1.63it/s]
 67%|██████▋   | 8871/13317 [1:31:27<45:28,  1.63it/s]
                                                      
{'loss': 0.1821, 'grad_norm': 0.5044851899147034, 'learning_rate': 3.028162446278194e-06, 'epoch': 2.0}

 67%|██████▋   | 8871/13317 [1:31:27<45:28,  1.63it/s]
 67%|██████▋   | 8872/13317 [1:31:27<45:27,  1.63it/s]
                                                      
{'loss': 0.225, 'grad_norm': 0.5907055735588074, 'learning_rate': 3.026958102105144e-06, 'epoch': 2.0}

 67%|██████▋   | 8872/13317 [1:31:27<45:27,  1.63it/s]
 67%|██████▋   | 8873/13317 [1:31:28<45:26,  1.63it/s]
                                                      
{'loss': 0.1645, 'grad_norm': 0.48345640301704407, 'learning_rate': 3.0257538935009643e-06, 'epoch': 2.0}

 67%|██████▋   | 8873/13317 [1:31:28<45:26,  1.63it/s]
 67%|██████▋   | 8874/13317 [1:31:29<45:26,  1.63it/s]
                                                      
{'loss': 0.1702, 'grad_norm': 0.4663739502429962, 'learning_rate': 3.024549820548394e-06, 'epoch': 2.0}

 67%|██████▋   | 8874/13317 [1:31:29<45:26,  1.63it/s]
 67%|██████▋   | 8875/13317 [1:31:29<45:24,  1.63it/s]
                                                      
{'loss': 0.1578, 'grad_norm': 0.4371885359287262, 'learning_rate': 3.023345883330171e-06, 'epoch': 2.0}

 67%|██████▋   | 8875/13317 [1:31:29<45:24,  1.63it/s]
 67%|██████▋   | 8876/13317 [1:31:30<45:23,  1.63it/s]
                                                      
{'loss': 0.2518, 'grad_norm': 0.6041123270988464, 'learning_rate': 3.0221420819290117e-06, 'epoch': 2.0}

 67%|██████▋   | 8876/13317 [1:31:30<45:23,  1.63it/s]
 67%|██████▋   | 8877/13317 [1:31:30<45:25,  1.63it/s]
                                                      
{'loss': 0.1913, 'grad_norm': 0.509492814540863, 'learning_rate': 3.020938416427634e-06, 'epoch': 2.0}

 67%|██████▋   | 8877/13317 [1:31:31<45:25,  1.63it/s]
 67%|██████▋   | 8878/13317 [1:31:31<45:20,  1.63it/s]
                                                      
{'loss': 0.162, 'grad_norm': 0.45143938064575195, 'learning_rate': 3.019734886908743e-06, 'epoch': 2.0}

 67%|██████▋   | 8878/13317 [1:31:31<45:20,  1.63it/s]
 67%|██████▋   | 8879/13317 [1:31:32<45:26,  1.63it/s]
                                                      
{'loss': 0.1728, 'grad_norm': 0.46946343779563904, 'learning_rate': 3.0185314934550314e-06, 'epoch': 2.0}

 67%|██████▋   | 8879/13317 [1:31:32<45:26,  1.63it/s]
 67%|██████▋   | 8880/13317 [1:31:32<45:24,  1.63it/s]
                                                      
{'loss': 0.1947, 'grad_norm': 0.4505583345890045, 'learning_rate': 3.017328236149187e-06, 'epoch': 2.0}

 67%|██████▋   | 8880/13317 [1:31:32<45:24,  1.63it/s]
 67%|██████▋   | 8881/13317 [1:31:33<46:26,  1.59it/s]
                                                      
{'loss': 0.1666, 'grad_norm': 0.44451576471328735, 'learning_rate': 3.016125115073884e-06, 'epoch': 2.0}

 67%|██████▋   | 8881/13317 [1:31:33<46:26,  1.59it/s]
 67%|██████▋   | 8882/13317 [1:31:33<46:17,  1.60it/s]
                                                      
{'loss': 0.1653, 'grad_norm': 0.42785146832466125, 'learning_rate': 3.0149221303117947e-06, 'epoch': 2.0}

 67%|██████▋   | 8882/13317 [1:31:34<46:17,  1.60it/s]
 67%|██████▋   | 8883/13317 [1:31:34<46:02,  1.61it/s]
                                                      
{'loss': 0.1473, 'grad_norm': 0.40036851167678833, 'learning_rate': 3.0137192819455696e-06, 'epoch': 2.0}

 67%|██████▋   | 8883/13317 [1:31:34<46:02,  1.61it/s]
 67%|██████▋   | 8884/13317 [1:31:35<45:50,  1.61it/s]
                                                      
{'loss': 0.1476, 'grad_norm': 0.42862802743911743, 'learning_rate': 3.012516570057863e-06, 'epoch': 2.0}

 67%|██████▋   | 8884/13317 [1:31:35<45:50,  1.61it/s]
 67%|██████▋   | 8885/13317 [1:31:35<45:41,  1.62it/s]
                                                      
{'loss': 0.1098, 'grad_norm': 0.3515554964542389, 'learning_rate': 3.0113139947313117e-06, 'epoch': 2.0}

 67%|██████▋   | 8885/13317 [1:31:35<45:41,  1.62it/s]
 67%|██████▋   | 8886/13317 [1:31:36<45:35,  1.62it/s]
                                                      
{'loss': 0.1727, 'grad_norm': 0.47751978039741516, 'learning_rate': 3.010111556048545e-06, 'epoch': 2.0}

 67%|██████▋   | 8886/13317 [1:31:36<45:35,  1.62it/s]
 67%|██████▋   | 8887/13317 [1:31:37<45:30,  1.62it/s]
                                                      
{'loss': 0.1732, 'grad_norm': 0.43321511149406433, 'learning_rate': 3.008909254092184e-06, 'epoch': 2.0}

 67%|██████▋   | 8887/13317 [1:31:37<45:30,  1.62it/s]
 67%|██████▋   | 8888/13317 [1:31:37<45:27,  1.62it/s]
                                                      
{'loss': 0.1494, 'grad_norm': 0.4313478469848633, 'learning_rate': 3.00770708894484e-06, 'epoch': 2.0}

 67%|██████▋   | 8888/13317 [1:31:37<45:27,  1.62it/s]
 67%|██████▋   | 8889/13317 [1:31:38<45:24,  1.62it/s]
                                                      
{'loss': 0.1664, 'grad_norm': 0.4537353217601776, 'learning_rate': 3.006505060689112e-06, 'epoch': 2.0}

 67%|██████▋   | 8889/13317 [1:31:38<45:24,  1.62it/s]
 67%|██████▋   | 8890/13317 [1:31:38<45:22,  1.63it/s]
                                                      
{'loss': 0.1639, 'grad_norm': 0.4590117335319519, 'learning_rate': 3.005303169407594e-06, 'epoch': 2.0}

 67%|██████▋   | 8890/13317 [1:31:39<45:22,  1.63it/s]
 67%|██████▋   | 8891/13317 [1:31:39<45:23,  1.63it/s]
                                                      
{'loss': 0.1566, 'grad_norm': 0.446291983127594, 'learning_rate': 3.0041014151828684e-06, 'epoch': 2.0}

 67%|██████▋   | 8891/13317 [1:31:39<45:23,  1.63it/s]
 67%|██████▋   | 8892/13317 [1:31:40<45:20,  1.63it/s]
                                                      
{'loss': 0.1469, 'grad_norm': 0.44301214814186096, 'learning_rate': 3.0028997980975104e-06, 'epoch': 2.0}

 67%|██████▋   | 8892/13317 [1:31:40<45:20,  1.63it/s]
 67%|██████▋   | 8893/13317 [1:31:40<45:22,  1.62it/s]
                                                      
{'loss': 0.1278, 'grad_norm': 0.4054727256298065, 'learning_rate': 3.001698318234081e-06, 'epoch': 2.0}

 67%|██████▋   | 8893/13317 [1:31:40<45:22,  1.62it/s]
 67%|██████▋   | 8894/13317 [1:31:41<45:20,  1.63it/s]
                                                      
{'loss': 0.1727, 'grad_norm': 0.6490105390548706, 'learning_rate': 3.0004969756751344e-06, 'epoch': 2.0}

 67%|██████▋   | 8894/13317 [1:31:41<45:20,  1.63it/s]
 67%|██████▋   | 8895/13317 [1:31:41<45:17,  1.63it/s]
                                                      
{'loss': 0.173, 'grad_norm': 0.521726667881012, 'learning_rate': 2.999295770503219e-06, 'epoch': 2.0}

 67%|██████▋   | 8895/13317 [1:31:42<45:17,  1.63it/s]
 67%|██████▋   | 8896/13317 [1:31:42<45:19,  1.63it/s]
                                                      
{'loss': 0.1522, 'grad_norm': 0.4607597291469574, 'learning_rate': 2.9980947028008643e-06, 'epoch': 2.0}

 67%|██████▋   | 8896/13317 [1:31:42<45:19,  1.63it/s]
 67%|██████▋   | 8897/13317 [1:31:43<45:15,  1.63it/s]
                                                      
{'loss': 0.1333, 'grad_norm': 0.4046030640602112, 'learning_rate': 2.996893772650602e-06, 'epoch': 2.0}

 67%|██████▋   | 8897/13317 [1:31:43<45:15,  1.63it/s]
 67%|██████▋   | 8898/13317 [1:31:43<45:16,  1.63it/s]
                                                      
{'loss': 0.1745, 'grad_norm': 0.513903796672821, 'learning_rate': 2.9956929801349484e-06, 'epoch': 2.0}

 67%|██████▋   | 8898/13317 [1:31:43<45:16,  1.63it/s]
 67%|██████▋   | 8899/13317 [1:31:44<45:13,  1.63it/s]
                                                      
{'loss': 0.1317, 'grad_norm': 0.38937535881996155, 'learning_rate': 2.9944923253364066e-06, 'epoch': 2.0}

 67%|██████▋   | 8899/13317 [1:31:44<45:13,  1.63it/s]
 67%|██████▋   | 8900/13317 [1:31:45<45:12,  1.63it/s]
                                                      
{'loss': 0.121, 'grad_norm': 0.3933415710926056, 'learning_rate': 2.993291808337478e-06, 'epoch': 2.0}

 67%|██████▋   | 8900/13317 [1:31:45<45:12,  1.63it/s]
 67%|██████▋   | 8901/13317 [1:31:45<45:11,  1.63it/s]
                                                      
{'loss': 0.126, 'grad_norm': 0.3665746748447418, 'learning_rate': 2.9920914292206477e-06, 'epoch': 2.01}

 67%|██████▋   | 8901/13317 [1:31:45<45:11,  1.63it/s]
 67%|██████▋   | 8902/13317 [1:31:46<45:11,  1.63it/s]
                                                      
{'loss': 0.1501, 'grad_norm': 0.5246037244796753, 'learning_rate': 2.9908911880684e-06, 'epoch': 2.01}

 67%|██████▋   | 8902/13317 [1:31:46<45:11,  1.63it/s]
 67%|██████▋   | 8903/13317 [1:31:46<45:16,  1.62it/s]
                                                      
{'loss': 0.1708, 'grad_norm': 0.5069417953491211, 'learning_rate': 2.9896910849631965e-06, 'epoch': 2.01}

 67%|██████▋   | 8903/13317 [1:31:47<45:16,  1.62it/s]
 67%|██████▋   | 8904/13317 [1:31:47<45:14,  1.63it/s]
                                                      
{'loss': 0.2195, 'grad_norm': 0.6269458532333374, 'learning_rate': 2.9884911199875027e-06, 'epoch': 2.01}

 67%|██████▋   | 8904/13317 [1:31:47<45:14,  1.63it/s]
 67%|██████▋   | 8905/13317 [1:31:48<45:11,  1.63it/s]
                                                      
{'loss': 0.1754, 'grad_norm': 0.5233111381530762, 'learning_rate': 2.987291293223768e-06, 'epoch': 2.01}

 67%|██████▋   | 8905/13317 [1:31:48<45:11,  1.63it/s]
 67%|██████▋   | 8906/13317 [1:31:48<45:09,  1.63it/s]
                                                      
{'loss': 0.1265, 'grad_norm': 0.42482081055641174, 'learning_rate': 2.9860916047544308e-06, 'epoch': 2.01}

 67%|██████▋   | 8906/13317 [1:31:48<45:09,  1.63it/s]
 67%|██████▋   | 8907/13317 [1:31:49<45:07,  1.63it/s]
                                                      
{'loss': 0.1638, 'grad_norm': 0.4665152132511139, 'learning_rate': 2.9848920546619247e-06, 'epoch': 2.01}

 67%|██████▋   | 8907/13317 [1:31:49<45:07,  1.63it/s]
 67%|██████▋   | 8908/13317 [1:31:49<45:09,  1.63it/s]
                                                      
{'loss': 0.1282, 'grad_norm': 0.44795548915863037, 'learning_rate': 2.9836926430286695e-06, 'epoch': 2.01}

 67%|██████▋   | 8908/13317 [1:31:50<45:09,  1.63it/s]
 67%|██████▋   | 8909/13317 [1:31:50<45:09,  1.63it/s]
                                                      
{'loss': 0.1534, 'grad_norm': 0.5217428207397461, 'learning_rate': 2.9824933699370817e-06, 'epoch': 2.01}

 67%|██████▋   | 8909/13317 [1:31:50<45:09,  1.63it/s]
 67%|██████▋   | 8910/13317 [1:31:51<45:08,  1.63it/s]
                                                      
{'loss': 0.1357, 'grad_norm': 0.41153472661972046, 'learning_rate': 2.9812942354695583e-06, 'epoch': 2.01}

 67%|██████▋   | 8910/13317 [1:31:51<45:08,  1.63it/s]
 67%|██████▋   | 8911/13317 [1:31:51<45:09,  1.63it/s]
                                                      
{'loss': 0.1307, 'grad_norm': 0.4180853068828583, 'learning_rate': 2.980095239708496e-06, 'epoch': 2.01}

 67%|██████▋   | 8911/13317 [1:31:51<45:09,  1.63it/s]
 67%|██████▋   | 8912/13317 [1:31:52<45:07,  1.63it/s]
                                                      
{'loss': 0.1511, 'grad_norm': 0.45283764600753784, 'learning_rate': 2.978896382736279e-06, 'epoch': 2.01}

 67%|██████▋   | 8912/13317 [1:31:52<45:07,  1.63it/s]
 67%|██████▋   | 8913/13317 [1:31:53<45:11,  1.62it/s]
                                                      
{'loss': 0.1363, 'grad_norm': 0.43553826212882996, 'learning_rate': 2.9776976646352797e-06, 'epoch': 2.01}

 67%|██████▋   | 8913/13317 [1:31:53<45:11,  1.62it/s]
 67%|██████▋   | 8914/13317 [1:31:53<45:09,  1.63it/s]
                                                      
{'loss': 0.1513, 'grad_norm': 0.4710219204425812, 'learning_rate': 2.9764990854878624e-06, 'epoch': 2.01}

 67%|██████▋   | 8914/13317 [1:31:53<45:09,  1.63it/s]
 67%|██████▋   | 8915/13317 [1:31:54<45:06,  1.63it/s]
                                                      
{'loss': 0.1534, 'grad_norm': 0.5057545900344849, 'learning_rate': 2.9753006453763844e-06, 'epoch': 2.01}

 67%|██████▋   | 8915/13317 [1:31:54<45:06,  1.63it/s]
 67%|██████▋   | 8916/13317 [1:31:54<45:04,  1.63it/s]
                                                      
{'loss': 0.1478, 'grad_norm': 0.48622608184814453, 'learning_rate': 2.9741023443831897e-06, 'epoch': 2.01}

 67%|██████▋   | 8916/13317 [1:31:55<45:04,  1.63it/s]
 67%|██████▋   | 8917/13317 [1:31:55<45:03,  1.63it/s]
                                                      
{'loss': 0.1456, 'grad_norm': 0.5491385459899902, 'learning_rate': 2.9729041825906124e-06, 'epoch': 2.01}

 67%|██████▋   | 8917/13317 [1:31:55<45:03,  1.63it/s]
 67%|██████▋   | 8918/13317 [1:31:56<45:03,  1.63it/s]
                                                      
{'loss': 0.1163, 'grad_norm': 0.4501831531524658, 'learning_rate': 2.971706160080983e-06, 'epoch': 2.01}

 67%|██████▋   | 8918/13317 [1:31:56<45:03,  1.63it/s]
 67%|██████▋   | 8919/13317 [1:31:56<45:03,  1.63it/s]
                                                      
{'loss': 0.1441, 'grad_norm': 0.47125619649887085, 'learning_rate': 2.970508276936616e-06, 'epoch': 2.01}

 67%|██████▋   | 8919/13317 [1:31:56<45:03,  1.63it/s]
 67%|██████▋   | 8920/13317 [1:31:57<45:02,  1.63it/s]
                                                      
{'loss': 0.1548, 'grad_norm': 0.5453885197639465, 'learning_rate': 2.969310533239819e-06, 'epoch': 2.01}

 67%|██████▋   | 8920/13317 [1:31:57<45:02,  1.63it/s]
 67%|██████▋   | 8921/13317 [1:31:57<45:00,  1.63it/s]
                                                      
{'loss': 0.1585, 'grad_norm': 0.5243396162986755, 'learning_rate': 2.9681129290728882e-06, 'epoch': 2.01}

 67%|██████▋   | 8921/13317 [1:31:58<45:00,  1.63it/s]
 67%|██████▋   | 8922/13317 [1:31:58<44:57,  1.63it/s]
                                                      
{'loss': 0.1671, 'grad_norm': 0.5578039288520813, 'learning_rate': 2.9669154645181166e-06, 'epoch': 2.01}

 67%|██████▋   | 8922/13317 [1:31:58<44:57,  1.63it/s]
 67%|██████▋   | 8923/13317 [1:31:59<44:58,  1.63it/s]
                                                      
{'loss': 0.1715, 'grad_norm': 0.5466580390930176, 'learning_rate': 2.9657181396577762e-06, 'epoch': 2.01}

 67%|██████▋   | 8923/13317 [1:31:59<44:58,  1.63it/s]
 67%|██████▋   | 8924/13317 [1:31:59<44:55,  1.63it/s]
                                                      
{'loss': 0.1364, 'grad_norm': 0.46838587522506714, 'learning_rate': 2.9645209545741394e-06, 'epoch': 2.01}

 67%|██████▋   | 8924/13317 [1:31:59<44:55,  1.63it/s]
 67%|██████▋   | 8925/13317 [1:32:00<44:55,  1.63it/s]
                                                      
{'loss': 0.14, 'grad_norm': 0.4352295696735382, 'learning_rate': 2.9633239093494653e-06, 'epoch': 2.01}

 67%|██████▋   | 8925/13317 [1:32:00<44:55,  1.63it/s]
 67%|██████▋   | 8926/13317 [1:32:01<44:54,  1.63it/s]
                                                      
{'loss': 0.1321, 'grad_norm': 0.4918535351753235, 'learning_rate': 2.9621270040660043e-06, 'epoch': 2.01}

 67%|██████▋   | 8926/13317 [1:32:01<44:54,  1.63it/s]
 67%|██████▋   | 8927/13317 [1:32:01<44:53,  1.63it/s]
                                                      
{'loss': 0.1572, 'grad_norm': 0.6687697172164917, 'learning_rate': 2.9609302388059947e-06, 'epoch': 2.01}

 67%|██████▋   | 8927/13317 [1:32:01<44:53,  1.63it/s]
 67%|██████▋   | 8928/13317 [1:32:02<44:51,  1.63it/s]
                                                      
{'loss': 0.1614, 'grad_norm': 0.5483740568161011, 'learning_rate': 2.959733613651666e-06, 'epoch': 2.01}

 67%|██████▋   | 8928/13317 [1:32:02<44:51,  1.63it/s]
 67%|██████▋   | 8929/13317 [1:32:02<44:51,  1.63it/s]
                                                      
{'loss': 0.1992, 'grad_norm': 0.6728695034980774, 'learning_rate': 2.9585371286852447e-06, 'epoch': 2.01}

 67%|██████▋   | 8929/13317 [1:32:03<44:51,  1.63it/s]
 67%|██████▋   | 8930/13317 [1:32:03<44:51,  1.63it/s]
                                                      
{'loss': 0.1912, 'grad_norm': 0.5527303814888, 'learning_rate': 2.9573407839889345e-06, 'epoch': 2.01}

 67%|██████▋   | 8930/13317 [1:32:03<44:51,  1.63it/s]
 67%|██████▋   | 8931/13317 [1:32:04<44:50,  1.63it/s]
                                                      
{'loss': 0.1373, 'grad_norm': 0.44865304231643677, 'learning_rate': 2.956144579644942e-06, 'epoch': 2.01}

 67%|██████▋   | 8931/13317 [1:32:04<44:50,  1.63it/s]
 67%|██████▋   | 8932/13317 [1:32:04<44:50,  1.63it/s]
                                                      
{'loss': 0.1535, 'grad_norm': 0.5422680974006653, 'learning_rate': 2.9549485157354573e-06, 'epoch': 2.01}

 67%|██████▋   | 8932/13317 [1:32:04<44:50,  1.63it/s]
 67%|██████▋   | 8933/13317 [1:32:05<44:49,  1.63it/s]
                                                      
{'loss': 0.1316, 'grad_norm': 0.46315205097198486, 'learning_rate': 2.953752592342663e-06, 'epoch': 2.01}

 67%|██████▋   | 8933/13317 [1:32:05<44:49,  1.63it/s]
 67%|██████▋   | 8934/13317 [1:32:05<44:52,  1.63it/s]
                                                      
{'loss': 0.1471, 'grad_norm': 0.561182975769043, 'learning_rate': 2.952556809548731e-06, 'epoch': 2.01}

 67%|██████▋   | 8934/13317 [1:32:06<44:52,  1.63it/s]
 67%|██████▋   | 8935/13317 [1:32:06<44:52,  1.63it/s]
                                                      
{'loss': 0.1598, 'grad_norm': 0.590100109577179, 'learning_rate': 2.9513611674358244e-06, 'epoch': 2.01}

 67%|██████▋   | 8935/13317 [1:32:06<44:52,  1.63it/s]
 67%|██████▋   | 8936/13317 [1:32:07<44:51,  1.63it/s]
                                                      
{'loss': 0.1309, 'grad_norm': 0.500572681427002, 'learning_rate': 2.9501656660861e-06, 'epoch': 2.01}

 67%|██████▋   | 8936/13317 [1:32:07<44:51,  1.63it/s]
 67%|██████▋   | 8937/13317 [1:32:07<44:50,  1.63it/s]
                                                      
{'loss': 0.1722, 'grad_norm': 0.4859059154987335, 'learning_rate': 2.948970305581695e-06, 'epoch': 2.01}

 67%|██████▋   | 8937/13317 [1:32:07<44:50,  1.63it/s]
 67%|██████▋   | 8938/13317 [1:32:08<44:48,  1.63it/s]
                                                      
{'loss': 0.1975, 'grad_norm': 0.6015129089355469, 'learning_rate': 2.947775086004747e-06, 'epoch': 2.01}

 67%|██████▋   | 8938/13317 [1:32:08<44:48,  1.63it/s]
 67%|██████▋   | 8939/13317 [1:32:09<44:45,  1.63it/s]
                                                      
{'loss': 0.1476, 'grad_norm': 0.5200459361076355, 'learning_rate': 2.946580007437382e-06, 'epoch': 2.01}

 67%|██████▋   | 8939/13317 [1:32:09<44:45,  1.63it/s]
 67%|██████▋   | 8940/13317 [1:32:09<44:45,  1.63it/s]
                                                      
{'loss': 0.1485, 'grad_norm': 0.46508464217185974, 'learning_rate': 2.9453850699617116e-06, 'epoch': 2.01}

 67%|██████▋   | 8940/13317 [1:32:09<44:45,  1.63it/s]
 67%|██████▋   | 8941/13317 [1:32:10<44:45,  1.63it/s]
                                                      
{'loss': 0.1446, 'grad_norm': 0.49875015020370483, 'learning_rate': 2.9441902736598402e-06, 'epoch': 2.01}

 67%|██████▋   | 8941/13317 [1:32:10<44:45,  1.63it/s]
 67%|██████▋   | 8942/13317 [1:32:10<44:44,  1.63it/s]
                                                      
{'loss': 0.1648, 'grad_norm': 0.5731536149978638, 'learning_rate': 2.9429956186138643e-06, 'epoch': 2.01}

 67%|██████▋   | 8942/13317 [1:32:10<44:44,  1.63it/s]
 67%|██████▋   | 8943/13317 [1:32:11<44:43,  1.63it/s]
                                                      
{'loss': 0.1596, 'grad_norm': 0.5858882665634155, 'learning_rate': 2.941801104905871e-06, 'epoch': 2.01}

 67%|██████▋   | 8943/13317 [1:32:11<44:43,  1.63it/s]
 67%|██████▋   | 8944/13317 [1:32:12<44:43,  1.63it/s]
                                                      
{'loss': 0.1416, 'grad_norm': 0.5021359324455261, 'learning_rate': 2.940606732617933e-06, 'epoch': 2.01}

 67%|██████▋   | 8944/13317 [1:32:12<44:43,  1.63it/s]
 67%|██████▋   | 8945/13317 [1:32:12<44:42,  1.63it/s]
                                                      
{'loss': 0.1415, 'grad_norm': 0.5052130818367004, 'learning_rate': 2.9394125018321185e-06, 'epoch': 2.02}

 67%|██████▋   | 8945/13317 [1:32:12<44:42,  1.63it/s]
 67%|██████▋   | 8946/13317 [1:32:13<44:41,  1.63it/s]
                                                      
{'loss': 0.1489, 'grad_norm': 0.513068437576294, 'learning_rate': 2.9382184126304834e-06, 'epoch': 2.02}

 67%|██████▋   | 8946/13317 [1:32:13<44:41,  1.63it/s]
 67%|██████▋   | 8947/13317 [1:32:13<44:39,  1.63it/s]
                                                      
{'loss': 0.1514, 'grad_norm': 0.5112431049346924, 'learning_rate': 2.937024465095073e-06, 'epoch': 2.02}

 67%|██████▋   | 8947/13317 [1:32:14<44:39,  1.63it/s]
 67%|██████▋   | 8948/13317 [1:32:14<44:39,  1.63it/s]
                                                      
{'loss': 0.1771, 'grad_norm': 0.6108816862106323, 'learning_rate': 2.935830659307924e-06, 'epoch': 2.02}

 67%|██████▋   | 8948/13317 [1:32:14<44:39,  1.63it/s]
 67%|██████▋   | 8949/13317 [1:32:15<44:39,  1.63it/s]
                                                      
{'loss': 0.1374, 'grad_norm': 0.4947526156902313, 'learning_rate': 2.934636995351068e-06, 'epoch': 2.02}

 67%|██████▋   | 8949/13317 [1:32:15<44:39,  1.63it/s]
 67%|██████▋   | 8950/13317 [1:32:15<44:38,  1.63it/s]
                                                      
{'loss': 0.1282, 'grad_norm': 0.43129515647888184, 'learning_rate': 2.9334434733065147e-06, 'epoch': 2.02}

 67%|██████▋   | 8950/13317 [1:32:15<44:38,  1.63it/s]
 67%|██████▋   | 8951/13317 [1:32:16<44:38,  1.63it/s]
                                                      
{'loss': 0.1659, 'grad_norm': 0.5725809931755066, 'learning_rate': 2.9322500932562774e-06, 'epoch': 2.02}

 67%|██████▋   | 8951/13317 [1:32:16<44:38,  1.63it/s]
 67%|██████▋   | 8952/13317 [1:32:16<44:37,  1.63it/s]
                                                      
{'loss': 0.1138, 'grad_norm': 0.4648684859275818, 'learning_rate': 2.9310568552823526e-06, 'epoch': 2.02}

 67%|██████▋   | 8952/13317 [1:32:17<44:37,  1.63it/s]
 67%|██████▋   | 8953/13317 [1:32:17<44:39,  1.63it/s]
                                                      
{'loss': 0.1355, 'grad_norm': 0.48715558648109436, 'learning_rate': 2.929863759466729e-06, 'epoch': 2.02}

 67%|██████▋   | 8953/13317 [1:32:17<44:39,  1.63it/s]
 67%|██████▋   | 8954/13317 [1:32:18<45:27,  1.60it/s]
                                                      
{'loss': 0.1311, 'grad_norm': 0.5032682418823242, 'learning_rate': 2.9286708058913826e-06, 'epoch': 2.02}

 67%|██████▋   | 8954/13317 [1:32:18<45:27,  1.60it/s]
 67%|██████▋   | 8955/13317 [1:32:18<46:34,  1.56it/s]
                                                      
{'loss': 0.1311, 'grad_norm': 0.4624532163143158, 'learning_rate': 2.927477994638283e-06, 'epoch': 2.02}

 67%|██████▋   | 8955/13317 [1:32:19<46:34,  1.56it/s]
 67%|██████▋   | 8956/13317 [1:32:19<47:18,  1.54it/s]
                                                      
{'loss': 0.1479, 'grad_norm': 0.5182405710220337, 'learning_rate': 2.926285325789392e-06, 'epoch': 2.02}

 67%|██████▋   | 8956/13317 [1:32:19<47:18,  1.54it/s]
 67%|██████▋   | 8957/13317 [1:32:20<47:33,  1.53it/s]
                                                      
{'loss': 0.1414, 'grad_norm': 0.5005488395690918, 'learning_rate': 2.925092799426653e-06, 'epoch': 2.02}

 67%|██████▋   | 8957/13317 [1:32:20<47:33,  1.53it/s]
 67%|██████▋   | 8958/13317 [1:32:20<46:39,  1.56it/s]
                                                      
{'loss': 0.1256, 'grad_norm': 0.41922059655189514, 'learning_rate': 2.9239004156320094e-06, 'epoch': 2.02}

 67%|██████▋   | 8958/13317 [1:32:21<46:39,  1.56it/s]
 67%|██████▋   | 8959/13317 [1:32:21<46:00,  1.58it/s]
                                                      
{'loss': 0.1478, 'grad_norm': 0.647802472114563, 'learning_rate': 2.922708174487391e-06, 'epoch': 2.02}

 67%|██████▋   | 8959/13317 [1:32:21<46:00,  1.58it/s]
 67%|██████▋   | 8960/13317 [1:32:22<45:33,  1.59it/s]
                                                      
{'loss': 0.1381, 'grad_norm': 0.5391071438789368, 'learning_rate': 2.9215160760747136e-06, 'epoch': 2.02}

 67%|██████▋   | 8960/13317 [1:32:22<45:33,  1.59it/s]
 67%|██████▋   | 8961/13317 [1:32:22<45:18,  1.60it/s]
                                                      
{'loss': 0.1518, 'grad_norm': 0.5053214430809021, 'learning_rate': 2.920324120475889e-06, 'epoch': 2.02}

 67%|██████▋   | 8961/13317 [1:32:22<45:18,  1.60it/s]
 67%|██████▋   | 8962/13317 [1:32:23<45:06,  1.61it/s]
                                                      
{'loss': 0.1943, 'grad_norm': 0.6203603744506836, 'learning_rate': 2.9191323077728173e-06, 'epoch': 2.02}

 67%|██████▋   | 8962/13317 [1:32:23<45:06,  1.61it/s]
 67%|██████▋   | 8963/13317 [1:32:23<44:55,  1.62it/s]
                                                      
{'loss': 0.2145, 'grad_norm': 0.5586323142051697, 'learning_rate': 2.91794063804739e-06, 'epoch': 2.02}

 67%|██████▋   | 8963/13317 [1:32:24<44:55,  1.62it/s]
 67%|██████▋   | 8964/13317 [1:32:24<44:47,  1.62it/s]
                                                      
{'loss': 0.1628, 'grad_norm': 0.5787076950073242, 'learning_rate': 2.916749111381486e-06, 'epoch': 2.02}

 67%|██████▋   | 8964/13317 [1:32:24<44:47,  1.62it/s]
 67%|██████▋   | 8965/13317 [1:32:25<44:41,  1.62it/s]
                                                      
{'loss': 0.1878, 'grad_norm': 0.5815331339836121, 'learning_rate': 2.9155577278569747e-06, 'epoch': 2.02}

 67%|██████▋   | 8965/13317 [1:32:25<44:41,  1.62it/s]
 67%|██████▋   | 8966/13317 [1:32:25<44:40,  1.62it/s]
                                                      
{'loss': 0.1721, 'grad_norm': 0.5321115851402283, 'learning_rate': 2.9143664875557193e-06, 'epoch': 2.02}

 67%|██████▋   | 8966/13317 [1:32:25<44:40,  1.62it/s]
 67%|██████▋   | 8967/13317 [1:32:26<44:39,  1.62it/s]
                                                      
{'loss': 0.1529, 'grad_norm': 0.5929206609725952, 'learning_rate': 2.913175390559567e-06, 'epoch': 2.02}

 67%|██████▋   | 8967/13317 [1:32:26<44:39,  1.62it/s]
 67%|██████▋   | 8968/13317 [1:32:27<44:35,  1.63it/s]
                                                      
{'loss': 0.1668, 'grad_norm': 0.5728535056114197, 'learning_rate': 2.9119844369503635e-06, 'epoch': 2.02}

 67%|██████▋   | 8968/13317 [1:32:27<44:35,  1.63it/s]
 67%|██████▋   | 8969/13317 [1:32:27<44:33,  1.63it/s]
                                                      
{'loss': 0.1388, 'grad_norm': 0.48125553131103516, 'learning_rate': 2.910793626809935e-06, 'epoch': 2.02}

 67%|██████▋   | 8969/13317 [1:32:27<44:33,  1.63it/s]
 67%|██████▋   | 8970/13317 [1:32:28<44:30,  1.63it/s]
                                                      
{'loss': 0.1779, 'grad_norm': 0.5941422581672668, 'learning_rate': 2.9096029602201077e-06, 'epoch': 2.02}

 67%|██████▋   | 8970/13317 [1:32:28<44:30,  1.63it/s]
 67%|██████▋   | 8971/13317 [1:32:28<44:32,  1.63it/s]
                                                      
{'loss': 0.1818, 'grad_norm': 0.5897356867790222, 'learning_rate': 2.90841243726269e-06, 'epoch': 2.02}

 67%|██████▋   | 8971/13317 [1:32:28<44:32,  1.63it/s]
 67%|██████▋   | 8972/13317 [1:32:29<44:43,  1.62it/s]
                                                      
{'loss': 0.149, 'grad_norm': 0.4842950701713562, 'learning_rate': 2.9072220580194822e-06, 'epoch': 2.02}

 67%|██████▋   | 8972/13317 [1:32:29<44:43,  1.62it/s]
 67%|██████▋   | 8973/13317 [1:32:30<44:38,  1.62it/s]
                                                      
{'loss': 0.1708, 'grad_norm': 0.5748940110206604, 'learning_rate': 2.9060318225722794e-06, 'epoch': 2.02}

 67%|██████▋   | 8973/13317 [1:32:30<44:38,  1.62it/s]
 67%|██████▋   | 8974/13317 [1:32:30<44:34,  1.62it/s]
                                                      
{'loss': 0.1338, 'grad_norm': 0.4516615569591522, 'learning_rate': 2.90484173100286e-06, 'epoch': 2.02}

 67%|██████▋   | 8974/13317 [1:32:30<44:34,  1.62it/s]
 67%|██████▋   | 8975/13317 [1:32:31<44:30,  1.63it/s]
                                                      
{'loss': 0.1627, 'grad_norm': 0.5219327807426453, 'learning_rate': 2.903651783393e-06, 'epoch': 2.02}

 67%|██████▋   | 8975/13317 [1:32:31<44:30,  1.63it/s]
 67%|██████▋   | 8976/13317 [1:32:31<44:33,  1.62it/s]
                                                      
{'loss': 0.1562, 'grad_norm': 0.6024295091629028, 'learning_rate': 2.902461979824458e-06, 'epoch': 2.02}

 67%|██████▋   | 8976/13317 [1:32:32<44:33,  1.62it/s]
 67%|██████▋   | 8977/13317 [1:32:32<44:35,  1.62it/s]
                                                      
{'loss': 0.1292, 'grad_norm': 0.4363348186016083, 'learning_rate': 2.9012723203789863e-06, 'epoch': 2.02}

 67%|██████▋   | 8977/13317 [1:32:32<44:35,  1.62it/s]
 67%|██████▋   | 8978/13317 [1:32:33<44:31,  1.62it/s]
                                                      
{'loss': 0.1383, 'grad_norm': 0.5174410343170166, 'learning_rate': 2.900082805138329e-06, 'epoch': 2.02}

 67%|██████▋   | 8978/13317 [1:32:33<44:31,  1.62it/s]
 67%|██████▋   | 8979/13317 [1:32:33<44:27,  1.63it/s]
                                                      
{'loss': 0.1126, 'grad_norm': 0.4104403257369995, 'learning_rate': 2.8988934341842153e-06, 'epoch': 2.02}

 67%|██████▋   | 8979/13317 [1:32:33<44:27,  1.63it/s]
 67%|██████▋   | 8980/13317 [1:32:34<44:24,  1.63it/s]
                                                      
{'loss': 0.1751, 'grad_norm': 0.5883663892745972, 'learning_rate': 2.8977042075983715e-06, 'epoch': 2.02}

 67%|██████▋   | 8980/13317 [1:32:34<44:24,  1.63it/s]
 67%|██████▋   | 8981/13317 [1:32:35<44:28,  1.63it/s]
                                                      
{'loss': 0.1532, 'grad_norm': 0.5967925786972046, 'learning_rate': 2.896515125462506e-06, 'epoch': 2.02}

 67%|██████▋   | 8981/13317 [1:32:35<44:28,  1.63it/s]
 67%|██████▋   | 8982/13317 [1:32:35<44:26,  1.63it/s]
                                                      
{'loss': 0.1864, 'grad_norm': 0.5857529640197754, 'learning_rate': 2.8953261878583263e-06, 'epoch': 2.02}

 67%|██████▋   | 8982/13317 [1:32:35<44:26,  1.63it/s]
 67%|██████▋   | 8983/13317 [1:32:36<44:23,  1.63it/s]
                                                      
{'loss': 0.1589, 'grad_norm': 0.5088075399398804, 'learning_rate': 2.8941373948675213e-06, 'epoch': 2.02}

 67%|██████▋   | 8983/13317 [1:32:36<44:23,  1.63it/s]
 67%|██████▋   | 8984/13317 [1:32:36<44:21,  1.63it/s]
                                                      
{'loss': 0.1852, 'grad_norm': 0.6072152853012085, 'learning_rate': 2.8929487465717725e-06, 'epoch': 2.02}

 67%|██████▋   | 8984/13317 [1:32:36<44:21,  1.63it/s]
 67%|██████▋   | 8985/13317 [1:32:37<44:20,  1.63it/s]
                                                      
{'loss': 0.1374, 'grad_norm': 0.45862898230552673, 'learning_rate': 2.8917602430527548e-06, 'epoch': 2.02}

 67%|██████▋   | 8985/13317 [1:32:37<44:20,  1.63it/s]
 67%|██████▋   | 8986/13317 [1:32:38<44:19,  1.63it/s]
                                                      
{'loss': 0.1882, 'grad_norm': 0.7007378935813904, 'learning_rate': 2.8905718843921326e-06, 'epoch': 2.02}

 67%|██████▋   | 8986/13317 [1:32:38<44:19,  1.63it/s]
 67%|██████▋   | 8987/13317 [1:32:38<44:18,  1.63it/s]
                                                      
{'loss': 0.1683, 'grad_norm': 0.5688397288322449, 'learning_rate': 2.889383670671557e-06, 'epoch': 2.02}

 67%|██████▋   | 8987/13317 [1:32:38<44:18,  1.63it/s]
 67%|██████▋   | 8988/13317 [1:32:39<44:17,  1.63it/s]
                                                      
{'loss': 0.1152, 'grad_norm': 0.4543035328388214, 'learning_rate': 2.8881956019726697e-06, 'epoch': 2.02}

 67%|██████▋   | 8988/13317 [1:32:39<44:17,  1.63it/s]
 68%|██████▊   | 8989/13317 [1:32:39<44:16,  1.63it/s]
                                                      
{'loss': 0.1416, 'grad_norm': 0.49085313081741333, 'learning_rate': 2.887007678377106e-06, 'epoch': 2.03}

 68%|██████▊   | 8989/13317 [1:32:40<44:16,  1.63it/s]
 68%|██████▊   | 8990/13317 [1:32:40<44:15,  1.63it/s]
                                                      
{'loss': 0.1211, 'grad_norm': 0.5596198439598083, 'learning_rate': 2.8858198999664877e-06, 'epoch': 2.03}

 68%|██████▊   | 8990/13317 [1:32:40<44:15,  1.63it/s]
 68%|██████▊   | 8991/13317 [1:32:41<44:17,  1.63it/s]
                                                      
{'loss': 0.1543, 'grad_norm': 0.5358420014381409, 'learning_rate': 2.8846322668224248e-06, 'epoch': 2.03}

 68%|██████▊   | 8991/13317 [1:32:41<44:17,  1.63it/s]
 68%|██████▊   | 8992/13317 [1:32:41<44:16,  1.63it/s]
                                                      
{'loss': 0.1484, 'grad_norm': 0.5309452414512634, 'learning_rate': 2.883444779026524e-06, 'epoch': 2.03}

 68%|██████▊   | 8992/13317 [1:32:41<44:16,  1.63it/s]
 68%|██████▊   | 8993/13317 [1:32:42<44:13,  1.63it/s]
                                                      
{'loss': 0.1876, 'grad_norm': 0.5891040563583374, 'learning_rate': 2.8822574366603804e-06, 'epoch': 2.03}

 68%|██████▊   | 8993/13317 [1:32:42<44:13,  1.63it/s]
 68%|██████▊   | 8994/13317 [1:32:43<44:13,  1.63it/s]
                                                      
{'loss': 0.2548, 'grad_norm': 0.7087313532829285, 'learning_rate': 2.8810702398055707e-06, 'epoch': 2.03}

 68%|██████▊   | 8994/13317 [1:32:43<44:13,  1.63it/s]
 68%|██████▊   | 8995/13317 [1:32:43<44:11,  1.63it/s]
                                                      
{'loss': 0.1607, 'grad_norm': 0.5149721503257751, 'learning_rate': 2.879883188543671e-06, 'epoch': 2.03}

 68%|██████▊   | 8995/13317 [1:32:43<44:11,  1.63it/s]
 68%|██████▊   | 8996/13317 [1:32:44<44:09,  1.63it/s]
                                                      
{'loss': 0.1284, 'grad_norm': 0.49393561482429504, 'learning_rate': 2.878696282956246e-06, 'epoch': 2.03}

 68%|██████▊   | 8996/13317 [1:32:44<44:09,  1.63it/s]
 68%|██████▊   | 8997/13317 [1:32:44<44:09,  1.63it/s]
                                                      
{'loss': 0.1667, 'grad_norm': 0.48004284501075745, 'learning_rate': 2.8775095231248475e-06, 'epoch': 2.03}

 68%|██████▊   | 8997/13317 [1:32:44<44:09,  1.63it/s]
 68%|██████▊   | 8998/13317 [1:32:45<44:08,  1.63it/s]
                                                      
{'loss': 0.1528, 'grad_norm': 0.6348469853401184, 'learning_rate': 2.8763229091310164e-06, 'epoch': 2.03}

 68%|██████▊   | 8998/13317 [1:32:45<44:08,  1.63it/s]
 68%|██████▊   | 8999/13317 [1:32:46<44:07,  1.63it/s]
                                                      
{'loss': 0.1861, 'grad_norm': 0.6506682634353638, 'learning_rate': 2.8751364410562864e-06, 'epoch': 2.03}

 68%|██████▊   | 8999/13317 [1:32:46<44:07,  1.63it/s]
 68%|██████▊   | 9000/13317 [1:32:46<44:08,  1.63it/s]
                                                      
{'loss': 0.1609, 'grad_norm': 0.543276846408844, 'learning_rate': 2.8739501189821856e-06, 'epoch': 2.03}

 68%|██████▊   | 9000/13317 [1:32:46<44:08,  1.63it/s]

  0%|          | 0/13 [00:00<?, ?it/s]

 15%|█▌        | 2/13 [00:00<00:00, 13.37it/s]

 31%|███       | 4/13 [00:00<00:01,  8.41it/s]

 38%|███▊      | 5/13 [00:00<00:01,  7.81it/s]

 46%|████▌     | 6/13 [00:00<00:00,  7.43it/s]

 54%|█████▍    | 7/13 [00:00<00:00,  7.19it/s]

 62%|██████▏   | 8/13 [00:01<00:00,  7.02it/s]

 69%|██████▉   | 9/13 [00:01<00:00,  6.91it/s]

 77%|███████▋  | 10/13 [00:01<00:00,  6.84it/s]

 85%|████████▍ | 11/13 [00:01<00:00,  6.79it/s]

 92%|█████████▏| 12/13 [00:01<00:00,  6.75it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]
                                                      

                                               
{'eval_loss': 0.5465837717056274, 'eval_runtime': 1.9514, 'eval_samples_per_second': 102.489, 'eval_steps_per_second': 6.662, 'epoch': 2.03}

 68%|██████▊   | 9000/13317 [1:32:48<44:08,  1.63it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]

                                               
 68%|██████▊   | 9001/13317 [1:32:49<1:26:20,  1.20s/it]
                                                        
{'loss': 0.197, 'grad_norm': 0.5920767188072205, 'learning_rate': 2.872763942990219e-06, 'epoch': 2.03}

 68%|██████▊   | 9001/13317 [1:32:49<1:26:20,  1.20s/it]
 68%|██████▊   | 9002/13317 [1:32:49<1:13:39,  1.02s/it]
                                                        
{'loss': 0.1812, 'grad_norm': 0.6106680035591125, 'learning_rate': 2.8715779131618925e-06, 'epoch': 2.03}

 68%|██████▊   | 9002/13317 [1:32:49<1:13:39,  1.02s/it]
 68%|██████▊   | 9003/13317 [1:32:50<1:04:47,  1.11it/s]
                                                        
{'loss': 0.1456, 'grad_norm': 0.5116019248962402, 'learning_rate': 2.870392029578702e-06, 'epoch': 2.03}

 68%|██████▊   | 9003/13317 [1:32:50<1:04:47,  1.11it/s]
 68%|██████▊   | 9004/13317 [1:32:51<58:34,  1.23it/s]  
                                                      
{'loss': 0.1529, 'grad_norm': 0.6495400667190552, 'learning_rate': 2.8692062923221274e-06, 'epoch': 2.03}

 68%|██████▊   | 9004/13317 [1:32:51<58:34,  1.23it/s]
 68%|██████▊   | 9005/13317 [1:32:51<54:13,  1.33it/s]
                                                      
{'loss': 0.1815, 'grad_norm': 0.490772008895874, 'learning_rate': 2.8680207014736396e-06, 'epoch': 2.03}

 68%|██████▊   | 9005/13317 [1:32:51<54:13,  1.33it/s]
 68%|██████▊   | 9006/13317 [1:32:52<51:11,  1.40it/s]
                                                      
{'loss': 0.1518, 'grad_norm': 0.4911167621612549, 'learning_rate': 2.8668352571147036e-06, 'epoch': 2.03}

 68%|██████▊   | 9006/13317 [1:32:52<51:11,  1.40it/s]
 68%|██████▊   | 9007/13317 [1:32:52<49:03,  1.46it/s]
                                                      
{'loss': 0.1591, 'grad_norm': 0.5364722013473511, 'learning_rate': 2.865649959326774e-06, 'epoch': 2.03}

 68%|██████▊   | 9007/13317 [1:32:53<49:03,  1.46it/s]
 68%|██████▊   | 9008/13317 [1:32:53<47:32,  1.51it/s]
                                                      
{'loss': 0.181, 'grad_norm': 0.5133891105651855, 'learning_rate': 2.8644648081912906e-06, 'epoch': 2.03}

 68%|██████▊   | 9008/13317 [1:32:53<47:32,  1.51it/s]
 68%|██████▊   | 9009/13317 [1:32:54<46:31,  1.54it/s]
                                                      
{'loss': 0.1536, 'grad_norm': 0.5263482928276062, 'learning_rate': 2.863279803789684e-06, 'epoch': 2.03}

 68%|██████▊   | 9009/13317 [1:32:54<46:31,  1.54it/s]
 68%|██████▊   | 9010/13317 [1:32:54<45:45,  1.57it/s]
                                                      
{'loss': 0.1617, 'grad_norm': 0.5683949589729309, 'learning_rate': 2.8620949462033813e-06, 'epoch': 2.03}

 68%|██████▊   | 9010/13317 [1:32:54<45:45,  1.57it/s]
 68%|██████▊   | 9011/13317 [1:32:55<45:15,  1.59it/s]
                                                      
{'loss': 0.1379, 'grad_norm': 0.46637415885925293, 'learning_rate': 2.8609102355137924e-06, 'epoch': 2.03}

 68%|██████▊   | 9011/13317 [1:32:55<45:15,  1.59it/s]
 68%|██████▊   | 9012/13317 [1:32:56<44:52,  1.60it/s]
                                                      
{'loss': 0.1585, 'grad_norm': 0.5478501915931702, 'learning_rate': 2.8597256718023168e-06, 'epoch': 2.03}

 68%|██████▊   | 9012/13317 [1:32:56<44:52,  1.60it/s]
 68%|██████▊   | 9013/13317 [1:32:56<44:36,  1.61it/s]
                                                      
{'loss': 0.163, 'grad_norm': 0.5715697407722473, 'learning_rate': 2.85854125515035e-06, 'epoch': 2.03}

 68%|██████▊   | 9013/13317 [1:32:56<44:36,  1.61it/s]
 68%|██████▊   | 9014/13317 [1:32:57<44:26,  1.61it/s]
                                                      
{'loss': 0.1428, 'grad_norm': 0.509544849395752, 'learning_rate': 2.8573569856392745e-06, 'epoch': 2.03}

 68%|██████▊   | 9014/13317 [1:32:57<44:26,  1.61it/s]
 68%|██████▊   | 9015/13317 [1:32:57<44:18,  1.62it/s]
                                                      
{'loss': 0.14, 'grad_norm': 0.5204402208328247, 'learning_rate': 2.856172863350461e-06, 'epoch': 2.03}

 68%|██████▊   | 9015/13317 [1:32:57<44:18,  1.62it/s]
 68%|██████▊   | 9016/13317 [1:32:58<44:13,  1.62it/s]
                                                      
{'loss': 0.152, 'grad_norm': 0.5372000932693481, 'learning_rate': 2.8549888883652687e-06, 'epoch': 2.03}

 68%|██████▊   | 9016/13317 [1:32:58<44:13,  1.62it/s]
 68%|██████▊   | 9017/13317 [1:32:59<44:08,  1.62it/s]
                                                      
{'loss': 0.1391, 'grad_norm': 0.534463107585907, 'learning_rate': 2.8538050607650535e-06, 'epoch': 2.03}

 68%|██████▊   | 9017/13317 [1:32:59<44:08,  1.62it/s]
 68%|██████▊   | 9018/13317 [1:32:59<44:06,  1.62it/s]
                                                      
{'loss': 0.145, 'grad_norm': 0.5226626992225647, 'learning_rate': 2.8526213806311552e-06, 'epoch': 2.03}

 68%|██████▊   | 9018/13317 [1:32:59<44:06,  1.62it/s]
 68%|██████▊   | 9019/13317 [1:33:00<44:02,  1.63it/s]
                                                      
{'loss': 0.1334, 'grad_norm': 0.537613570690155, 'learning_rate': 2.8514378480449033e-06, 'epoch': 2.03}

 68%|██████▊   | 9019/13317 [1:33:00<44:02,  1.63it/s]
 68%|██████▊   | 9020/13317 [1:33:00<44:01,  1.63it/s]
                                                      
{'loss': 0.1528, 'grad_norm': 0.5923119187355042, 'learning_rate': 2.8502544630876215e-06, 'epoch': 2.03}

 68%|██████▊   | 9020/13317 [1:33:01<44:01,  1.63it/s]
 68%|██████▊   | 9021/13317 [1:33:01<43:59,  1.63it/s]
                                                      
{'loss': 0.1471, 'grad_norm': 0.4598279893398285, 'learning_rate': 2.849071225840619e-06, 'epoch': 2.03}

 68%|██████▊   | 9021/13317 [1:33:01<43:59,  1.63it/s]
 68%|██████▊   | 9022/13317 [1:33:02<43:57,  1.63it/s]
                                                      
{'loss': 0.1469, 'grad_norm': 0.4931190311908722, 'learning_rate': 2.847888136385199e-06, 'epoch': 2.03}

 68%|██████▊   | 9022/13317 [1:33:02<43:57,  1.63it/s]
 68%|██████▊   | 9023/13317 [1:33:02<43:56,  1.63it/s]
                                                      
{'loss': 0.1387, 'grad_norm': 0.46956154704093933, 'learning_rate': 2.846705194802649e-06, 'epoch': 2.03}

 68%|██████▊   | 9023/13317 [1:33:02<43:56,  1.63it/s]
 68%|██████▊   | 9024/13317 [1:33:03<43:56,  1.63it/s]
                                                      
{'loss': 0.1734, 'grad_norm': 0.6062358021736145, 'learning_rate': 2.8455224011742537e-06, 'epoch': 2.03}

 68%|██████▊   | 9024/13317 [1:33:03<43:56,  1.63it/s]
 68%|██████▊   | 9025/13317 [1:33:03<43:53,  1.63it/s]
                                                      
{'loss': 0.136, 'grad_norm': 0.5443840622901917, 'learning_rate': 2.8443397555812814e-06, 'epoch': 2.03}

 68%|██████▊   | 9025/13317 [1:33:04<43:53,  1.63it/s]
 68%|██████▊   | 9026/13317 [1:33:04<43:53,  1.63it/s]
                                                      
{'loss': 0.1501, 'grad_norm': 0.4948100447654724, 'learning_rate': 2.8431572581049905e-06, 'epoch': 2.03}

 68%|██████▊   | 9026/13317 [1:33:04<43:53,  1.63it/s]
 68%|██████▊   | 9027/13317 [1:33:05<43:52,  1.63it/s]
                                                      
{'loss': 0.1615, 'grad_norm': 0.6006219983100891, 'learning_rate': 2.8419749088266352e-06, 'epoch': 2.03}

 68%|██████▊   | 9027/13317 [1:33:05<43:52,  1.63it/s]
 68%|██████▊   | 9028/13317 [1:33:05<43:52,  1.63it/s]
                                                      
{'loss': 0.1807, 'grad_norm': 0.6456959843635559, 'learning_rate': 2.840792707827451e-06, 'epoch': 2.03}

 68%|██████▊   | 9028/13317 [1:33:05<43:52,  1.63it/s]
 68%|██████▊   | 9029/13317 [1:33:06<43:52,  1.63it/s]
                                                      
{'loss': 0.133, 'grad_norm': 0.48599958419799805, 'learning_rate': 2.8396106551886716e-06, 'epoch': 2.03}

 68%|██████▊   | 9029/13317 [1:33:06<43:52,  1.63it/s]
 68%|██████▊   | 9030/13317 [1:33:07<43:52,  1.63it/s]
                                                      
{'loss': 0.1676, 'grad_norm': 0.5337496995925903, 'learning_rate': 2.838428750991513e-06, 'epoch': 2.03}

 68%|██████▊   | 9030/13317 [1:33:07<43:52,  1.63it/s]
 68%|██████▊   | 9031/13317 [1:33:07<43:53,  1.63it/s]
                                                      
{'loss': 0.1285, 'grad_norm': 0.4658252000808716, 'learning_rate': 2.8372469953171876e-06, 'epoch': 2.03}

 68%|██████▊   | 9031/13317 [1:33:07<43:53,  1.63it/s]
 68%|██████▊   | 9032/13317 [1:33:08<43:52,  1.63it/s]
                                                      
{'loss': 0.188, 'grad_norm': 0.6044343709945679, 'learning_rate': 2.836065388246894e-06, 'epoch': 2.03}

 68%|██████▊   | 9032/13317 [1:33:08<43:52,  1.63it/s]
 68%|██████▊   | 9033/13317 [1:33:08<43:53,  1.63it/s]
                                                      
{'loss': 0.1973, 'grad_norm': 0.483792781829834, 'learning_rate': 2.834883929861818e-06, 'epoch': 2.03}

 68%|██████▊   | 9033/13317 [1:33:09<43:53,  1.63it/s]
 68%|██████▊   | 9034/13317 [1:33:09<43:52,  1.63it/s]
                                                      
{'loss': 0.1441, 'grad_norm': 0.5563597679138184, 'learning_rate': 2.8337026202431428e-06, 'epoch': 2.04}

 68%|██████▊   | 9034/13317 [1:33:09<43:52,  1.63it/s]
 68%|██████▊   | 9035/13317 [1:33:10<43:51,  1.63it/s]
                                                      
{'loss': 0.1356, 'grad_norm': 0.5040873885154724, 'learning_rate': 2.832521459472033e-06, 'epoch': 2.04}

 68%|██████▊   | 9035/13317 [1:33:10<43:51,  1.63it/s]
 68%|██████▊   | 9036/13317 [1:33:10<43:49,  1.63it/s]
                                                      
{'loss': 0.16, 'grad_norm': 0.49584493041038513, 'learning_rate': 2.83134044762965e-06, 'epoch': 2.04}

 68%|██████▊   | 9036/13317 [1:33:10<43:49,  1.63it/s]
 68%|██████▊   | 9037/13317 [1:33:11<43:48,  1.63it/s]
                                                      
{'loss': 0.1115, 'grad_norm': 0.40354979038238525, 'learning_rate': 2.830159584797141e-06, 'epoch': 2.04}

 68%|██████▊   | 9037/13317 [1:33:11<43:48,  1.63it/s]
 68%|██████▊   | 9038/13317 [1:33:11<43:48,  1.63it/s]
                                                      
{'loss': 0.1377, 'grad_norm': 0.5104973316192627, 'learning_rate': 2.828978871055641e-06, 'epoch': 2.04}

 68%|██████▊   | 9038/13317 [1:33:12<43:48,  1.63it/s]
 68%|██████▊   | 9039/13317 [1:33:12<43:49,  1.63it/s]
                                                      
{'loss': 0.1669, 'grad_norm': 0.6075260043144226, 'learning_rate': 2.827798306486279e-06, 'epoch': 2.04}

 68%|██████▊   | 9039/13317 [1:33:12<43:49,  1.63it/s]
 68%|██████▊   | 9040/13317 [1:33:13<43:46,  1.63it/s]
                                                      
{'loss': 0.1317, 'grad_norm': 0.5263929963111877, 'learning_rate': 2.8266178911701757e-06, 'epoch': 2.04}

 68%|██████▊   | 9040/13317 [1:33:13<43:46,  1.63it/s]
 68%|██████▊   | 9041/13317 [1:33:13<43:48,  1.63it/s]
                                                      
{'loss': 0.173, 'grad_norm': 0.5507851243019104, 'learning_rate': 2.8254376251884353e-06, 'epoch': 2.04}

 68%|██████▊   | 9041/13317 [1:33:13<43:48,  1.63it/s]
 68%|██████▊   | 9042/13317 [1:33:14<43:46,  1.63it/s]
                                                      
{'loss': 0.1311, 'grad_norm': 0.4647568464279175, 'learning_rate': 2.8242575086221523e-06, 'epoch': 2.04}

 68%|██████▊   | 9042/13317 [1:33:14<43:46,  1.63it/s]
 68%|██████▊   | 9043/13317 [1:33:15<43:44,  1.63it/s]
                                                      
{'loss': 0.1478, 'grad_norm': 0.5597816705703735, 'learning_rate': 2.8230775415524182e-06, 'epoch': 2.04}

 68%|██████▊   | 9043/13317 [1:33:15<43:44,  1.63it/s]
 68%|██████▊   | 9044/13317 [1:33:15<43:41,  1.63it/s]
                                                      
{'loss': 0.1625, 'grad_norm': 0.5454680919647217, 'learning_rate': 2.8218977240603065e-06, 'epoch': 2.04}

 68%|██████▊   | 9044/13317 [1:33:15<43:41,  1.63it/s]
 68%|██████▊   | 9045/13317 [1:33:16<43:40,  1.63it/s]
                                                      
{'loss': 0.1456, 'grad_norm': 0.48294714093208313, 'learning_rate': 2.8207180562268805e-06, 'epoch': 2.04}

 68%|██████▊   | 9045/13317 [1:33:16<43:40,  1.63it/s]
 68%|██████▊   | 9046/13317 [1:33:16<43:41,  1.63it/s]
                                                      
{'loss': 0.1564, 'grad_norm': 0.5939539670944214, 'learning_rate': 2.819538538133199e-06, 'epoch': 2.04}

 68%|██████▊   | 9046/13317 [1:33:17<43:41,  1.63it/s]
 68%|██████▊   | 9047/13317 [1:33:17<43:41,  1.63it/s]
                                                      
{'loss': 0.1552, 'grad_norm': 0.5996304750442505, 'learning_rate': 2.8183591698603098e-06, 'epoch': 2.04}

 68%|██████▊   | 9047/13317 [1:33:17<43:41,  1.63it/s]
 68%|██████▊   | 9048/13317 [1:33:18<43:39,  1.63it/s]
                                                      
{'loss': 0.1472, 'grad_norm': 0.6129876971244812, 'learning_rate': 2.817179951489244e-06, 'epoch': 2.04}

 68%|██████▊   | 9048/13317 [1:33:18<43:39,  1.63it/s]
 68%|██████▊   | 9049/13317 [1:33:18<43:39,  1.63it/s]
                                                      
{'loss': 0.1301, 'grad_norm': 0.46207472681999207, 'learning_rate': 2.8160008831010266e-06, 'epoch': 2.04}

 68%|██████▊   | 9049/13317 [1:33:18<43:39,  1.63it/s]
 68%|██████▊   | 9050/13317 [1:33:19<43:37,  1.63it/s]
                                                      
{'loss': 0.1151, 'grad_norm': 0.4404259920120239, 'learning_rate': 2.814821964776675e-06, 'epoch': 2.04}

 68%|██████▊   | 9050/13317 [1:33:19<43:37,  1.63it/s]
 68%|██████▊   | 9051/13317 [1:33:19<43:37,  1.63it/s]
                                                      
{'loss': 0.1318, 'grad_norm': 0.6945632100105286, 'learning_rate': 2.8136431965971913e-06, 'epoch': 2.04}

 68%|██████▊   | 9051/13317 [1:33:20<43:37,  1.63it/s]
 68%|██████▊   | 9052/13317 [1:33:20<43:36,  1.63it/s]
                                                      
{'loss': 0.144, 'grad_norm': 0.5617467164993286, 'learning_rate': 2.8124645786435665e-06, 'epoch': 2.04}

 68%|██████▊   | 9052/13317 [1:33:20<43:36,  1.63it/s]
 68%|██████▊   | 9053/13317 [1:33:21<43:35,  1.63it/s]
                                                      
{'loss': 0.1674, 'grad_norm': 0.6669450402259827, 'learning_rate': 2.8112861109967883e-06, 'epoch': 2.04}

 68%|██████▊   | 9053/13317 [1:33:21<43:35,  1.63it/s]
 68%|██████▊   | 9054/13317 [1:33:21<43:34,  1.63it/s]
                                                      
{'loss': 0.1025, 'grad_norm': 0.4255969524383545, 'learning_rate': 2.810107793737833e-06, 'epoch': 2.04}

 68%|██████▊   | 9054/13317 [1:33:21<43:34,  1.63it/s]
 68%|██████▊   | 9055/13317 [1:33:22<43:34,  1.63it/s]
                                                      
{'loss': 0.1417, 'grad_norm': 0.48760804533958435, 'learning_rate': 2.808929626947654e-06, 'epoch': 2.04}

 68%|██████▊   | 9055/13317 [1:33:22<43:34,  1.63it/s]
 68%|██████▊   | 9056/13317 [1:33:23<43:34,  1.63it/s]
                                                      
{'loss': 0.1812, 'grad_norm': 0.5833709239959717, 'learning_rate': 2.80775161070721e-06, 'epoch': 2.04}

 68%|██████▊   | 9056/13317 [1:33:23<43:34,  1.63it/s]
 68%|██████▊   | 9057/13317 [1:33:23<43:34,  1.63it/s]
                                                      
{'loss': 0.1553, 'grad_norm': 0.5729191899299622, 'learning_rate': 2.806573745097444e-06, 'epoch': 2.04}

 68%|██████▊   | 9057/13317 [1:33:23<43:34,  1.63it/s]
 68%|██████▊   | 9058/13317 [1:33:24<43:32,  1.63it/s]
                                                      
{'loss': 0.1457, 'grad_norm': 0.5460982918739319, 'learning_rate': 2.805396030199286e-06, 'epoch': 2.04}

 68%|██████▊   | 9058/13317 [1:33:24<43:32,  1.63it/s]
 68%|██████▊   | 9059/13317 [1:33:24<43:32,  1.63it/s]
                                                      
{'loss': 0.1767, 'grad_norm': 0.5949220657348633, 'learning_rate': 2.8042184660936556e-06, 'epoch': 2.04}

 68%|██████▊   | 9059/13317 [1:33:24<43:32,  1.63it/s]
 68%|██████▊   | 9060/13317 [1:33:25<43:33,  1.63it/s]
                                                      
{'loss': 0.1294, 'grad_norm': 0.49007827043533325, 'learning_rate': 2.803041052861466e-06, 'epoch': 2.04}

 68%|██████▊   | 9060/13317 [1:33:25<43:33,  1.63it/s]
 68%|██████▊   | 9061/13317 [1:33:26<43:32,  1.63it/s]
                                                      
{'loss': 0.1219, 'grad_norm': 0.44217821955680847, 'learning_rate': 2.80186379058362e-06, 'epoch': 2.04}

 68%|██████▊   | 9061/13317 [1:33:26<43:32,  1.63it/s]
 68%|██████▊   | 9062/13317 [1:33:26<43:32,  1.63it/s]
                                                      
{'loss': 0.1819, 'grad_norm': 0.6263560652732849, 'learning_rate': 2.8006866793410055e-06, 'epoch': 2.04}

 68%|██████▊   | 9062/13317 [1:33:26<43:32,  1.63it/s]
 68%|██████▊   | 9063/13317 [1:33:27<43:30,  1.63it/s]
                                                      
{'loss': 0.1108, 'grad_norm': 0.4157686233520508, 'learning_rate': 2.799509719214501e-06, 'epoch': 2.04}

 68%|██████▊   | 9063/13317 [1:33:27<43:30,  1.63it/s]
 68%|██████▊   | 9064/13317 [1:33:27<43:28,  1.63it/s]
                                                      
{'loss': 0.1789, 'grad_norm': 0.5682604312896729, 'learning_rate': 2.7983329102849797e-06, 'epoch': 2.04}

 68%|██████▊   | 9064/13317 [1:33:28<43:28,  1.63it/s]
 68%|██████▊   | 9065/13317 [1:33:28<43:27,  1.63it/s]
                                                      
{'loss': 0.1292, 'grad_norm': 0.4913271367549896, 'learning_rate': 2.7971562526332996e-06, 'epoch': 2.04}

 68%|██████▊   | 9065/13317 [1:33:28<43:27,  1.63it/s]
 68%|██████▊   | 9066/13317 [1:33:29<43:28,  1.63it/s]
                                                      
{'loss': 0.1404, 'grad_norm': 0.6475685238838196, 'learning_rate': 2.7959797463403065e-06, 'epoch': 2.04}

 68%|██████▊   | 9066/13317 [1:33:29<43:28,  1.63it/s]
 68%|██████▊   | 9067/13317 [1:33:29<43:26,  1.63it/s]
                                                      
{'loss': 0.1572, 'grad_norm': 0.5674029588699341, 'learning_rate': 2.7948033914868415e-06, 'epoch': 2.04}

 68%|██████▊   | 9067/13317 [1:33:29<43:26,  1.63it/s]
 68%|██████▊   | 9068/13317 [1:33:30<43:25,  1.63it/s]
                                                      
{'loss': 0.1298, 'grad_norm': 0.5201994776725769, 'learning_rate': 2.7936271881537345e-06, 'epoch': 2.04}

 68%|██████▊   | 9068/13317 [1:33:30<43:25,  1.63it/s]
 68%|██████▊   | 9069/13317 [1:33:30<43:23,  1.63it/s]
                                                      
{'loss': 0.1734, 'grad_norm': 0.6683054566383362, 'learning_rate': 2.792451136421801e-06, 'epoch': 2.04}

 68%|██████▊   | 9069/13317 [1:33:31<43:23,  1.63it/s]
 68%|██████▊   | 9070/13317 [1:33:31<43:25,  1.63it/s]
                                                      
{'loss': 0.1615, 'grad_norm': 0.5334722399711609, 'learning_rate': 2.7912752363718464e-06, 'epoch': 2.04}

 68%|██████▊   | 9070/13317 [1:33:31<43:25,  1.63it/s]
 68%|██████▊   | 9071/13317 [1:33:32<43:26,  1.63it/s]
                                                      
{'loss': 0.1732, 'grad_norm': 0.580101728439331, 'learning_rate': 2.790099488084672e-06, 'epoch': 2.04}

 68%|██████▊   | 9071/13317 [1:33:32<43:26,  1.63it/s]
 68%|██████▊   | 9072/13317 [1:33:32<43:24,  1.63it/s]
                                                      
{'loss': 0.168, 'grad_norm': 0.635360062122345, 'learning_rate': 2.788923891641061e-06, 'epoch': 2.04}

 68%|██████▊   | 9072/13317 [1:33:32<43:24,  1.63it/s]
 68%|██████▊   | 9073/13317 [1:33:33<43:23,  1.63it/s]
                                                      
{'loss': 0.1244, 'grad_norm': 0.47988736629486084, 'learning_rate': 2.787748447121788e-06, 'epoch': 2.04}

 68%|██████▊   | 9073/13317 [1:33:33<43:23,  1.63it/s]
 68%|██████▊   | 9074/13317 [1:33:34<43:22,  1.63it/s]
                                                      
{'loss': 0.149, 'grad_norm': 0.5711203813552856, 'learning_rate': 2.7865731546076203e-06, 'epoch': 2.04}

 68%|██████▊   | 9074/13317 [1:33:34<43:22,  1.63it/s]
 68%|██████▊   | 9075/13317 [1:33:34<43:21,  1.63it/s]
                                                      
{'loss': 0.1429, 'grad_norm': 0.570212185382843, 'learning_rate': 2.785398014179315e-06, 'epoch': 2.04}

 68%|██████▊   | 9075/13317 [1:33:34<43:21,  1.63it/s]
 68%|██████▊   | 9076/13317 [1:33:35<43:20,  1.63it/s]
                                                      
{'loss': 0.1499, 'grad_norm': 0.5567415356636047, 'learning_rate': 2.784223025917615e-06, 'epoch': 2.04}

 68%|██████▊   | 9076/13317 [1:33:35<43:20,  1.63it/s]
 68%|██████▊   | 9077/13317 [1:33:35<43:22,  1.63it/s]
                                                      
{'loss': 0.1645, 'grad_norm': 0.5709004998207092, 'learning_rate': 2.7830481899032523e-06, 'epoch': 2.04}

 68%|██████▊   | 9077/13317 [1:33:36<43:22,  1.63it/s]
 68%|██████▊   | 9078/13317 [1:33:36<43:21,  1.63it/s]
                                                      
{'loss': 0.1598, 'grad_norm': 0.581677258014679, 'learning_rate': 2.7818735062169538e-06, 'epoch': 2.05}

 68%|██████▊   | 9078/13317 [1:33:36<43:21,  1.63it/s]
 68%|██████▊   | 9079/13317 [1:33:37<43:22,  1.63it/s]
                                                      
{'loss': 0.1605, 'grad_norm': 0.6076658964157104, 'learning_rate': 2.7806989749394313e-06, 'epoch': 2.05}

 68%|██████▊   | 9079/13317 [1:33:37<43:22,  1.63it/s]
 68%|██████▊   | 9080/13317 [1:33:37<43:21,  1.63it/s]
                                                      
{'loss': 0.1441, 'grad_norm': 0.5648558139801025, 'learning_rate': 2.779524596151386e-06, 'epoch': 2.05}

 68%|██████▊   | 9080/13317 [1:33:37<43:21,  1.63it/s]
 68%|██████▊   | 9081/13317 [1:33:38<43:22,  1.63it/s]
                                                      
{'loss': 0.1784, 'grad_norm': 0.565828800201416, 'learning_rate': 2.7783503699335135e-06, 'epoch': 2.05}

 68%|██████▊   | 9081/13317 [1:33:38<43:22,  1.63it/s]
 68%|██████▊   | 9082/13317 [1:33:38<43:20,  1.63it/s]
                                                      
{'loss': 0.1779, 'grad_norm': 0.7543521523475647, 'learning_rate': 2.7771762963664915e-06, 'epoch': 2.05}

 68%|██████▊   | 9082/13317 [1:33:39<43:20,  1.63it/s]
 68%|██████▊   | 9083/13317 [1:33:39<43:18,  1.63it/s]
                                                      
{'loss': 0.1139, 'grad_norm': 0.5027368068695068, 'learning_rate': 2.776002375530996e-06, 'epoch': 2.05}

 68%|██████▊   | 9083/13317 [1:33:39<43:18,  1.63it/s]
 68%|██████▊   | 9084/13317 [1:33:40<43:18,  1.63it/s]
                                                      
{'loss': 0.192, 'grad_norm': 0.6767992973327637, 'learning_rate': 2.7748286075076834e-06, 'epoch': 2.05}

 68%|██████▊   | 9084/13317 [1:33:40<43:18,  1.63it/s]
 68%|██████▊   | 9085/13317 [1:33:40<43:17,  1.63it/s]
                                                      
{'loss': 0.155, 'grad_norm': 0.5553813576698303, 'learning_rate': 2.7736549923772083e-06, 'epoch': 2.05}

 68%|██████▊   | 9085/13317 [1:33:40<43:17,  1.63it/s]
 68%|██████▊   | 9086/13317 [1:33:41<43:17,  1.63it/s]
                                                      
{'loss': 0.1119, 'grad_norm': 0.4560610353946686, 'learning_rate': 2.7724815302202055e-06, 'epoch': 2.05}

 68%|██████▊   | 9086/13317 [1:33:41<43:17,  1.63it/s]
 68%|██████▊   | 9087/13317 [1:33:42<43:17,  1.63it/s]
                                                      
{'loss': 0.1862, 'grad_norm': 0.6680421829223633, 'learning_rate': 2.771308221117309e-06, 'epoch': 2.05}

 68%|██████▊   | 9087/13317 [1:33:42<43:17,  1.63it/s]
 68%|██████▊   | 9088/13317 [1:33:42<43:23,  1.62it/s]
                                                      
{'loss': 0.1334, 'grad_norm': 0.5291731357574463, 'learning_rate': 2.7701350651491354e-06, 'epoch': 2.05}

 68%|██████▊   | 9088/13317 [1:33:42<43:23,  1.62it/s]
 68%|██████▊   | 9089/13317 [1:33:43<43:19,  1.63it/s]
                                                      
{'loss': 0.1422, 'grad_norm': 0.5161295533180237, 'learning_rate': 2.7689620623962915e-06, 'epoch': 2.05}

 68%|██████▊   | 9089/13317 [1:33:43<43:19,  1.63it/s]
 68%|██████▊   | 9090/13317 [1:33:43<43:18,  1.63it/s]
                                                      
{'loss': 0.1394, 'grad_norm': 0.5156646370887756, 'learning_rate': 2.767789212939378e-06, 'epoch': 2.05}

 68%|██████▊   | 9090/13317 [1:33:44<43:18,  1.63it/s]
 68%|██████▊   | 9091/13317 [1:33:44<43:18,  1.63it/s]
                                                      
{'loss': 0.1401, 'grad_norm': 0.5176277160644531, 'learning_rate': 2.7666165168589788e-06, 'epoch': 2.05}

 68%|██████▊   | 9091/13317 [1:33:44<43:18,  1.63it/s]
 68%|██████▊   | 9092/13317 [1:33:45<43:15,  1.63it/s]
                                                      
{'loss': 0.1583, 'grad_norm': 0.6724377274513245, 'learning_rate': 2.765443974235674e-06, 'epoch': 2.05}

 68%|██████▊   | 9092/13317 [1:33:45<43:15,  1.63it/s]
 68%|██████▊   | 9093/13317 [1:33:45<43:15,  1.63it/s]
                                                      
{'loss': 0.1229, 'grad_norm': 0.4674677550792694, 'learning_rate': 2.764271585150027e-06, 'epoch': 2.05}

 68%|██████▊   | 9093/13317 [1:33:45<43:15,  1.63it/s]
 68%|██████▊   | 9094/13317 [1:33:46<43:14,  1.63it/s]
                                                      
{'loss': 0.1254, 'grad_norm': 0.4917107820510864, 'learning_rate': 2.7630993496825954e-06, 'epoch': 2.05}

 68%|██████▊   | 9094/13317 [1:33:46<43:14,  1.63it/s]
 68%|██████▊   | 9095/13317 [1:33:46<43:14,  1.63it/s]
                                                      
{'loss': 0.1446, 'grad_norm': 0.5460302829742432, 'learning_rate': 2.761927267913924e-06, 'epoch': 2.05}

 68%|██████▊   | 9095/13317 [1:33:47<43:14,  1.63it/s]
 68%|██████▊   | 9096/13317 [1:33:47<43:12,  1.63it/s]
                                                      
{'loss': 0.141, 'grad_norm': 0.47360557317733765, 'learning_rate': 2.7607553399245434e-06, 'epoch': 2.05}

 68%|██████▊   | 9096/13317 [1:33:47<43:12,  1.63it/s]
 68%|██████▊   | 9097/13317 [1:33:48<43:13,  1.63it/s]
                                                      
{'loss': 0.1533, 'grad_norm': 0.582574188709259, 'learning_rate': 2.7595835657949834e-06, 'epoch': 2.05}

 68%|██████▊   | 9097/13317 [1:33:48<43:13,  1.63it/s]
 68%|██████▊   | 9098/13317 [1:33:48<43:10,  1.63it/s]
                                                      
{'loss': 0.1364, 'grad_norm': 0.5999854207038879, 'learning_rate': 2.7584119456057532e-06, 'epoch': 2.05}

 68%|██████▊   | 9098/13317 [1:33:48<43:10,  1.63it/s]
 68%|██████▊   | 9099/13317 [1:33:49<43:08,  1.63it/s]
                                                      
{'loss': 0.1704, 'grad_norm': 0.5830715894699097, 'learning_rate': 2.7572404794373558e-06, 'epoch': 2.05}

 68%|██████▊   | 9099/13317 [1:33:49<43:08,  1.63it/s]
 68%|██████▊   | 9100/13317 [1:33:50<43:07,  1.63it/s]
                                                      
{'loss': 0.1357, 'grad_norm': 0.5473541617393494, 'learning_rate': 2.7560691673702837e-06, 'epoch': 2.05}

 68%|██████▊   | 9100/13317 [1:33:50<43:07,  1.63it/s]
 68%|██████▊   | 9101/13317 [1:33:50<43:06,  1.63it/s]
                                                      
{'loss': 0.1326, 'grad_norm': 0.4636263847351074, 'learning_rate': 2.7548980094850208e-06, 'epoch': 2.05}

 68%|██████▊   | 9101/13317 [1:33:50<43:06,  1.63it/s]
 68%|██████▊   | 9102/13317 [1:33:51<43:06,  1.63it/s]
                                                      
{'loss': 0.1499, 'grad_norm': 0.5916093587875366, 'learning_rate': 2.7537270058620365e-06, 'epoch': 2.05}

 68%|██████▊   | 9102/13317 [1:33:51<43:06,  1.63it/s]
 68%|██████▊   | 9103/13317 [1:33:51<43:07,  1.63it/s]
                                                      
{'loss': 0.1643, 'grad_norm': 0.6221763491630554, 'learning_rate': 2.7525561565817883e-06, 'epoch': 2.05}

 68%|██████▊   | 9103/13317 [1:33:52<43:07,  1.63it/s]
 68%|██████▊   | 9104/13317 [1:33:52<43:06,  1.63it/s]
                                                      
{'loss': 0.2016, 'grad_norm': 0.6313110589981079, 'learning_rate': 2.751385461724731e-06, 'epoch': 2.05}

 68%|██████▊   | 9104/13317 [1:33:52<43:06,  1.63it/s]
 68%|██████▊   | 9105/13317 [1:33:53<43:05,  1.63it/s]
                                                      
{'loss': 0.1557, 'grad_norm': 0.6928877234458923, 'learning_rate': 2.7502149213713013e-06, 'epoch': 2.05}

 68%|██████▊   | 9105/13317 [1:33:53<43:05,  1.63it/s]
 68%|██████▊   | 9106/13317 [1:33:53<43:05,  1.63it/s]
                                                      
{'loss': 0.1624, 'grad_norm': 0.5508126616477966, 'learning_rate': 2.7490445356019246e-06, 'epoch': 2.05}

 68%|██████▊   | 9106/13317 [1:33:53<43:05,  1.63it/s]
 68%|██████▊   | 9107/13317 [1:33:54<43:03,  1.63it/s]
                                                      
{'loss': 0.1612, 'grad_norm': 0.6079183220863342, 'learning_rate': 2.7478743044970228e-06, 'epoch': 2.05}

 68%|██████▊   | 9107/13317 [1:33:54<43:03,  1.63it/s]
 68%|██████▊   | 9108/13317 [1:33:54<43:03,  1.63it/s]
                                                      
{'loss': 0.1504, 'grad_norm': 0.5971160531044006, 'learning_rate': 2.7467042281370038e-06, 'epoch': 2.05}

 68%|██████▊   | 9108/13317 [1:33:55<43:03,  1.63it/s]
 68%|██████▊   | 9109/13317 [1:33:55<43:02,  1.63it/s]
                                                      
{'loss': 0.2123, 'grad_norm': 0.6055189967155457, 'learning_rate': 2.745534306602263e-06, 'epoch': 2.05}

 68%|██████▊   | 9109/13317 [1:33:55<43:02,  1.63it/s]
 68%|██████▊   | 9110/13317 [1:33:56<43:01,  1.63it/s]
                                                      
{'loss': 0.1585, 'grad_norm': 0.5313714146614075, 'learning_rate': 2.7443645399731834e-06, 'epoch': 2.05}

 68%|██████▊   | 9110/13317 [1:33:56<43:01,  1.63it/s]
 68%|██████▊   | 9111/13317 [1:33:56<43:01,  1.63it/s]
                                                      
{'loss': 0.1141, 'grad_norm': 0.41166815161705017, 'learning_rate': 2.743194928330145e-06, 'epoch': 2.05}

 68%|██████▊   | 9111/13317 [1:33:56<43:01,  1.63it/s]
 68%|██████▊   | 9112/13317 [1:33:57<43:00,  1.63it/s]
                                                      
{'loss': 0.136, 'grad_norm': 0.5339260101318359, 'learning_rate': 2.742025471753511e-06, 'epoch': 2.05}

 68%|██████▊   | 9112/13317 [1:33:57<43:00,  1.63it/s]
 68%|██████▊   | 9113/13317 [1:33:58<42:59,  1.63it/s]
                                                      
{'loss': 0.1786, 'grad_norm': 0.6812987923622131, 'learning_rate': 2.7408561703236326e-06, 'epoch': 2.05}

 68%|██████▊   | 9113/13317 [1:33:58<42:59,  1.63it/s]
 68%|██████▊   | 9114/13317 [1:33:58<42:59,  1.63it/s]
                                                      
{'loss': 0.1922, 'grad_norm': 0.6494225263595581, 'learning_rate': 2.7396870241208552e-06, 'epoch': 2.05}

 68%|██████▊   | 9114/13317 [1:33:58<42:59,  1.63it/s]
 68%|██████▊   | 9115/13317 [1:33:59<42:57,  1.63it/s]
                                                      
{'loss': 0.1416, 'grad_norm': 0.5891035199165344, 'learning_rate': 2.7385180332255147e-06, 'epoch': 2.05}

 68%|██████▊   | 9115/13317 [1:33:59<42:57,  1.63it/s]
 68%|██████▊   | 9116/13317 [1:33:59<42:57,  1.63it/s]
                                                      
{'loss': 0.1929, 'grad_norm': 0.6075280904769897, 'learning_rate': 2.737349197717929e-06, 'epoch': 2.05}

 68%|██████▊   | 9116/13317 [1:33:59<42:57,  1.63it/s]
 68%|██████▊   | 9117/13317 [1:34:00<42:57,  1.63it/s]
                                                      
{'loss': 0.1283, 'grad_norm': 0.5045675039291382, 'learning_rate': 2.7361805176784095e-06, 'epoch': 2.05}

 68%|██████▊   | 9117/13317 [1:34:00<42:57,  1.63it/s]
 68%|██████▊   | 9118/13317 [1:34:01<42:59,  1.63it/s]
                                                      
{'loss': 0.1321, 'grad_norm': 0.5365733504295349, 'learning_rate': 2.735011993187258e-06, 'epoch': 2.05}

 68%|██████▊   | 9118/13317 [1:34:01<42:59,  1.63it/s]
 68%|██████▊   | 9119/13317 [1:34:01<43:02,  1.63it/s]
                                                      
{'loss': 0.1522, 'grad_norm': 0.5027294754981995, 'learning_rate': 2.733843624324769e-06, 'epoch': 2.05}

 68%|██████▊   | 9119/13317 [1:34:01<43:02,  1.63it/s]
 68%|██████▊   | 9120/13317 [1:34:02<43:01,  1.63it/s]
                                                      
{'loss': 0.1621, 'grad_norm': 0.6397231221199036, 'learning_rate': 2.732675411171213e-06, 'epoch': 2.05}

 68%|██████▊   | 9120/13317 [1:34:02<43:01,  1.63it/s]
 68%|██████▊   | 9121/13317 [1:34:02<43:02,  1.62it/s]
                                                      
{'loss': 0.1313, 'grad_norm': 0.47090062499046326, 'learning_rate': 2.7315073538068636e-06, 'epoch': 2.05}

 68%|██████▊   | 9121/13317 [1:34:03<43:02,  1.62it/s]
 68%|██████▊   | 9122/13317 [1:34:03<43:04,  1.62it/s]
                                                      
{'loss': 0.124, 'grad_norm': 0.6265190839767456, 'learning_rate': 2.73033945231198e-06, 'epoch': 2.05}

 68%|██████▊   | 9122/13317 [1:34:03<43:04,  1.62it/s]
 69%|██████▊   | 9123/13317 [1:34:04<43:01,  1.62it/s]
                                                      
{'loss': 0.1637, 'grad_norm': 0.5894726514816284, 'learning_rate': 2.7291717067668068e-06, 'epoch': 2.06}

 69%|██████▊   | 9123/13317 [1:34:04<43:01,  1.62it/s]
 69%|██████▊   | 9124/13317 [1:34:04<43:03,  1.62it/s]
                                                      
{'loss': 0.1516, 'grad_norm': 0.5683713555335999, 'learning_rate': 2.72800411725158e-06, 'epoch': 2.06}

 69%|██████▊   | 9124/13317 [1:34:04<43:03,  1.62it/s]
 69%|██████▊   | 9125/13317 [1:34:05<42:59,  1.63it/s]
                                                      
{'loss': 0.1624, 'grad_norm': 0.6376332640647888, 'learning_rate': 2.7268366838465286e-06, 'epoch': 2.06}

 69%|██████▊   | 9125/13317 [1:34:05<42:59,  1.63it/s]
 69%|██████▊   | 9126/13317 [1:34:06<42:59,  1.62it/s]
                                                      
{'loss': 0.1285, 'grad_norm': 0.4910074770450592, 'learning_rate': 2.7256694066318656e-06, 'epoch': 2.06}

 69%|██████▊   | 9126/13317 [1:34:06<42:59,  1.62it/s]
 69%|██████▊   | 9127/13317 [1:34:06<42:58,  1.63it/s]
                                                      
{'loss': 0.2026, 'grad_norm': 0.6836909651756287, 'learning_rate': 2.7245022856877933e-06, 'epoch': 2.06}

 69%|██████▊   | 9127/13317 [1:34:06<42:58,  1.63it/s]
 69%|██████▊   | 9128/13317 [1:34:07<42:55,  1.63it/s]
                                                      
{'loss': 0.1636, 'grad_norm': 0.6266347765922546, 'learning_rate': 2.723335321094507e-06, 'epoch': 2.06}

 69%|██████▊   | 9128/13317 [1:34:07<42:55,  1.63it/s]
 69%|██████▊   | 9129/13317 [1:34:07<42:57,  1.62it/s]
                                                      
{'loss': 0.1555, 'grad_norm': 0.48001840710639954, 'learning_rate': 2.7221685129321925e-06, 'epoch': 2.06}

 69%|██████▊   | 9129/13317 [1:34:07<42:57,  1.62it/s]
 69%|██████▊   | 9130/13317 [1:34:08<42:55,  1.63it/s]
                                                      
{'loss': 0.1243, 'grad_norm': 0.4118208885192871, 'learning_rate': 2.7210018612810186e-06, 'epoch': 2.06}

 69%|██████▊   | 9130/13317 [1:34:08<42:55,  1.63it/s]
 69%|██████▊   | 9131/13317 [1:34:09<42:59,  1.62it/s]
                                                      
{'loss': 0.1472, 'grad_norm': 0.5331549048423767, 'learning_rate': 2.719835366221146e-06, 'epoch': 2.06}

 69%|██████▊   | 9131/13317 [1:34:09<42:59,  1.62it/s]
 69%|██████▊   | 9132/13317 [1:34:09<43:54,  1.59it/s]
                                                      
{'loss': 0.1557, 'grad_norm': 0.5098785758018494, 'learning_rate': 2.7186690278327277e-06, 'epoch': 2.06}

 69%|██████▊   | 9132/13317 [1:34:09<43:54,  1.59it/s]
 69%|██████▊   | 9133/13317 [1:34:10<45:05,  1.55it/s]
                                                      
{'loss': 0.1463, 'grad_norm': 0.5243009924888611, 'learning_rate': 2.717502846195903e-06, 'epoch': 2.06}

 69%|██████▊   | 9133/13317 [1:34:10<45:05,  1.55it/s]
 69%|██████▊   | 9134/13317 [1:34:11<45:38,  1.53it/s]
                                                      
{'loss': 0.1761, 'grad_norm': 0.6051859259605408, 'learning_rate': 2.7163368213907975e-06, 'epoch': 2.06}

 69%|██████▊   | 9134/13317 [1:34:11<45:38,  1.53it/s]
 69%|██████▊   | 9135/13317 [1:34:11<45:37,  1.53it/s]
                                                      
{'loss': 0.1794, 'grad_norm': 0.6763930320739746, 'learning_rate': 2.715170953497532e-06, 'epoch': 2.06}

 69%|██████▊   | 9135/13317 [1:34:11<45:37,  1.53it/s]
 69%|██████▊   | 9136/13317 [1:34:12<44:57,  1.55it/s]
                                                      
{'loss': 0.1584, 'grad_norm': 0.5696524977684021, 'learning_rate': 2.714005242596217e-06, 'epoch': 2.06}

 69%|██████▊   | 9136/13317 [1:34:12<44:57,  1.55it/s]
 69%|██████▊   | 9137/13317 [1:34:12<44:15,  1.57it/s]
                                                      
{'loss': 0.1414, 'grad_norm': 0.5404953360557556, 'learning_rate': 2.712839688766946e-06, 'epoch': 2.06}

 69%|██████▊   | 9137/13317 [1:34:13<44:15,  1.57it/s]
 69%|██████▊   | 9138/13317 [1:34:13<43:46,  1.59it/s]
                                                      
{'loss': 0.1754, 'grad_norm': 0.5952796339988708, 'learning_rate': 2.711674292089804e-06, 'epoch': 2.06}

 69%|██████▊   | 9138/13317 [1:34:13<43:46,  1.59it/s]
 69%|██████▊   | 9139/13317 [1:34:14<43:32,  1.60it/s]
                                                      
{'loss': 0.1153, 'grad_norm': 0.4737461805343628, 'learning_rate': 2.7105090526448686e-06, 'epoch': 2.06}

 69%|██████▊   | 9139/13317 [1:34:14<43:32,  1.60it/s]
 69%|██████▊   | 9140/13317 [1:34:14<43:19,  1.61it/s]
                                                      
{'loss': 0.1362, 'grad_norm': 0.5042037963867188, 'learning_rate': 2.7093439705122005e-06, 'epoch': 2.06}

 69%|██████▊   | 9140/13317 [1:34:14<43:19,  1.61it/s]
 69%|██████▊   | 9141/13317 [1:34:15<43:08,  1.61it/s]
                                                      
{'loss': 0.1296, 'grad_norm': 0.48214372992515564, 'learning_rate': 2.7081790457718582e-06, 'epoch': 2.06}

 69%|██████▊   | 9141/13317 [1:34:15<43:08,  1.61it/s]
 69%|██████▊   | 9142/13317 [1:34:16<42:59,  1.62it/s]
                                                      
{'loss': 0.1525, 'grad_norm': 0.570961594581604, 'learning_rate': 2.7070142785038813e-06, 'epoch': 2.06}

 69%|██████▊   | 9142/13317 [1:34:16<42:59,  1.62it/s]
 69%|██████▊   | 9143/13317 [1:34:16<42:54,  1.62it/s]
                                                      
{'loss': 0.1439, 'grad_norm': 0.6884613037109375, 'learning_rate': 2.7058496687882994e-06, 'epoch': 2.06}

 69%|██████▊   | 9143/13317 [1:34:16<42:54,  1.62it/s]
 69%|██████▊   | 9144/13317 [1:34:17<42:53,  1.62it/s]
                                                      
{'loss': 0.1606, 'grad_norm': 0.6417830586433411, 'learning_rate': 2.7046852167051387e-06, 'epoch': 2.06}

 69%|██████▊   | 9144/13317 [1:34:17<42:53,  1.62it/s]
 69%|██████▊   | 9145/13317 [1:34:17<42:50,  1.62it/s]
                                                      
{'loss': 0.1516, 'grad_norm': 1.0842711925506592, 'learning_rate': 2.703520922334405e-06, 'epoch': 2.06}

 69%|██████▊   | 9145/13317 [1:34:18<42:50,  1.62it/s]
 69%|██████▊   | 9146/13317 [1:34:18<42:47,  1.62it/s]
                                                      
{'loss': 0.136, 'grad_norm': 0.5081523060798645, 'learning_rate': 2.702356785756101e-06, 'epoch': 2.06}

 69%|██████▊   | 9146/13317 [1:34:18<42:47,  1.62it/s]
 69%|██████▊   | 9147/13317 [1:34:19<42:46,  1.62it/s]
                                                      
{'loss': 0.1516, 'grad_norm': 0.5556577444076538, 'learning_rate': 2.701192807050212e-06, 'epoch': 2.06}

 69%|██████▊   | 9147/13317 [1:34:19<42:46,  1.62it/s]
 69%|██████▊   | 9148/13317 [1:34:19<42:44,  1.63it/s]
                                                      
{'loss': 0.1936, 'grad_norm': 0.7811702489852905, 'learning_rate': 2.700028986296719e-06, 'epoch': 2.06}

 69%|██████▊   | 9148/13317 [1:34:19<42:44,  1.63it/s]
 69%|██████▊   | 9149/13317 [1:34:20<42:57,  1.62it/s]
                                                      
{'loss': 0.1454, 'grad_norm': 0.5179230570793152, 'learning_rate': 2.6988653235755875e-06, 'epoch': 2.06}

 69%|██████▊   | 9149/13317 [1:34:20<42:57,  1.62it/s]
 69%|██████▊   | 9150/13317 [1:34:20<42:50,  1.62it/s]
                                                      
{'loss': 0.1644, 'grad_norm': 0.5898194909095764, 'learning_rate': 2.6977018189667704e-06, 'epoch': 2.06}

 69%|██████▊   | 9150/13317 [1:34:21<42:50,  1.62it/s]
 69%|██████▊   | 9151/13317 [1:34:21<42:45,  1.62it/s]
                                                      
{'loss': 0.1591, 'grad_norm': 0.5653402209281921, 'learning_rate': 2.696538472550218e-06, 'epoch': 2.06}

 69%|██████▊   | 9151/13317 [1:34:21<42:45,  1.62it/s]
 69%|██████▊   | 9152/13317 [1:34:22<42:43,  1.62it/s]
                                                      
{'loss': 0.2036, 'grad_norm': 0.6021991968154907, 'learning_rate': 2.69537528440586e-06, 'epoch': 2.06}

 69%|██████▊   | 9152/13317 [1:34:22<42:43,  1.62it/s]
 69%|██████▊   | 9153/13317 [1:34:22<42:41,  1.63it/s]
                                                      
{'loss': 0.1357, 'grad_norm': 0.4802173376083374, 'learning_rate': 2.694212254613624e-06, 'epoch': 2.06}

 69%|██████▊   | 9153/13317 [1:34:22<42:41,  1.63it/s]
 69%|██████▊   | 9154/13317 [1:34:23<42:40,  1.63it/s]
                                                      
{'loss': 0.1361, 'grad_norm': 0.47573164105415344, 'learning_rate': 2.693049383253418e-06, 'epoch': 2.06}

 69%|██████▊   | 9154/13317 [1:34:23<42:40,  1.63it/s]
 69%|██████▊   | 9155/13317 [1:34:24<42:38,  1.63it/s]
                                                      
{'loss': 0.1686, 'grad_norm': 0.6737121939659119, 'learning_rate': 2.691886670405147e-06, 'epoch': 2.06}

 69%|██████▊   | 9155/13317 [1:34:24<42:38,  1.63it/s]
 69%|██████▉   | 9156/13317 [1:34:24<42:36,  1.63it/s]
                                                      
{'loss': 0.1646, 'grad_norm': 0.6189615726470947, 'learning_rate': 2.6907241161487006e-06, 'epoch': 2.06}

 69%|██████▉   | 9156/13317 [1:34:24<42:36,  1.63it/s]
 69%|██████▉   | 9157/13317 [1:34:25<42:34,  1.63it/s]
                                                      
{'loss': 0.2178, 'grad_norm': 0.752666711807251, 'learning_rate': 2.6895617205639568e-06, 'epoch': 2.06}

 69%|██████▉   | 9157/13317 [1:34:25<42:34,  1.63it/s]
 69%|██████▉   | 9158/13317 [1:34:25<42:33,  1.63it/s]
                                                      
{'loss': 0.1581, 'grad_norm': 0.5907101631164551, 'learning_rate': 2.688399483730788e-06, 'epoch': 2.06}

 69%|██████▉   | 9158/13317 [1:34:26<42:33,  1.63it/s]
 69%|██████▉   | 9159/13317 [1:34:26<42:35,  1.63it/s]
                                                      
{'loss': 0.1294, 'grad_norm': 0.5026516914367676, 'learning_rate': 2.6872374057290505e-06, 'epoch': 2.06}

 69%|██████▉   | 9159/13317 [1:34:26<42:35,  1.63it/s]
 69%|██████▉   | 9160/13317 [1:34:27<42:32,  1.63it/s]
                                                      
{'loss': 0.1741, 'grad_norm': 0.5950522422790527, 'learning_rate': 2.68607548663859e-06, 'epoch': 2.06}

 69%|██████▉   | 9160/13317 [1:34:27<42:32,  1.63it/s]
 69%|██████▉   | 9161/13317 [1:34:27<42:31,  1.63it/s]
                                                      
{'loss': 0.1507, 'grad_norm': 0.5568944811820984, 'learning_rate': 2.6849137265392423e-06, 'epoch': 2.06}

 69%|██████▉   | 9161/13317 [1:34:27<42:31,  1.63it/s]
 69%|██████▉   | 9162/13317 [1:34:28<42:33,  1.63it/s]
                                                      
{'loss': 0.1339, 'grad_norm': 0.6346309185028076, 'learning_rate': 2.6837521255108366e-06, 'epoch': 2.06}

 69%|██████▉   | 9162/13317 [1:34:28<42:33,  1.63it/s]
 69%|██████▉   | 9163/13317 [1:34:28<42:31,  1.63it/s]
                                                      
{'loss': 0.1411, 'grad_norm': 0.4652269780635834, 'learning_rate': 2.682590683633185e-06, 'epoch': 2.06}

 69%|██████▉   | 9163/13317 [1:34:29<42:31,  1.63it/s]
 69%|██████▉   | 9164/13317 [1:34:29<42:29,  1.63it/s]
                                                      
{'loss': 0.1414, 'grad_norm': 0.5000174641609192, 'learning_rate': 2.681429400986088e-06, 'epoch': 2.06}

 69%|██████▉   | 9164/13317 [1:34:29<42:29,  1.63it/s]
 69%|██████▉   | 9165/13317 [1:34:30<42:31,  1.63it/s]
                                                      
{'loss': 0.111, 'grad_norm': 0.4758138954639435, 'learning_rate': 2.680268277649341e-06, 'epoch': 2.06}

 69%|██████▉   | 9165/13317 [1:34:30<42:31,  1.63it/s]
 69%|██████▉   | 9166/13317 [1:34:30<42:31,  1.63it/s]
                                                      
{'loss': 0.1584, 'grad_norm': 0.5742637515068054, 'learning_rate': 2.679107313702729e-06, 'epoch': 2.06}

 69%|██████▉   | 9166/13317 [1:34:30<42:31,  1.63it/s]
 69%|██████▉   | 9167/13317 [1:34:31<42:30,  1.63it/s]
                                                      
{'loss': 0.1399, 'grad_norm': 0.5696808099746704, 'learning_rate': 2.6779465092260145e-06, 'epoch': 2.07}

 69%|██████▉   | 9167/13317 [1:34:31<42:30,  1.63it/s]
 69%|██████▉   | 9168/13317 [1:34:32<42:29,  1.63it/s]
                                                      
{'loss': 0.1368, 'grad_norm': 0.4606858491897583, 'learning_rate': 2.6767858642989606e-06, 'epoch': 2.07}

 69%|██████▉   | 9168/13317 [1:34:32<42:29,  1.63it/s]
 69%|██████▉   | 9169/13317 [1:34:32<42:26,  1.63it/s]
                                                      
{'loss': 0.1557, 'grad_norm': 0.5674687623977661, 'learning_rate': 2.6756253790013194e-06, 'epoch': 2.07}

 69%|██████▉   | 9169/13317 [1:34:32<42:26,  1.63it/s]
 69%|██████▉   | 9170/13317 [1:34:33<42:25,  1.63it/s]
                                                      
{'loss': 0.1353, 'grad_norm': 0.531514585018158, 'learning_rate': 2.6744650534128245e-06, 'epoch': 2.07}

 69%|██████▉   | 9170/13317 [1:34:33<42:25,  1.63it/s]
 69%|██████▉   | 9171/13317 [1:34:33<42:23,  1.63it/s]
                                                      
{'loss': 0.1448, 'grad_norm': 0.5160131454467773, 'learning_rate': 2.6733048876132027e-06, 'epoch': 2.07}

 69%|██████▉   | 9171/13317 [1:34:34<42:23,  1.63it/s]
 69%|██████▉   | 9172/13317 [1:34:34<42:23,  1.63it/s]
                                                      
{'loss': 0.1481, 'grad_norm': 0.6403546333312988, 'learning_rate': 2.6721448816821697e-06, 'epoch': 2.07}

 69%|██████▉   | 9172/13317 [1:34:34<42:23,  1.63it/s]
 69%|██████▉   | 9173/13317 [1:34:35<42:23,  1.63it/s]
                                                      
{'loss': 0.1475, 'grad_norm': 0.5931554436683655, 'learning_rate': 2.6709850356994353e-06, 'epoch': 2.07}

 69%|██████▉   | 9173/13317 [1:34:35<42:23,  1.63it/s]
 69%|██████▉   | 9174/13317 [1:34:35<42:23,  1.63it/s]
                                                      
{'loss': 0.1095, 'grad_norm': 0.43663281202316284, 'learning_rate': 2.6698253497446853e-06, 'epoch': 2.07}

 69%|██████▉   | 9174/13317 [1:34:35<42:23,  1.63it/s]
 69%|██████▉   | 9175/13317 [1:34:36<42:22,  1.63it/s]
                                                      
{'loss': 0.1535, 'grad_norm': 0.578450083732605, 'learning_rate': 2.668665823897606e-06, 'epoch': 2.07}

 69%|██████▉   | 9175/13317 [1:34:36<42:22,  1.63it/s]
 69%|██████▉   | 9176/13317 [1:34:36<42:20,  1.63it/s]
                                                      
{'loss': 0.1512, 'grad_norm': 0.5713630318641663, 'learning_rate': 2.6675064582378708e-06, 'epoch': 2.07}

 69%|██████▉   | 9176/13317 [1:34:37<42:20,  1.63it/s]
 69%|██████▉   | 9177/13317 [1:34:37<42:18,  1.63it/s]
                                                      
{'loss': 0.1408, 'grad_norm': 0.5295129418373108, 'learning_rate': 2.6663472528451382e-06, 'epoch': 2.07}

 69%|██████▉   | 9177/13317 [1:34:37<42:18,  1.63it/s]
 69%|██████▉   | 9178/13317 [1:34:38<42:19,  1.63it/s]
                                                      
{'loss': 0.1614, 'grad_norm': 0.5940789580345154, 'learning_rate': 2.6651882077990574e-06, 'epoch': 2.07}

 69%|██████▉   | 9178/13317 [1:34:38<42:19,  1.63it/s]
 69%|██████▉   | 9179/13317 [1:34:38<42:19,  1.63it/s]
                                                      
{'loss': 0.1426, 'grad_norm': 0.5043054223060608, 'learning_rate': 2.664029323179268e-06, 'epoch': 2.07}

 69%|██████▉   | 9179/13317 [1:34:38<42:19,  1.63it/s]
 69%|██████▉   | 9180/13317 [1:34:39<42:18,  1.63it/s]
                                                      
{'loss': 0.1429, 'grad_norm': 0.5362478494644165, 'learning_rate': 2.6628705990654012e-06, 'epoch': 2.07}

 69%|██████▉   | 9180/13317 [1:34:39<42:18,  1.63it/s]
 69%|██████▉   | 9181/13317 [1:34:40<42:18,  1.63it/s]
                                                      
{'loss': 0.1504, 'grad_norm': 0.5144618153572083, 'learning_rate': 2.6617120355370667e-06, 'epoch': 2.07}

 69%|██████▉   | 9181/13317 [1:34:40<42:18,  1.63it/s]
 69%|██████▉   | 9182/13317 [1:34:40<42:18,  1.63it/s]
                                                      
{'loss': 0.1829, 'grad_norm': 0.6012714505195618, 'learning_rate': 2.6605536326738735e-06, 'epoch': 2.07}

 69%|██████▉   | 9182/13317 [1:34:40<42:18,  1.63it/s]
 69%|██████▉   | 9183/13317 [1:34:41<42:16,  1.63it/s]
                                                      
{'loss': 0.1219, 'grad_norm': 0.4515117108821869, 'learning_rate': 2.659395390555418e-06, 'epoch': 2.07}

 69%|██████▉   | 9183/13317 [1:34:41<42:16,  1.63it/s]
 69%|██████▉   | 9184/13317 [1:34:41<42:14,  1.63it/s]
                                                      
{'loss': 0.1341, 'grad_norm': 0.44484686851501465, 'learning_rate': 2.6582373092612825e-06, 'epoch': 2.07}

 69%|██████▉   | 9184/13317 [1:34:41<42:14,  1.63it/s]
 69%|██████▉   | 9185/13317 [1:34:42<42:13,  1.63it/s]
                                                      
{'loss': 0.1598, 'grad_norm': 0.5983505845069885, 'learning_rate': 2.6570793888710354e-06, 'epoch': 2.07}

 69%|██████▉   | 9185/13317 [1:34:42<42:13,  1.63it/s]
 69%|██████▉   | 9186/13317 [1:34:43<42:12,  1.63it/s]
                                                      
{'loss': 0.1966, 'grad_norm': 0.6022180914878845, 'learning_rate': 2.6559216294642446e-06, 'epoch': 2.07}

 69%|██████▉   | 9186/13317 [1:34:43<42:12,  1.63it/s]
 69%|██████▉   | 9187/13317 [1:34:43<42:13,  1.63it/s]
                                                      
{'loss': 0.1411, 'grad_norm': 0.4910987615585327, 'learning_rate': 2.6547640311204553e-06, 'epoch': 2.07}

 69%|██████▉   | 9187/13317 [1:34:43<42:13,  1.63it/s]
 69%|██████▉   | 9188/13317 [1:34:44<42:14,  1.63it/s]
                                                      
{'loss': 0.1577, 'grad_norm': 0.5242408514022827, 'learning_rate': 2.6536065939192107e-06, 'epoch': 2.07}

 69%|██████▉   | 9188/13317 [1:34:44<42:14,  1.63it/s]
 69%|██████▉   | 9189/13317 [1:34:44<42:16,  1.63it/s]
                                                      
{'loss': 0.139, 'grad_norm': 0.5113378763198853, 'learning_rate': 2.6524493179400353e-06, 'epoch': 2.07}

 69%|██████▉   | 9189/13317 [1:34:45<42:16,  1.63it/s]
 69%|██████▉   | 9190/13317 [1:34:45<42:15,  1.63it/s]
                                                      
{'loss': 0.1715, 'grad_norm': 0.6376482844352722, 'learning_rate': 2.65129220326245e-06, 'epoch': 2.07}

 69%|██████▉   | 9190/13317 [1:34:45<42:15,  1.63it/s]
 69%|██████▉   | 9191/13317 [1:34:46<42:17,  1.63it/s]
                                                      
{'loss': 0.1445, 'grad_norm': 0.5055913329124451, 'learning_rate': 2.650135249965958e-06, 'epoch': 2.07}

 69%|██████▉   | 9191/13317 [1:34:46<42:17,  1.63it/s]
 69%|██████▉   | 9192/13317 [1:34:46<42:31,  1.62it/s]
                                                      
{'loss': 0.1502, 'grad_norm': 0.5180767774581909, 'learning_rate': 2.6489784581300542e-06, 'epoch': 2.07}

 69%|██████▉   | 9192/13317 [1:34:46<42:31,  1.62it/s]
 69%|██████▉   | 9193/13317 [1:34:47<42:24,  1.62it/s]
                                                      
{'loss': 0.1744, 'grad_norm': 0.5536616444587708, 'learning_rate': 2.647821827834225e-06, 'epoch': 2.07}

 69%|██████▉   | 9193/13317 [1:34:47<42:24,  1.62it/s]
 69%|██████▉   | 9194/13317 [1:34:48<42:24,  1.62it/s]
                                                      
{'loss': 0.2305, 'grad_norm': 0.6507523655891418, 'learning_rate': 2.6466653591579393e-06, 'epoch': 2.07}

 69%|██████▉   | 9194/13317 [1:34:48<42:24,  1.62it/s]
 69%|██████▉   | 9195/13317 [1:34:48<42:21,  1.62it/s]
                                                      
{'loss': 0.1018, 'grad_norm': 0.4578722417354584, 'learning_rate': 2.645509052180663e-06, 'epoch': 2.07}

 69%|██████▉   | 9195/13317 [1:34:48<42:21,  1.62it/s]
 69%|██████▉   | 9196/13317 [1:34:49<42:15,  1.63it/s]
                                                      
{'loss': 0.1312, 'grad_norm': 0.4865834712982178, 'learning_rate': 2.644352906981843e-06, 'epoch': 2.07}

 69%|██████▉   | 9196/13317 [1:34:49<42:15,  1.63it/s]
 69%|██████▉   | 9197/13317 [1:34:49<42:13,  1.63it/s]
                                                      
{'loss': 0.1404, 'grad_norm': 0.5900213718414307, 'learning_rate': 2.6431969236409218e-06, 'epoch': 2.07}

 69%|██████▉   | 9197/13317 [1:34:49<42:13,  1.63it/s]
 69%|██████▉   | 9198/13317 [1:34:50<42:11,  1.63it/s]
                                                      
{'loss': 0.171, 'grad_norm': 0.5474534630775452, 'learning_rate': 2.642041102237326e-06, 'epoch': 2.07}

 69%|██████▉   | 9198/13317 [1:34:50<42:11,  1.63it/s]
 69%|██████▉   | 9199/13317 [1:34:51<42:12,  1.63it/s]
                                                      
{'loss': 0.1656, 'grad_norm': 0.5819205641746521, 'learning_rate': 2.6408854428504714e-06, 'epoch': 2.07}

 69%|██████▉   | 9199/13317 [1:34:51<42:12,  1.63it/s]
 69%|██████▉   | 9200/13317 [1:34:51<42:09,  1.63it/s]
                                                      
{'loss': 0.1354, 'grad_norm': 0.4924975633621216, 'learning_rate': 2.639729945559767e-06, 'epoch': 2.07}

 69%|██████▉   | 9200/13317 [1:34:51<42:09,  1.63it/s]
 69%|██████▉   | 9201/13317 [1:34:52<42:07,  1.63it/s]
                                                      
{'loss': 0.1508, 'grad_norm': 0.6541138291358948, 'learning_rate': 2.6385746104446047e-06, 'epoch': 2.07}

 69%|██████▉   | 9201/13317 [1:34:52<42:07,  1.63it/s]
 69%|██████▉   | 9202/13317 [1:34:52<42:06,  1.63it/s]
                                                      
{'loss': 0.175, 'grad_norm': 0.6600496172904968, 'learning_rate': 2.637419437584372e-06, 'epoch': 2.07}

 69%|██████▉   | 9202/13317 [1:34:53<42:06,  1.63it/s]
 69%|██████▉   | 9203/13317 [1:34:53<42:04,  1.63it/s]
                                                      
{'loss': 0.1336, 'grad_norm': 0.5848424434661865, 'learning_rate': 2.6362644270584392e-06, 'epoch': 2.07}

 69%|██████▉   | 9203/13317 [1:34:53<42:04,  1.63it/s]
 69%|██████▉   | 9204/13317 [1:34:54<42:06,  1.63it/s]
                                                      
{'loss': 0.1355, 'grad_norm': 0.4834032952785492, 'learning_rate': 2.6351095789461657e-06, 'epoch': 2.07}

 69%|██████▉   | 9204/13317 [1:34:54<42:06,  1.63it/s]
 69%|██████▉   | 9205/13317 [1:34:54<42:05,  1.63it/s]
                                                      
{'loss': 0.1551, 'grad_norm': 0.5837188363075256, 'learning_rate': 2.633954893326906e-06, 'epoch': 2.07}

 69%|██████▉   | 9205/13317 [1:34:54<42:05,  1.63it/s]
 69%|██████▉   | 9206/13317 [1:34:55<42:03,  1.63it/s]
                                                      
{'loss': 0.1655, 'grad_norm': 0.5476884245872498, 'learning_rate': 2.6328003702799955e-06, 'epoch': 2.07}

 69%|██████▉   | 9206/13317 [1:34:55<42:03,  1.63it/s]
 69%|██████▉   | 9207/13317 [1:34:56<42:03,  1.63it/s]
                                                      
{'loss': 0.176, 'grad_norm': 0.6203933954238892, 'learning_rate': 2.6316460098847656e-06, 'epoch': 2.07}

 69%|██████▉   | 9207/13317 [1:34:56<42:03,  1.63it/s]
 69%|██████▉   | 9208/13317 [1:34:56<42:01,  1.63it/s]
                                                      
{'loss': 0.1905, 'grad_norm': 0.6095455884933472, 'learning_rate': 2.6304918122205293e-06, 'epoch': 2.07}

 69%|██████▉   | 9208/13317 [1:34:56<42:01,  1.63it/s]
 69%|██████▉   | 9209/13317 [1:34:57<42:05,  1.63it/s]
                                                      
{'loss': 0.1526, 'grad_norm': 0.4737702012062073, 'learning_rate': 2.629337777366597e-06, 'epoch': 2.07}

 69%|██████▉   | 9209/13317 [1:34:57<42:05,  1.63it/s]
 69%|██████▉   | 9210/13317 [1:34:57<42:03,  1.63it/s]
                                                      
{'loss': 0.1572, 'grad_norm': 0.566830039024353, 'learning_rate': 2.6281839054022594e-06, 'epoch': 2.07}

 69%|██████▉   | 9210/13317 [1:34:57<42:03,  1.63it/s]
 69%|██████▉   | 9211/13317 [1:34:58<42:01,  1.63it/s]
                                                      
{'loss': 0.1388, 'grad_norm': 0.554639995098114, 'learning_rate': 2.627030196406799e-06, 'epoch': 2.08}

 69%|██████▉   | 9211/13317 [1:34:58<42:01,  1.63it/s]
 69%|██████▉   | 9212/13317 [1:34:59<41:59,  1.63it/s]
                                                      
{'loss': 0.1363, 'grad_norm': 0.5575802326202393, 'learning_rate': 2.625876650459492e-06, 'epoch': 2.08}

 69%|██████▉   | 9212/13317 [1:34:59<41:59,  1.63it/s]
 69%|██████▉   | 9213/13317 [1:34:59<42:00,  1.63it/s]
                                                      
{'loss': 0.1309, 'grad_norm': 0.4934747517108917, 'learning_rate': 2.624723267639594e-06, 'epoch': 2.08}

 69%|██████▉   | 9213/13317 [1:34:59<42:00,  1.63it/s]
 69%|██████▉   | 9214/13317 [1:35:00<42:01,  1.63it/s]
                                                      
{'loss': 0.1586, 'grad_norm': 0.5845583081245422, 'learning_rate': 2.62357004802636e-06, 'epoch': 2.08}

 69%|██████▉   | 9214/13317 [1:35:00<42:01,  1.63it/s]
 69%|██████▉   | 9215/13317 [1:35:00<41:59,  1.63it/s]
                                                      
{'loss': 0.2089, 'grad_norm': 0.6264524459838867, 'learning_rate': 2.6224169916990232e-06, 'epoch': 2.08}

 69%|██████▉   | 9215/13317 [1:35:01<41:59,  1.63it/s]
 69%|██████▉   | 9216/13317 [1:35:01<41:57,  1.63it/s]
                                                      
{'loss': 0.1886, 'grad_norm': 0.5718518495559692, 'learning_rate': 2.6212640987368155e-06, 'epoch': 2.08}

 69%|██████▉   | 9216/13317 [1:35:01<41:57,  1.63it/s]
 69%|██████▉   | 9217/13317 [1:35:02<41:57,  1.63it/s]
                                                      
{'loss': 0.1576, 'grad_norm': 0.5960370302200317, 'learning_rate': 2.6201113692189506e-06, 'epoch': 2.08}

 69%|██████▉   | 9217/13317 [1:35:02<41:57,  1.63it/s]
 69%|██████▉   | 9218/13317 [1:35:02<41:57,  1.63it/s]
                                                      
{'loss': 0.1548, 'grad_norm': 0.5905475616455078, 'learning_rate': 2.6189588032246314e-06, 'epoch': 2.08}

 69%|██████▉   | 9218/13317 [1:35:02<41:57,  1.63it/s]
 69%|██████▉   | 9219/13317 [1:35:03<42:10,  1.62it/s]
                                                      
{'loss': 0.1294, 'grad_norm': 0.4952625036239624, 'learning_rate': 2.617806400833053e-06, 'epoch': 2.08}

 69%|██████▉   | 9219/13317 [1:35:03<42:10,  1.62it/s]
 69%|██████▉   | 9220/13317 [1:35:04<42:07,  1.62it/s]
                                                      
{'loss': 0.1436, 'grad_norm': 0.5330820083618164, 'learning_rate': 2.6166541621234028e-06, 'epoch': 2.08}

 69%|██████▉   | 9220/13317 [1:35:04<42:07,  1.62it/s]
 69%|██████▉   | 9221/13317 [1:35:04<42:03,  1.62it/s]
                                                      
{'loss': 0.1181, 'grad_norm': 0.4300166964530945, 'learning_rate': 2.6155020871748427e-06, 'epoch': 2.08}

 69%|██████▉   | 9221/13317 [1:35:04<42:03,  1.62it/s]
 69%|██████▉   | 9222/13317 [1:35:05<41:58,  1.63it/s]
                                                      
{'loss': 0.166, 'grad_norm': 0.6594939827919006, 'learning_rate': 2.6143501760665357e-06, 'epoch': 2.08}

 69%|██████▉   | 9222/13317 [1:35:05<41:58,  1.63it/s]
 69%|██████▉   | 9223/13317 [1:35:05<41:55,  1.63it/s]
                                                      
{'loss': 0.1292, 'grad_norm': 0.5706667900085449, 'learning_rate': 2.613198428877634e-06, 'epoch': 2.08}

 69%|██████▉   | 9223/13317 [1:35:05<41:55,  1.63it/s]
 69%|██████▉   | 9224/13317 [1:35:06<41:56,  1.63it/s]
                                                      
{'loss': 0.1914, 'grad_norm': 0.61237633228302, 'learning_rate': 2.612046845687272e-06, 'epoch': 2.08}

 69%|██████▉   | 9224/13317 [1:35:06<41:56,  1.63it/s]
 69%|██████▉   | 9225/13317 [1:35:07<42:04,  1.62it/s]
                                                      
{'loss': 0.1442, 'grad_norm': 0.5107315182685852, 'learning_rate': 2.6108954265745738e-06, 'epoch': 2.08}

 69%|██████▉   | 9225/13317 [1:35:07<42:04,  1.62it/s]
 69%|██████▉   | 9226/13317 [1:35:07<42:00,  1.62it/s]
                                                      
{'loss': 0.1663, 'grad_norm': 0.5485401749610901, 'learning_rate': 2.609744171618656e-06, 'epoch': 2.08}

 69%|██████▉   | 9226/13317 [1:35:07<42:00,  1.62it/s]
 69%|██████▉   | 9227/13317 [1:35:08<41:56,  1.63it/s]
                                                      
{'loss': 0.1737, 'grad_norm': 0.6056250929832458, 'learning_rate': 2.608593080898626e-06, 'epoch': 2.08}

 69%|██████▉   | 9227/13317 [1:35:08<41:56,  1.63it/s]
 69%|██████▉   | 9228/13317 [1:35:08<41:53,  1.63it/s]
                                                      
{'loss': 0.1295, 'grad_norm': 0.5062000155448914, 'learning_rate': 2.607442154493568e-06, 'epoch': 2.08}

 69%|██████▉   | 9228/13317 [1:35:09<41:53,  1.63it/s]
 69%|██████▉   | 9229/13317 [1:35:09<41:55,  1.63it/s]
                                                      
{'loss': 0.131, 'grad_norm': 0.4912949502468109, 'learning_rate': 2.6062913924825674e-06, 'epoch': 2.08}

 69%|██████▉   | 9229/13317 [1:35:09<41:55,  1.63it/s]
 69%|██████▉   | 9230/13317 [1:35:10<41:54,  1.63it/s]
                                                      
{'loss': 0.1586, 'grad_norm': 0.5201298594474792, 'learning_rate': 2.605140794944694e-06, 'epoch': 2.08}

 69%|██████▉   | 9230/13317 [1:35:10<41:54,  1.63it/s]
 69%|██████▉   | 9231/13317 [1:35:10<41:53,  1.63it/s]
                                                      
{'loss': 0.1375, 'grad_norm': 0.4836471974849701, 'learning_rate': 2.6039903619590057e-06, 'epoch': 2.08}

 69%|██████▉   | 9231/13317 [1:35:10<41:53,  1.63it/s]
 69%|██████▉   | 9232/13317 [1:35:11<41:52,  1.63it/s]
                                                      
{'loss': 0.1375, 'grad_norm': 0.5532448887825012, 'learning_rate': 2.6028400936045473e-06, 'epoch': 2.08}

 69%|██████▉   | 9232/13317 [1:35:11<41:52,  1.63it/s]
 69%|██████▉   | 9233/13317 [1:35:11<41:50,  1.63it/s]
                                                      
{'loss': 0.1793, 'grad_norm': 0.6195665001869202, 'learning_rate': 2.6016899899603554e-06, 'epoch': 2.08}

 69%|██████▉   | 9233/13317 [1:35:12<41:50,  1.63it/s]
 69%|██████▉   | 9234/13317 [1:35:12<41:54,  1.62it/s]
                                                      
{'loss': 0.1856, 'grad_norm': 0.6175163388252258, 'learning_rate': 2.600540051105459e-06, 'epoch': 2.08}

 69%|██████▉   | 9234/13317 [1:35:12<41:54,  1.62it/s]
 69%|██████▉   | 9235/13317 [1:35:13<41:52,  1.62it/s]
                                                      
{'loss': 0.1823, 'grad_norm': 0.6163903474807739, 'learning_rate': 2.5993902771188627e-06, 'epoch': 2.08}

 69%|██████▉   | 9235/13317 [1:35:13<41:52,  1.62it/s]
 69%|██████▉   | 9236/13317 [1:35:13<41:49,  1.63it/s]
                                                      
{'loss': 0.1429, 'grad_norm': 0.48094117641448975, 'learning_rate': 2.5982406680795733e-06, 'epoch': 2.08}

 69%|██████▉   | 9236/13317 [1:35:13<41:49,  1.63it/s]
 69%|██████▉   | 9237/13317 [1:35:14<41:47,  1.63it/s]
                                                      
{'loss': 0.135, 'grad_norm': 0.5013766288757324, 'learning_rate': 2.5970912240665815e-06, 'epoch': 2.08}

 69%|██████▉   | 9237/13317 [1:35:14<41:47,  1.63it/s]
 69%|██████▉   | 9238/13317 [1:35:15<41:44,  1.63it/s]
                                                      
{'loss': 0.192, 'grad_norm': 0.6809030175209045, 'learning_rate': 2.5959419451588653e-06, 'epoch': 2.08}

 69%|██████▉   | 9238/13317 [1:35:15<41:44,  1.63it/s]
 69%|██████▉   | 9239/13317 [1:35:15<41:45,  1.63it/s]
                                                      
{'loss': 0.2256, 'grad_norm': 0.6431143879890442, 'learning_rate': 2.59479283143539e-06, 'epoch': 2.08}

 69%|██████▉   | 9239/13317 [1:35:15<41:45,  1.63it/s]
 69%|██████▉   | 9240/13317 [1:35:16<41:43,  1.63it/s]
                                                      
{'loss': 0.1309, 'grad_norm': 0.518124520778656, 'learning_rate': 2.5936438829751143e-06, 'epoch': 2.08}

 69%|██████▉   | 9240/13317 [1:35:16<41:43,  1.63it/s]
 69%|██████▉   | 9241/13317 [1:35:16<41:41,  1.63it/s]
                                                      
{'loss': 0.1194, 'grad_norm': 0.4414824843406677, 'learning_rate': 2.592495099856984e-06, 'epoch': 2.08}

 69%|██████▉   | 9241/13317 [1:35:17<41:41,  1.63it/s]
 69%|██████▉   | 9242/13317 [1:35:17<41:52,  1.62it/s]
                                                      
{'loss': 0.1875, 'grad_norm': 0.6660076379776001, 'learning_rate': 2.591346482159932e-06, 'epoch': 2.08}

 69%|██████▉   | 9242/13317 [1:35:17<41:52,  1.62it/s]
 69%|██████▉   | 9243/13317 [1:35:18<41:47,  1.62it/s]
                                                      
{'loss': 0.1246, 'grad_norm': 0.4660392701625824, 'learning_rate': 2.5901980299628776e-06, 'epoch': 2.08}

 69%|██████▉   | 9243/13317 [1:35:18<41:47,  1.62it/s]
 69%|██████▉   | 9244/13317 [1:35:18<41:47,  1.62it/s]
                                                      
{'loss': 0.1594, 'grad_norm': 0.7968282699584961, 'learning_rate': 2.589049743344736e-06, 'epoch': 2.08}

 69%|██████▉   | 9244/13317 [1:35:18<41:47,  1.62it/s]
 69%|██████▉   | 9245/13317 [1:35:19<41:45,  1.63it/s]
                                                      
{'loss': 0.139, 'grad_norm': 0.5630068182945251, 'learning_rate': 2.5879016223844043e-06, 'epoch': 2.08}

 69%|██████▉   | 9245/13317 [1:35:19<41:45,  1.63it/s]
 69%|██████▉   | 9246/13317 [1:35:19<41:43,  1.63it/s]
                                                      
{'loss': 0.1773, 'grad_norm': 0.5935696363449097, 'learning_rate': 2.5867536671607695e-06, 'epoch': 2.08}

 69%|██████▉   | 9246/13317 [1:35:20<41:43,  1.63it/s]
 69%|██████▉   | 9247/13317 [1:35:20<41:41,  1.63it/s]
                                                      
{'loss': 0.1457, 'grad_norm': 0.5329090356826782, 'learning_rate': 2.5856058777527115e-06, 'epoch': 2.08}

 69%|██████▉   | 9247/13317 [1:35:20<41:41,  1.63it/s]
 69%|██████▉   | 9248/13317 [1:35:21<41:40,  1.63it/s]
                                                      
{'loss': 0.1393, 'grad_norm': 0.49066588282585144, 'learning_rate': 2.5844582542390915e-06, 'epoch': 2.08}

 69%|██████▉   | 9248/13317 [1:35:21<41:40,  1.63it/s]
 69%|██████▉   | 9249/13317 [1:35:21<41:40,  1.63it/s]
                                                      
{'loss': 0.1373, 'grad_norm': 0.5304970145225525, 'learning_rate': 2.5833107966987676e-06, 'epoch': 2.08}

 69%|██████▉   | 9249/13317 [1:35:21<41:40,  1.63it/s]
 69%|██████▉   | 9250/13317 [1:35:22<41:39,  1.63it/s]
                                                      
{'loss': 0.1485, 'grad_norm': 0.5304846167564392, 'learning_rate': 2.582163505210578e-06, 'epoch': 2.08}

 69%|██████▉   | 9250/13317 [1:35:22<41:39,  1.63it/s]
 69%|██████▉   | 9251/13317 [1:35:23<41:37,  1.63it/s]
                                                      
{'loss': 0.1989, 'grad_norm': 0.6729828119277954, 'learning_rate': 2.5810163798533583e-06, 'epoch': 2.08}

 69%|██████▉   | 9251/13317 [1:35:23<41:37,  1.63it/s]
 69%|██████▉   | 9252/13317 [1:35:23<41:37,  1.63it/s]
                                                      
{'loss': 0.1298, 'grad_norm': 0.46260055899620056, 'learning_rate': 2.5798694207059263e-06, 'epoch': 2.08}

 69%|██████▉   | 9252/13317 [1:35:23<41:37,  1.63it/s]
 69%|██████▉   | 9253/13317 [1:35:24<41:34,  1.63it/s]
                                                      
{'loss': 0.1451, 'grad_norm': 0.54020756483078, 'learning_rate': 2.578722627847087e-06, 'epoch': 2.08}

 69%|██████▉   | 9253/13317 [1:35:24<41:34,  1.63it/s]
 69%|██████▉   | 9254/13317 [1:35:24<41:37,  1.63it/s]
                                                      
{'loss': 0.1761, 'grad_norm': 0.6126853823661804, 'learning_rate': 2.5775760013556426e-06, 'epoch': 2.08}

 69%|██████▉   | 9254/13317 [1:35:25<41:37,  1.63it/s]
 69%|██████▉   | 9255/13317 [1:35:25<41:35,  1.63it/s]
                                                      
{'loss': 0.1607, 'grad_norm': 0.5465840101242065, 'learning_rate': 2.576429541310374e-06, 'epoch': 2.08}

 69%|██████▉   | 9255/13317 [1:35:25<41:35,  1.63it/s]
 70%|██████▉   | 9256/13317 [1:35:26<41:33,  1.63it/s]
                                                      
{'loss': 0.1691, 'grad_norm': 0.6577282547950745, 'learning_rate': 2.575283247790059e-06, 'epoch': 2.09}

 70%|██████▉   | 9256/13317 [1:35:26<41:33,  1.63it/s]
 70%|██████▉   | 9257/13317 [1:35:26<41:32,  1.63it/s]
                                                      
{'loss': 0.1531, 'grad_norm': 0.5429136753082275, 'learning_rate': 2.574137120873456e-06, 'epoch': 2.09}

 70%|██████▉   | 9257/13317 [1:35:26<41:32,  1.63it/s]
 70%|██████▉   | 9258/13317 [1:35:27<41:32,  1.63it/s]
                                                      
{'loss': 0.1559, 'grad_norm': 0.566487729549408, 'learning_rate': 2.572991160639321e-06, 'epoch': 2.09}

 70%|██████▉   | 9258/13317 [1:35:27<41:32,  1.63it/s]
 70%|██████▉   | 9259/13317 [1:35:27<41:34,  1.63it/s]
                                                      
{'loss': 0.1628, 'grad_norm': 0.5914864540100098, 'learning_rate': 2.57184536716639e-06, 'epoch': 2.09}

 70%|██████▉   | 9259/13317 [1:35:28<41:34,  1.63it/s]
 70%|██████▉   | 9260/13317 [1:35:28<41:32,  1.63it/s]
                                                      
{'loss': 0.1358, 'grad_norm': 0.46651703119277954, 'learning_rate': 2.57069974053339e-06, 'epoch': 2.09}

 70%|██████▉   | 9260/13317 [1:35:28<41:32,  1.63it/s]
 70%|██████▉   | 9261/13317 [1:35:29<41:31,  1.63it/s]
                                                      
{'loss': 0.1423, 'grad_norm': 0.5155518054962158, 'learning_rate': 2.569554280819042e-06, 'epoch': 2.09}

 70%|██████▉   | 9261/13317 [1:35:29<41:31,  1.63it/s]
 70%|██████▉   | 9262/13317 [1:35:29<41:30,  1.63it/s]
                                                      
{'loss': 0.1491, 'grad_norm': 0.5704249143600464, 'learning_rate': 2.568408988102047e-06, 'epoch': 2.09}

 70%|██████▉   | 9262/13317 [1:35:29<41:30,  1.63it/s]
 70%|██████▉   | 9263/13317 [1:35:30<41:28,  1.63it/s]
                                                      
{'loss': 0.1124, 'grad_norm': 0.4659474790096283, 'learning_rate': 2.5672638624611022e-06, 'epoch': 2.09}

 70%|██████▉   | 9263/13317 [1:35:30<41:28,  1.63it/s]
 70%|██████▉   | 9264/13317 [1:35:31<41:31,  1.63it/s]
                                                      
{'loss': 0.1407, 'grad_norm': 0.5730624198913574, 'learning_rate': 2.5661189039748887e-06, 'epoch': 2.09}

 70%|██████▉   | 9264/13317 [1:35:31<41:31,  1.63it/s]
 70%|██████▉   | 9265/13317 [1:35:31<41:28,  1.63it/s]
                                                      
{'loss': 0.1489, 'grad_norm': 0.4922029972076416, 'learning_rate': 2.5649741127220745e-06, 'epoch': 2.09}

 70%|██████▉   | 9265/13317 [1:35:31<41:28,  1.63it/s]
 70%|██████▉   | 9266/13317 [1:35:32<41:27,  1.63it/s]
                                                      
{'loss': 0.1322, 'grad_norm': 0.517259418964386, 'learning_rate': 2.5638294887813216e-06, 'epoch': 2.09}

 70%|██████▉   | 9266/13317 [1:35:32<41:27,  1.63it/s]
 70%|██████▉   | 9267/13317 [1:35:32<41:25,  1.63it/s]
                                                      
{'loss': 0.1313, 'grad_norm': 0.49361491203308105, 'learning_rate': 2.562685032231279e-06, 'epoch': 2.09}

 70%|██████▉   | 9267/13317 [1:35:33<41:25,  1.63it/s]
 70%|██████▉   | 9268/13317 [1:35:33<41:23,  1.63it/s]
                                                      
{'loss': 0.1488, 'grad_norm': 0.518802285194397, 'learning_rate': 2.5615407431505813e-06, 'epoch': 2.09}

 70%|██████▉   | 9268/13317 [1:35:33<41:23,  1.63it/s]
 70%|██████▉   | 9269/13317 [1:35:34<41:26,  1.63it/s]
                                                      
{'loss': 0.1391, 'grad_norm': 0.5501248240470886, 'learning_rate': 2.560396621617852e-06, 'epoch': 2.09}

 70%|██████▉   | 9269/13317 [1:35:34<41:26,  1.63it/s]
 70%|██████▉   | 9270/13317 [1:35:34<41:24,  1.63it/s]
                                                      
{'loss': 0.1346, 'grad_norm': 0.554878294467926, 'learning_rate': 2.559252667711707e-06, 'epoch': 2.09}

 70%|██████▉   | 9270/13317 [1:35:34<41:24,  1.63it/s]
 70%|██████▉   | 9271/13317 [1:35:35<41:24,  1.63it/s]
                                                      
{'loss': 0.1355, 'grad_norm': 0.5699017643928528, 'learning_rate': 2.558108881510747e-06, 'epoch': 2.09}

 70%|██████▉   | 9271/13317 [1:35:35<41:24,  1.63it/s]
 70%|██████▉   | 9272/13317 [1:35:35<41:23,  1.63it/s]
                                                      
{'loss': 0.1873, 'grad_norm': 0.5980398058891296, 'learning_rate': 2.5569652630935604e-06, 'epoch': 2.09}

 70%|██████▉   | 9272/13317 [1:35:36<41:23,  1.63it/s]
 70%|██████▉   | 9273/13317 [1:35:36<41:22,  1.63it/s]
                                                      
{'loss': 0.1793, 'grad_norm': 0.6269034147262573, 'learning_rate': 2.5558218125387274e-06, 'epoch': 2.09}

 70%|██████▉   | 9273/13317 [1:35:36<41:22,  1.63it/s]
 70%|██████▉   | 9274/13317 [1:35:37<41:21,  1.63it/s]
                                                      
{'loss': 0.1395, 'grad_norm': 0.5226629972457886, 'learning_rate': 2.5546785299248166e-06, 'epoch': 2.09}

 70%|██████▉   | 9274/13317 [1:35:37<41:21,  1.63it/s]
 70%|██████▉   | 9275/13317 [1:35:37<41:22,  1.63it/s]
                                                      
{'loss': 0.1849, 'grad_norm': 0.58719801902771, 'learning_rate': 2.5535354153303827e-06, 'epoch': 2.09}

 70%|██████▉   | 9275/13317 [1:35:37<41:22,  1.63it/s]
 70%|██████▉   | 9276/13317 [1:35:38<41:22,  1.63it/s]
                                                      
{'loss': 0.1374, 'grad_norm': 0.5621381402015686, 'learning_rate': 2.5523924688339676e-06, 'epoch': 2.09}

 70%|██████▉   | 9276/13317 [1:35:38<41:22,  1.63it/s]
 70%|██████▉   | 9277/13317 [1:35:39<41:19,  1.63it/s]
                                                      
{'loss': 0.1558, 'grad_norm': 0.642850935459137, 'learning_rate': 2.551249690514107e-06, 'epoch': 2.09}

 70%|██████▉   | 9277/13317 [1:35:39<41:19,  1.63it/s]
 70%|██████▉   | 9278/13317 [1:35:39<41:18,  1.63it/s]
                                                      
{'loss': 0.1305, 'grad_norm': 0.5005557537078857, 'learning_rate': 2.550107080449321e-06, 'epoch': 2.09}

 70%|██████▉   | 9278/13317 [1:35:39<41:18,  1.63it/s]
 70%|██████▉   | 9279/13317 [1:35:40<41:18,  1.63it/s]
                                                      
{'loss': 0.159, 'grad_norm': 0.5691969394683838, 'learning_rate': 2.5489646387181166e-06, 'epoch': 2.09}

 70%|██████▉   | 9279/13317 [1:35:40<41:18,  1.63it/s]
 70%|██████▉   | 9280/13317 [1:35:40<41:17,  1.63it/s]
                                                      
{'loss': 0.1458, 'grad_norm': 0.6093635559082031, 'learning_rate': 2.5478223653989933e-06, 'epoch': 2.09}

 70%|██████▉   | 9280/13317 [1:35:41<41:17,  1.63it/s]
 70%|██████▉   | 9281/13317 [1:35:41<41:16,  1.63it/s]
                                                      
{'loss': 0.2019, 'grad_norm': 0.600604772567749, 'learning_rate': 2.546680260570441e-06, 'epoch': 2.09}

 70%|██████▉   | 9281/13317 [1:35:41<41:16,  1.63it/s]
 70%|██████▉   | 9282/13317 [1:35:42<41:15,  1.63it/s]
                                                      
{'loss': 0.176, 'grad_norm': 0.6229565143585205, 'learning_rate': 2.5455383243109277e-06, 'epoch': 2.09}

 70%|██████▉   | 9282/13317 [1:35:42<41:15,  1.63it/s]
 70%|██████▉   | 9283/13317 [1:35:42<41:15,  1.63it/s]
                                                      
{'loss': 0.1809, 'grad_norm': 0.6474100947380066, 'learning_rate': 2.5443965566989194e-06, 'epoch': 2.09}

 70%|██████▉   | 9283/13317 [1:35:42<41:15,  1.63it/s]
 70%|██████▉   | 9284/13317 [1:35:43<41:14,  1.63it/s]
                                                      
{'loss': 0.1523, 'grad_norm': 0.5316413640975952, 'learning_rate': 2.5432549578128703e-06, 'epoch': 2.09}

 70%|██████▉   | 9284/13317 [1:35:43<41:14,  1.63it/s]
 70%|██████▉   | 9285/13317 [1:35:43<41:13,  1.63it/s]
                                                      
{'loss': 0.1678, 'grad_norm': 0.5930626392364502, 'learning_rate': 2.5421135277312175e-06, 'epoch': 2.09}

 70%|██████▉   | 9285/13317 [1:35:44<41:13,  1.63it/s]
 70%|██████▉   | 9286/13317 [1:35:44<41:12,  1.63it/s]
                                                      
{'loss': 0.1864, 'grad_norm': 0.6550801396369934, 'learning_rate': 2.5409722665323885e-06, 'epoch': 2.09}

 70%|██████▉   | 9286/13317 [1:35:44<41:12,  1.63it/s]
 70%|██████▉   | 9287/13317 [1:35:45<41:10,  1.63it/s]
                                                      
{'loss': 0.105, 'grad_norm': 0.4172692596912384, 'learning_rate': 2.5398311742948014e-06, 'epoch': 2.09}

 70%|██████▉   | 9287/13317 [1:35:45<41:10,  1.63it/s]
 70%|██████▉   | 9288/13317 [1:35:45<41:10,  1.63it/s]
                                                      
{'loss': 0.1409, 'grad_norm': 0.5306200981140137, 'learning_rate': 2.5386902510968627e-06, 'epoch': 2.09}

 70%|██████▉   | 9288/13317 [1:35:45<41:10,  1.63it/s]
 70%|██████▉   | 9289/13317 [1:35:46<41:10,  1.63it/s]
                                                      
{'loss': 0.1586, 'grad_norm': 0.5357498526573181, 'learning_rate': 2.5375494970169645e-06, 'epoch': 2.09}

 70%|██████▉   | 9289/13317 [1:35:46<41:10,  1.63it/s]
 70%|██████▉   | 9290/13317 [1:35:47<41:08,  1.63it/s]
                                                      
{'loss': 0.1818, 'grad_norm': 0.6085326671600342, 'learning_rate': 2.5364089121334866e-06, 'epoch': 2.09}

 70%|██████▉   | 9290/13317 [1:35:47<41:08,  1.63it/s]
 70%|██████▉   | 9291/13317 [1:35:47<41:08,  1.63it/s]
                                                      
{'loss': 0.1596, 'grad_norm': 0.4941132366657257, 'learning_rate': 2.5352684965248044e-06, 'epoch': 2.09}

 70%|██████▉   | 9291/13317 [1:35:47<41:08,  1.63it/s]
 70%|██████▉   | 9292/13317 [1:35:48<41:09,  1.63it/s]
                                                      
{'loss': 0.1398, 'grad_norm': 0.5302123427391052, 'learning_rate': 2.534128250269273e-06, 'epoch': 2.09}

 70%|██████▉   | 9292/13317 [1:35:48<41:09,  1.63it/s]
 70%|██████▉   | 9293/13317 [1:35:48<41:07,  1.63it/s]
                                                      
{'loss': 0.2086, 'grad_norm': 0.6743387579917908, 'learning_rate': 2.5329881734452373e-06, 'epoch': 2.09}

 70%|██████▉   | 9293/13317 [1:35:48<41:07,  1.63it/s]
 70%|██████▉   | 9294/13317 [1:35:49<41:08,  1.63it/s]
                                                      
{'loss': 0.1541, 'grad_norm': 0.5631728768348694, 'learning_rate': 2.5318482661310356e-06, 'epoch': 2.09}

 70%|██████▉   | 9294/13317 [1:35:49<41:08,  1.63it/s]
 70%|██████▉   | 9295/13317 [1:35:50<41:09,  1.63it/s]
                                                      
{'loss': 0.1447, 'grad_norm': 0.5597543120384216, 'learning_rate': 2.5307085284049933e-06, 'epoch': 2.09}

 70%|██████▉   | 9295/13317 [1:35:50<41:09,  1.63it/s]
 70%|██████▉   | 9296/13317 [1:35:50<42:15,  1.59it/s]
                                                      
{'loss': 0.1521, 'grad_norm': 0.5423039197921753, 'learning_rate': 2.52956896034542e-06, 'epoch': 2.09}

 70%|██████▉   | 9296/13317 [1:35:50<42:15,  1.59it/s]
 70%|██████▉   | 9297/13317 [1:35:51<41:55,  1.60it/s]
                                                      
{'loss': 0.122, 'grad_norm': 0.4863179922103882, 'learning_rate': 2.5284295620306153e-06, 'epoch': 2.09}

 70%|██████▉   | 9297/13317 [1:35:51<41:55,  1.60it/s]
 70%|██████▉   | 9298/13317 [1:35:51<41:40,  1.61it/s]
                                                      
{'loss': 0.1183, 'grad_norm': 0.5004451870918274, 'learning_rate': 2.5272903335388704e-06, 'epoch': 2.09}

 70%|██████▉   | 9298/13317 [1:35:52<41:40,  1.61it/s]
 70%|██████▉   | 9299/13317 [1:35:52<41:33,  1.61it/s]
                                                      
{'loss': 0.1388, 'grad_norm': 0.51177579164505, 'learning_rate': 2.5261512749484615e-06, 'epoch': 2.09}

 70%|██████▉   | 9299/13317 [1:35:52<41:33,  1.61it/s]
 70%|██████▉   | 9300/13317 [1:35:53<41:26,  1.62it/s]
                                                      
{'loss': 0.1574, 'grad_norm': 0.5822045803070068, 'learning_rate': 2.525012386337651e-06, 'epoch': 2.1}

 70%|██████▉   | 9300/13317 [1:35:53<41:26,  1.62it/s]
 70%|██████▉   | 9301/13317 [1:35:53<41:19,  1.62it/s]
                                                      
{'loss': 0.1614, 'grad_norm': 0.5469629764556885, 'learning_rate': 2.5238736677846955e-06, 'epoch': 2.1}

 70%|██████▉   | 9301/13317 [1:35:53<41:19,  1.62it/s]
 70%|██████▉   | 9302/13317 [1:35:54<41:13,  1.62it/s]
                                                      
{'loss': 0.169, 'grad_norm': 0.5908411145210266, 'learning_rate': 2.5227351193678385e-06, 'epoch': 2.1}

 70%|██████▉   | 9302/13317 [1:35:54<41:13,  1.62it/s]
 70%|██████▉   | 9303/13317 [1:35:55<41:08,  1.63it/s]
                                                      
{'loss': 0.1363, 'grad_norm': 0.5175959467887878, 'learning_rate': 2.5215967411653087e-06, 'epoch': 2.1}

 70%|██████▉   | 9303/13317 [1:35:55<41:08,  1.63it/s]
 70%|██████▉   | 9304/13317 [1:35:55<41:11,  1.62it/s]
                                                      
{'loss': 0.2048, 'grad_norm': 0.6347358822822571, 'learning_rate': 2.520458533255322e-06, 'epoch': 2.1}

 70%|██████▉   | 9304/13317 [1:35:55<41:11,  1.62it/s]
 70%|██████▉   | 9305/13317 [1:35:56<41:07,  1.63it/s]
                                                      
{'loss': 0.1384, 'grad_norm': 0.49396568536758423, 'learning_rate': 2.519320495716091e-06, 'epoch': 2.1}

 70%|██████▉   | 9305/13317 [1:35:56<41:07,  1.63it/s]
 70%|██████▉   | 9306/13317 [1:35:56<41:05,  1.63it/s]
                                                      
{'loss': 0.1474, 'grad_norm': 0.5414527654647827, 'learning_rate': 2.518182628625807e-06, 'epoch': 2.1}

 70%|██████▉   | 9306/13317 [1:35:57<41:05,  1.63it/s]
 70%|██████▉   | 9307/13317 [1:35:57<41:04,  1.63it/s]
                                                      
{'loss': 0.1441, 'grad_norm': 0.4820500314235687, 'learning_rate': 2.517044932062652e-06, 'epoch': 2.1}

 70%|██████▉   | 9307/13317 [1:35:57<41:04,  1.63it/s]
 70%|██████▉   | 9308/13317 [1:35:58<41:01,  1.63it/s]
                                                      
{'loss': 0.1961, 'grad_norm': 0.6044034957885742, 'learning_rate': 2.515907406104803e-06, 'epoch': 2.1}

 70%|██████▉   | 9308/13317 [1:35:58<41:01,  1.63it/s]
 70%|██████▉   | 9309/13317 [1:35:58<40:59,  1.63it/s]
                                                      
{'loss': 0.2072, 'grad_norm': 0.8854067921638489, 'learning_rate': 2.5147700508304146e-06, 'epoch': 2.1}

 70%|██████▉   | 9309/13317 [1:35:58<40:59,  1.63it/s]
 70%|██████▉   | 9310/13317 [1:35:59<40:58,  1.63it/s]
                                                      
{'loss': 0.1126, 'grad_norm': 0.44058895111083984, 'learning_rate': 2.5136328663176403e-06, 'epoch': 2.1}

 70%|██████▉   | 9310/13317 [1:35:59<40:58,  1.63it/s]
 70%|██████▉   | 9311/13317 [1:35:59<40:57,  1.63it/s]
                                                      
{'loss': 0.1238, 'grad_norm': 0.5030642747879028, 'learning_rate': 2.5124958526446114e-06, 'epoch': 2.1}

 70%|██████▉   | 9311/13317 [1:36:00<40:57,  1.63it/s]
 70%|██████▉   | 9312/13317 [1:36:00<40:57,  1.63it/s]
                                                      
{'loss': 0.1632, 'grad_norm': 0.621199369430542, 'learning_rate': 2.511359009889458e-06, 'epoch': 2.1}

 70%|██████▉   | 9312/13317 [1:36:00<40:57,  1.63it/s]
 70%|██████▉   | 9313/13317 [1:36:01<40:56,  1.63it/s]
                                                      
{'loss': 0.1427, 'grad_norm': 0.5022540092468262, 'learning_rate': 2.5102223381302902e-06, 'epoch': 2.1}

 70%|██████▉   | 9313/13317 [1:36:01<40:56,  1.63it/s]
 70%|██████▉   | 9314/13317 [1:36:01<40:55,  1.63it/s]
                                                      
{'loss': 0.1455, 'grad_norm': 0.599838376045227, 'learning_rate': 2.5090858374452086e-06, 'epoch': 2.1}

 70%|██████▉   | 9314/13317 [1:36:01<40:55,  1.63it/s]
 70%|██████▉   | 9315/13317 [1:36:02<40:54,  1.63it/s]
                                                      
{'loss': 0.1506, 'grad_norm': 0.5666728019714355, 'learning_rate': 2.507949507912306e-06, 'epoch': 2.1}

 70%|██████▉   | 9315/13317 [1:36:02<40:54,  1.63it/s]
 70%|██████▉   | 9316/13317 [1:36:03<40:54,  1.63it/s]
                                                      
{'loss': 0.1273, 'grad_norm': 0.5538309812545776, 'learning_rate': 2.506813349609657e-06, 'epoch': 2.1}

 70%|██████▉   | 9316/13317 [1:36:03<40:54,  1.63it/s]
 70%|██████▉   | 9317/13317 [1:36:03<40:55,  1.63it/s]
                                                      
{'loss': 0.1521, 'grad_norm': 0.7200771570205688, 'learning_rate': 2.5056773626153303e-06, 'epoch': 2.1}

 70%|██████▉   | 9317/13317 [1:36:03<40:55,  1.63it/s]
 70%|██████▉   | 9318/13317 [1:36:04<40:55,  1.63it/s]
                                                      
{'loss': 0.162, 'grad_norm': 0.5675994753837585, 'learning_rate': 2.5045415470073776e-06, 'epoch': 2.1}

 70%|██████▉   | 9318/13317 [1:36:04<40:55,  1.63it/s]
 70%|██████▉   | 9319/13317 [1:36:04<40:56,  1.63it/s]
                                                      
{'loss': 0.163, 'grad_norm': 0.5325003266334534, 'learning_rate': 2.5034059028638445e-06, 'epoch': 2.1}

 70%|██████▉   | 9319/13317 [1:36:04<40:56,  1.63it/s]
 70%|██████▉   | 9320/13317 [1:36:05<40:55,  1.63it/s]
                                                      
{'loss': 0.1251, 'grad_norm': 0.42249733209609985, 'learning_rate': 2.5022704302627588e-06, 'epoch': 2.1}

 70%|██████▉   | 9320/13317 [1:36:05<40:55,  1.63it/s]
 70%|██████▉   | 9321/13317 [1:36:06<40:55,  1.63it/s]
                                                      
{'loss': 0.1893, 'grad_norm': 0.5851512551307678, 'learning_rate': 2.5011351292821424e-06, 'epoch': 2.1}

 70%|██████▉   | 9321/13317 [1:36:06<40:55,  1.63it/s]
 70%|███████   | 9322/13317 [1:36:06<40:53,  1.63it/s]
                                                      
{'loss': 0.2331, 'grad_norm': 0.7107666730880737, 'learning_rate': 2.5000000000000015e-06, 'epoch': 2.1}

 70%|███████   | 9322/13317 [1:36:06<40:53,  1.63it/s]
 70%|███████   | 9323/13317 [1:36:07<40:51,  1.63it/s]
                                                      
{'loss': 0.1325, 'grad_norm': 0.4619589149951935, 'learning_rate': 2.4988650424943284e-06, 'epoch': 2.1}

 70%|███████   | 9323/13317 [1:36:07<40:51,  1.63it/s]
 70%|███████   | 9324/13317 [1:36:07<40:50,  1.63it/s]
                                                      
{'loss': 0.1295, 'grad_norm': 0.5284699201583862, 'learning_rate': 2.4977302568431115e-06, 'epoch': 2.1}

 70%|███████   | 9324/13317 [1:36:08<40:50,  1.63it/s]
 70%|███████   | 9325/13317 [1:36:08<40:49,  1.63it/s]
                                                      
{'loss': 0.1286, 'grad_norm': 0.4816587269306183, 'learning_rate': 2.496595643124321e-06, 'epoch': 2.1}

 70%|███████   | 9325/13317 [1:36:08<40:49,  1.63it/s]
 70%|███████   | 9326/13317 [1:36:09<40:48,  1.63it/s]
                                                      
{'loss': 0.0989, 'grad_norm': 0.4052530527114868, 'learning_rate': 2.495461201415914e-06, 'epoch': 2.1}

 70%|███████   | 9326/13317 [1:36:09<40:48,  1.63it/s]
 70%|███████   | 9327/13317 [1:36:09<40:47,  1.63it/s]
                                                      
{'loss': 0.1952, 'grad_norm': 0.6850550174713135, 'learning_rate': 2.4943269317958408e-06, 'epoch': 2.1}

 70%|███████   | 9327/13317 [1:36:09<40:47,  1.63it/s]
 70%|███████   | 9328/13317 [1:36:10<40:47,  1.63it/s]
                                                      
{'loss': 0.1441, 'grad_norm': 0.5366700291633606, 'learning_rate': 2.4931928343420397e-06, 'epoch': 2.1}

 70%|███████   | 9328/13317 [1:36:10<40:47,  1.63it/s]
 70%|███████   | 9329/13317 [1:36:11<40:47,  1.63it/s]
                                                      
{'loss': 0.1261, 'grad_norm': 0.4812491536140442, 'learning_rate': 2.4920589091324338e-06, 'epoch': 2.1}

 70%|███████   | 9329/13317 [1:36:11<40:47,  1.63it/s]
 70%|███████   | 9330/13317 [1:36:11<40:45,  1.63it/s]
                                                      
{'loss': 0.1549, 'grad_norm': 0.5749524235725403, 'learning_rate': 2.490925156244933e-06, 'epoch': 2.1}

 70%|███████   | 9330/13317 [1:36:11<40:45,  1.63it/s]
 70%|███████   | 9331/13317 [1:36:12<40:45,  1.63it/s]
                                                      
{'loss': 0.152, 'grad_norm': 0.5490036010742188, 'learning_rate': 2.4897915757574426e-06, 'epoch': 2.1}

 70%|███████   | 9331/13317 [1:36:12<40:45,  1.63it/s]
 70%|███████   | 9332/13317 [1:36:12<40:45,  1.63it/s]
                                                      
{'loss': 0.1594, 'grad_norm': 1.0090062618255615, 'learning_rate': 2.48865816774785e-06, 'epoch': 2.1}

 70%|███████   | 9332/13317 [1:36:12<40:45,  1.63it/s]
 70%|███████   | 9333/13317 [1:36:13<40:44,  1.63it/s]
                                                      
{'loss': 0.1572, 'grad_norm': 0.5489905476570129, 'learning_rate': 2.4875249322940294e-06, 'epoch': 2.1}

 70%|███████   | 9333/13317 [1:36:13<40:44,  1.63it/s]
 70%|███████   | 9334/13317 [1:36:14<40:45,  1.63it/s]
                                                      
{'loss': 0.2343, 'grad_norm': 0.6772655844688416, 'learning_rate': 2.486391869473848e-06, 'epoch': 2.1}

 70%|███████   | 9334/13317 [1:36:14<40:45,  1.63it/s]
 70%|███████   | 9335/13317 [1:36:14<40:44,  1.63it/s]
                                                      
{'loss': 0.1465, 'grad_norm': 0.552223801612854, 'learning_rate': 2.4852589793651636e-06, 'epoch': 2.1}

 70%|███████   | 9335/13317 [1:36:14<40:44,  1.63it/s]
 70%|███████   | 9336/13317 [1:36:15<40:42,  1.63it/s]
                                                      
{'loss': 0.1523, 'grad_norm': 0.5271050930023193, 'learning_rate': 2.4841262620458108e-06, 'epoch': 2.1}

 70%|███████   | 9336/13317 [1:36:15<40:42,  1.63it/s]
 70%|███████   | 9337/13317 [1:36:15<40:41,  1.63it/s]
                                                      
{'loss': 0.1973, 'grad_norm': 0.6012085676193237, 'learning_rate': 2.482993717593621e-06, 'epoch': 2.1}

 70%|███████   | 9337/13317 [1:36:16<40:41,  1.63it/s]
 70%|███████   | 9338/13317 [1:36:16<40:40,  1.63it/s]
                                                      
{'loss': 0.14, 'grad_norm': 0.6105419397354126, 'learning_rate': 2.4818613460864165e-06, 'epoch': 2.1}

 70%|███████   | 9338/13317 [1:36:16<40:40,  1.63it/s]
 70%|███████   | 9339/13317 [1:36:17<40:40,  1.63it/s]
                                                      
{'loss': 0.1499, 'grad_norm': 0.55092453956604, 'learning_rate': 2.4807291476019996e-06, 'epoch': 2.1}

 70%|███████   | 9339/13317 [1:36:17<40:40,  1.63it/s]
 70%|███████   | 9340/13317 [1:36:17<40:39,  1.63it/s]
                                                      
{'loss': 0.1297, 'grad_norm': 0.5248216390609741, 'learning_rate': 2.479597122218163e-06, 'epoch': 2.1}

 70%|███████   | 9340/13317 [1:36:17<40:39,  1.63it/s]
 70%|███████   | 9341/13317 [1:36:18<40:37,  1.63it/s]
                                                      
{'loss': 0.1381, 'grad_norm': 0.48298579454421997, 'learning_rate': 2.47846527001269e-06, 'epoch': 2.1}

 70%|███████   | 9341/13317 [1:36:18<40:37,  1.63it/s]
 70%|███████   | 9342/13317 [1:36:18<40:37,  1.63it/s]
                                                      
{'loss': 0.189, 'grad_norm': 0.5677077770233154, 'learning_rate': 2.477333591063354e-06, 'epoch': 2.1}

 70%|███████   | 9342/13317 [1:36:19<40:37,  1.63it/s]
 70%|███████   | 9343/13317 [1:36:19<40:36,  1.63it/s]
                                                      
{'loss': 0.1367, 'grad_norm': 0.5157226324081421, 'learning_rate': 2.47620208544791e-06, 'epoch': 2.1}

 70%|███████   | 9343/13317 [1:36:19<40:36,  1.63it/s]
 70%|███████   | 9344/13317 [1:36:20<40:36,  1.63it/s]
                                                      
{'loss': 0.1191, 'grad_norm': 0.43935874104499817, 'learning_rate': 2.475070753244104e-06, 'epoch': 2.1}

 70%|███████   | 9344/13317 [1:36:20<40:36,  1.63it/s]
 70%|███████   | 9345/13317 [1:36:20<40:36,  1.63it/s]
                                                      
{'loss': 0.1446, 'grad_norm': 0.5413407683372498, 'learning_rate': 2.473939594529671e-06, 'epoch': 2.11}

 70%|███████   | 9345/13317 [1:36:20<40:36,  1.63it/s]
 70%|███████   | 9346/13317 [1:36:21<40:36,  1.63it/s]
                                                      
{'loss': 0.1317, 'grad_norm': 0.5318446755409241, 'learning_rate': 2.4728086093823376e-06, 'epoch': 2.11}

 70%|███████   | 9346/13317 [1:36:21<40:36,  1.63it/s]
 70%|███████   | 9347/13317 [1:36:22<40:37,  1.63it/s]
                                                      
{'loss': 0.1619, 'grad_norm': 0.5277021527290344, 'learning_rate': 2.471677797879808e-06, 'epoch': 2.11}

 70%|███████   | 9347/13317 [1:36:22<40:37,  1.63it/s]
 70%|███████   | 9348/13317 [1:36:22<40:34,  1.63it/s]
                                                      
{'loss': 0.1539, 'grad_norm': 0.560069739818573, 'learning_rate': 2.4705471600997837e-06, 'epoch': 2.11}

 70%|███████   | 9348/13317 [1:36:22<40:34,  1.63it/s]
 70%|███████   | 9349/13317 [1:36:23<40:34,  1.63it/s]
                                                      
{'loss': 0.1482, 'grad_norm': 0.5002989768981934, 'learning_rate': 2.469416696119954e-06, 'epoch': 2.11}

 70%|███████   | 9349/13317 [1:36:23<40:34,  1.63it/s]
 70%|███████   | 9350/13317 [1:36:23<40:34,  1.63it/s]
                                                      
{'loss': 0.1292, 'grad_norm': 0.5602324604988098, 'learning_rate': 2.468286406017991e-06, 'epoch': 2.11}

 70%|███████   | 9350/13317 [1:36:24<40:34,  1.63it/s]
 70%|███████   | 9351/13317 [1:36:24<40:32,  1.63it/s]
                                                      
{'loss': 0.1555, 'grad_norm': 0.5822707414627075, 'learning_rate': 2.4671562898715567e-06, 'epoch': 2.11}

 70%|███████   | 9351/13317 [1:36:24<40:32,  1.63it/s]
 70%|███████   | 9352/13317 [1:36:25<40:32,  1.63it/s]
                                                      
{'loss': 0.1656, 'grad_norm': 0.5812645554542542, 'learning_rate': 2.466026347758305e-06, 'epoch': 2.11}

 70%|███████   | 9352/13317 [1:36:25<40:32,  1.63it/s]
 70%|███████   | 9353/13317 [1:36:25<40:32,  1.63it/s]
                                                      
{'loss': 0.1646, 'grad_norm': 0.5407654643058777, 'learning_rate': 2.464896579755874e-06, 'epoch': 2.11}

 70%|███████   | 9353/13317 [1:36:25<40:32,  1.63it/s]
 70%|███████   | 9354/13317 [1:36:26<40:30,  1.63it/s]
                                                      
{'loss': 0.2179, 'grad_norm': 0.7873445153236389, 'learning_rate': 2.463766985941888e-06, 'epoch': 2.11}

 70%|███████   | 9354/13317 [1:36:26<40:30,  1.63it/s]
 70%|███████   | 9355/13317 [1:36:26<40:30,  1.63it/s]
                                                      
{'loss': 0.1855, 'grad_norm': 0.5580340027809143, 'learning_rate': 2.4626375663939635e-06, 'epoch': 2.11}

 70%|███████   | 9355/13317 [1:36:27<40:30,  1.63it/s]
 70%|███████   | 9356/13317 [1:36:27<40:33,  1.63it/s]
                                                      
{'loss': 0.1516, 'grad_norm': 0.5191629528999329, 'learning_rate': 2.4615083211897062e-06, 'epoch': 2.11}

 70%|███████   | 9356/13317 [1:36:27<40:33,  1.63it/s]
 70%|███████   | 9357/13317 [1:36:28<40:31,  1.63it/s]
                                                      
{'loss': 0.1477, 'grad_norm': 0.5528790354728699, 'learning_rate': 2.460379250406706e-06, 'epoch': 2.11}

 70%|███████   | 9357/13317 [1:36:28<40:31,  1.63it/s]
 70%|███████   | 9358/13317 [1:36:28<40:29,  1.63it/s]
                                                      
{'loss': 0.1367, 'grad_norm': 0.5094370245933533, 'learning_rate': 2.4592503541225386e-06, 'epoch': 2.11}

 70%|███████   | 9358/13317 [1:36:28<40:29,  1.63it/s]
 70%|███████   | 9359/13317 [1:36:29<40:34,  1.63it/s]
                                                      
{'loss': 0.1772, 'grad_norm': 0.6211641430854797, 'learning_rate': 2.4581216324147763e-06, 'epoch': 2.11}

 70%|███████   | 9359/13317 [1:36:29<40:34,  1.63it/s]
 70%|███████   | 9360/13317 [1:36:30<40:29,  1.63it/s]
                                                      
{'loss': 0.1523, 'grad_norm': 0.5240804553031921, 'learning_rate': 2.456993085360971e-06, 'epoch': 2.11}

 70%|███████   | 9360/13317 [1:36:30<40:29,  1.63it/s]
 70%|███████   | 9361/13317 [1:36:30<40:29,  1.63it/s]
                                                      
{'loss': 0.1974, 'grad_norm': 0.6805075407028198, 'learning_rate': 2.455864713038665e-06, 'epoch': 2.11}

 70%|███████   | 9361/13317 [1:36:30<40:29,  1.63it/s]
 70%|███████   | 9362/13317 [1:36:31<40:28,  1.63it/s]
                                                      
{'loss': 0.1882, 'grad_norm': 0.521658718585968, 'learning_rate': 2.4547365155253904e-06, 'epoch': 2.11}

 70%|███████   | 9362/13317 [1:36:31<40:28,  1.63it/s]
 70%|███████   | 9363/13317 [1:36:31<40:26,  1.63it/s]
                                                      
{'loss': 0.1423, 'grad_norm': 0.5205925703048706, 'learning_rate': 2.4536084928986693e-06, 'epoch': 2.11}

 70%|███████   | 9363/13317 [1:36:31<40:26,  1.63it/s]
 70%|███████   | 9364/13317 [1:36:32<40:25,  1.63it/s]
                                                      
{'loss': 0.14, 'grad_norm': 0.5382764339447021, 'learning_rate': 2.452480645236006e-06, 'epoch': 2.11}

 70%|███████   | 9364/13317 [1:36:32<40:25,  1.63it/s]
 70%|███████   | 9365/13317 [1:36:33<40:24,  1.63it/s]
                                                      
{'loss': 0.1246, 'grad_norm': 0.46633628010749817, 'learning_rate': 2.4513529726148947e-06, 'epoch': 2.11}

 70%|███████   | 9365/13317 [1:36:33<40:24,  1.63it/s]
 70%|███████   | 9366/13317 [1:36:33<40:23,  1.63it/s]
                                                      
{'loss': 0.124, 'grad_norm': 0.4156206548213959, 'learning_rate': 2.4502254751128214e-06, 'epoch': 2.11}

 70%|███████   | 9366/13317 [1:36:33<40:23,  1.63it/s]
 70%|███████   | 9367/13317 [1:36:34<40:23,  1.63it/s]
                                                      
{'loss': 0.1675, 'grad_norm': 0.588591456413269, 'learning_rate': 2.4490981528072534e-06, 'epoch': 2.11}

 70%|███████   | 9367/13317 [1:36:34<40:23,  1.63it/s]
 70%|███████   | 9368/13317 [1:36:34<40:23,  1.63it/s]
                                                      
{'loss': 0.1421, 'grad_norm': 0.5463805198669434, 'learning_rate': 2.4479710057756533e-06, 'epoch': 2.11}

 70%|███████   | 9368/13317 [1:36:35<40:23,  1.63it/s]
 70%|███████   | 9369/13317 [1:36:35<40:22,  1.63it/s]
                                                      
{'loss': 0.1413, 'grad_norm': 0.4755219519138336, 'learning_rate': 2.4468440340954664e-06, 'epoch': 2.11}

 70%|███████   | 9369/13317 [1:36:35<40:22,  1.63it/s]
 70%|███████   | 9370/13317 [1:36:36<40:21,  1.63it/s]
                                                      
{'loss': 0.1186, 'grad_norm': 0.458421528339386, 'learning_rate': 2.4457172378441262e-06, 'epoch': 2.11}

 70%|███████   | 9370/13317 [1:36:36<40:21,  1.63it/s]
 70%|███████   | 9371/13317 [1:36:36<40:23,  1.63it/s]
                                                      
{'loss': 0.1343, 'grad_norm': 0.4789389371871948, 'learning_rate': 2.4445906170990584e-06, 'epoch': 2.11}

 70%|███████   | 9371/13317 [1:36:36<40:23,  1.63it/s]
 70%|███████   | 9372/13317 [1:36:37<40:19,  1.63it/s]
                                                      
{'loss': 0.1196, 'grad_norm': 0.3983357846736908, 'learning_rate': 2.4434641719376707e-06, 'epoch': 2.11}

 70%|███████   | 9372/13317 [1:36:37<40:19,  1.63it/s]
 70%|███████   | 9373/13317 [1:36:38<40:19,  1.63it/s]
                                                      
{'loss': 0.1477, 'grad_norm': 0.5257238149642944, 'learning_rate': 2.442337902437365e-06, 'epoch': 2.11}

 70%|███████   | 9373/13317 [1:36:38<40:19,  1.63it/s]
 70%|███████   | 9374/13317 [1:36:38<40:19,  1.63it/s]
                                                      
{'loss': 0.1649, 'grad_norm': 0.6176623702049255, 'learning_rate': 2.441211808675525e-06, 'epoch': 2.11}

 70%|███████   | 9374/13317 [1:36:38<40:19,  1.63it/s]
 70%|███████   | 9375/13317 [1:36:39<40:18,  1.63it/s]
                                                      
{'loss': 0.1452, 'grad_norm': 0.492859810590744, 'learning_rate': 2.440085890729528e-06, 'epoch': 2.11}

 70%|███████   | 9375/13317 [1:36:39<40:18,  1.63it/s]
 70%|███████   | 9376/13317 [1:36:39<40:18,  1.63it/s]
                                                      
{'loss': 0.2126, 'grad_norm': 0.7293538451194763, 'learning_rate': 2.438960148676735e-06, 'epoch': 2.11}

 70%|███████   | 9376/13317 [1:36:39<40:18,  1.63it/s]
 70%|███████   | 9377/13317 [1:36:40<40:17,  1.63it/s]
                                                      
{'loss': 0.1336, 'grad_norm': 0.5033314824104309, 'learning_rate': 2.437834582594495e-06, 'epoch': 2.11}

 70%|███████   | 9377/13317 [1:36:40<40:17,  1.63it/s]
 70%|███████   | 9378/13317 [1:36:41<40:16,  1.63it/s]
                                                      
{'loss': 0.1296, 'grad_norm': 0.4595862627029419, 'learning_rate': 2.43670919256015e-06, 'epoch': 2.11}

 70%|███████   | 9378/13317 [1:36:41<40:16,  1.63it/s]
 70%|███████   | 9379/13317 [1:36:41<40:14,  1.63it/s]
                                                      
{'loss': 0.1237, 'grad_norm': 0.5365263223648071, 'learning_rate': 2.4355839786510216e-06, 'epoch': 2.11}

 70%|███████   | 9379/13317 [1:36:41<40:14,  1.63it/s]
 70%|███████   | 9380/13317 [1:36:42<40:14,  1.63it/s]
                                                      
{'loss': 0.1644, 'grad_norm': 0.6194126009941101, 'learning_rate': 2.434458940944428e-06, 'epoch': 2.11}

 70%|███████   | 9380/13317 [1:36:42<40:14,  1.63it/s]
 70%|███████   | 9381/13317 [1:36:42<40:14,  1.63it/s]
                                                      
{'loss': 0.1418, 'grad_norm': 0.5479564666748047, 'learning_rate': 2.4333340795176685e-06, 'epoch': 2.11}

 70%|███████   | 9381/13317 [1:36:43<40:14,  1.63it/s]
 70%|███████   | 9382/13317 [1:36:43<40:13,  1.63it/s]
                                                      
{'loss': 0.166, 'grad_norm': 0.5541086792945862, 'learning_rate': 2.4322093944480353e-06, 'epoch': 2.11}

 70%|███████   | 9382/13317 [1:36:43<40:13,  1.63it/s]
 70%|███████   | 9383/13317 [1:36:44<40:13,  1.63it/s]
                                                      
{'loss': 0.1445, 'grad_norm': 0.5327193737030029, 'learning_rate': 2.431084885812805e-06, 'epoch': 2.11}

 70%|███████   | 9383/13317 [1:36:44<40:13,  1.63it/s]
 70%|███████   | 9384/13317 [1:36:44<40:12,  1.63it/s]
                                                      
{'loss': 0.1562, 'grad_norm': 0.6170597672462463, 'learning_rate': 2.4299605536892413e-06, 'epoch': 2.11}

 70%|███████   | 9384/13317 [1:36:44<40:12,  1.63it/s]
 70%|███████   | 9385/13317 [1:36:45<40:11,  1.63it/s]
                                                      
{'loss': 0.1877, 'grad_norm': 0.6235355734825134, 'learning_rate': 2.4288363981546013e-06, 'epoch': 2.11}

 70%|███████   | 9385/13317 [1:36:45<40:11,  1.63it/s]
 70%|███████   | 9386/13317 [1:36:45<40:10,  1.63it/s]
                                                      
{'loss': 0.1657, 'grad_norm': 0.6115406155586243, 'learning_rate': 2.427712419286125e-06, 'epoch': 2.11}

 70%|███████   | 9386/13317 [1:36:46<40:10,  1.63it/s]
 70%|███████   | 9387/13317 [1:36:46<40:11,  1.63it/s]
                                                      
{'loss': 0.1504, 'grad_norm': 0.6644530296325684, 'learning_rate': 2.426588617161039e-06, 'epoch': 2.11}

 70%|███████   | 9387/13317 [1:36:46<40:11,  1.63it/s]
 70%|███████   | 9388/13317 [1:36:47<40:09,  1.63it/s]
                                                      
{'loss': 0.1392, 'grad_norm': 0.602378249168396, 'learning_rate': 2.425464991856564e-06, 'epoch': 2.11}

 70%|███████   | 9388/13317 [1:36:47<40:09,  1.63it/s]
 71%|███████   | 9389/13317 [1:36:47<40:09,  1.63it/s]
                                                      
{'loss': 0.1232, 'grad_norm': 0.4735495448112488, 'learning_rate': 2.424341543449904e-06, 'epoch': 2.12}

 71%|███████   | 9389/13317 [1:36:47<40:09,  1.63it/s]
 71%|███████   | 9390/13317 [1:36:48<40:09,  1.63it/s]
                                                      
{'loss': 0.131, 'grad_norm': 0.5362032651901245, 'learning_rate': 2.4232182720182524e-06, 'epoch': 2.12}

 71%|███████   | 9390/13317 [1:36:48<40:09,  1.63it/s]
 71%|███████   | 9391/13317 [1:36:49<40:07,  1.63it/s]
                                                      
{'loss': 0.1602, 'grad_norm': 0.6042806506156921, 'learning_rate': 2.4220951776387863e-06, 'epoch': 2.12}

 71%|███████   | 9391/13317 [1:36:49<40:07,  1.63it/s]
 71%|███████   | 9392/13317 [1:36:49<40:08,  1.63it/s]
                                                      
{'loss': 0.1475, 'grad_norm': 0.5577574372291565, 'learning_rate': 2.4209722603886792e-06, 'epoch': 2.12}

 71%|███████   | 9392/13317 [1:36:49<40:08,  1.63it/s]
 71%|███████   | 9393/13317 [1:36:50<40:07,  1.63it/s]
                                                      
{'loss': 0.1378, 'grad_norm': 0.4984179437160492, 'learning_rate': 2.4198495203450844e-06, 'epoch': 2.12}

 71%|███████   | 9393/13317 [1:36:50<40:07,  1.63it/s]
 71%|███████   | 9394/13317 [1:36:50<40:06,  1.63it/s]
                                                      
{'loss': 0.1383, 'grad_norm': 0.49477311968803406, 'learning_rate': 2.4187269575851457e-06, 'epoch': 2.12}

 71%|███████   | 9394/13317 [1:36:51<40:06,  1.63it/s]
 71%|███████   | 9395/13317 [1:36:51<40:06,  1.63it/s]
                                                      
{'loss': 0.1397, 'grad_norm': 0.4695683717727661, 'learning_rate': 2.4176045721859953e-06, 'epoch': 2.12}

 71%|███████   | 9395/13317 [1:36:51<40:06,  1.63it/s]
 71%|███████   | 9396/13317 [1:36:52<40:06,  1.63it/s]
                                                      
{'loss': 0.1123, 'grad_norm': 0.4435540437698364, 'learning_rate': 2.416482364224755e-06, 'epoch': 2.12}

 71%|███████   | 9396/13317 [1:36:52<40:06,  1.63it/s]
 71%|███████   | 9397/13317 [1:36:52<40:05,  1.63it/s]
                                                      
{'loss': 0.175, 'grad_norm': 0.5606993436813354, 'learning_rate': 2.4153603337785323e-06, 'epoch': 2.12}

 71%|███████   | 9397/13317 [1:36:52<40:05,  1.63it/s]
 71%|███████   | 9398/13317 [1:36:53<40:05,  1.63it/s]
                                                      
{'loss': 0.1395, 'grad_norm': 0.5009104609489441, 'learning_rate': 2.4142384809244187e-06, 'epoch': 2.12}

 71%|███████   | 9398/13317 [1:36:53<40:05,  1.63it/s]
 71%|███████   | 9399/13317 [1:36:53<40:03,  1.63it/s]
                                                      
{'loss': 0.1584, 'grad_norm': 0.5509693622589111, 'learning_rate': 2.4131168057395e-06, 'epoch': 2.12}

 71%|███████   | 9399/13317 [1:36:54<40:03,  1.63it/s]
 71%|███████   | 9400/13317 [1:36:54<40:03,  1.63it/s]
                                                      
{'loss': 0.1374, 'grad_norm': 0.5549517869949341, 'learning_rate': 2.4119953083008507e-06, 'epoch': 2.12}

 71%|███████   | 9400/13317 [1:36:54<40:03,  1.63it/s]
 71%|███████   | 9401/13317 [1:36:55<40:02,  1.63it/s]
                                                      
{'loss': 0.1336, 'grad_norm': 0.4995281994342804, 'learning_rate': 2.4108739886855226e-06, 'epoch': 2.12}

 71%|███████   | 9401/13317 [1:36:55<40:02,  1.63it/s]
 71%|███████   | 9402/13317 [1:36:55<40:01,  1.63it/s]
                                                      
{'loss': 0.1319, 'grad_norm': 0.49877384305000305, 'learning_rate': 2.409752846970565e-06, 'epoch': 2.12}

 71%|███████   | 9402/13317 [1:36:55<40:01,  1.63it/s]
 71%|███████   | 9403/13317 [1:36:56<40:01,  1.63it/s]
                                                      
{'loss': 0.1663, 'grad_norm': 0.6157169938087463, 'learning_rate': 2.408631883233015e-06, 'epoch': 2.12}

 71%|███████   | 9403/13317 [1:36:56<40:01,  1.63it/s]
 71%|███████   | 9404/13317 [1:36:57<40:01,  1.63it/s]
                                                      
{'loss': 0.1199, 'grad_norm': 0.4629165828227997, 'learning_rate': 2.4075110975498927e-06, 'epoch': 2.12}

 71%|███████   | 9404/13317 [1:36:57<40:01,  1.63it/s]
 71%|███████   | 9405/13317 [1:36:57<40:00,  1.63it/s]
                                                      
{'loss': 0.1545, 'grad_norm': 0.5788499116897583, 'learning_rate': 2.4063904899982055e-06, 'epoch': 2.12}

 71%|███████   | 9405/13317 [1:36:57<40:00,  1.63it/s]
 71%|███████   | 9406/13317 [1:36:58<40:01,  1.63it/s]
                                                      
{'loss': 0.1589, 'grad_norm': 0.5814086198806763, 'learning_rate': 2.4052700606549533e-06, 'epoch': 2.12}

 71%|███████   | 9406/13317 [1:36:58<40:01,  1.63it/s]
 71%|███████   | 9407/13317 [1:36:58<40:00,  1.63it/s]
                                                      
{'loss': 0.1391, 'grad_norm': 0.4441385567188263, 'learning_rate': 2.4041498095971253e-06, 'epoch': 2.12}

 71%|███████   | 9407/13317 [1:36:58<40:00,  1.63it/s]
 71%|███████   | 9408/13317 [1:36:59<39:59,  1.63it/s]
                                                      
{'loss': 0.1406, 'grad_norm': 0.4937634766101837, 'learning_rate': 2.403029736901687e-06, 'epoch': 2.12}

 71%|███████   | 9408/13317 [1:36:59<39:59,  1.63it/s]
 71%|███████   | 9409/13317 [1:37:00<39:57,  1.63it/s]
                                                      
{'loss': 0.1554, 'grad_norm': 0.5437819361686707, 'learning_rate': 2.4019098426456023e-06, 'epoch': 2.12}

 71%|███████   | 9409/13317 [1:37:00<39:57,  1.63it/s]
 71%|███████   | 9410/13317 [1:37:00<39:57,  1.63it/s]
                                                      
{'loss': 0.1489, 'grad_norm': 0.5033108592033386, 'learning_rate': 2.4007901269058233e-06, 'epoch': 2.12}

 71%|███████   | 9410/13317 [1:37:00<39:57,  1.63it/s]
 71%|███████   | 9411/13317 [1:37:01<39:56,  1.63it/s]
                                                      
{'loss': 0.1679, 'grad_norm': 0.567689061164856, 'learning_rate': 2.3996705897592825e-06, 'epoch': 2.12}

 71%|███████   | 9411/13317 [1:37:01<39:56,  1.63it/s]
 71%|███████   | 9412/13317 [1:37:01<39:56,  1.63it/s]
                                                      
{'loss': 0.1622, 'grad_norm': 0.5837022662162781, 'learning_rate': 2.3985512312829035e-06, 'epoch': 2.12}

 71%|███████   | 9412/13317 [1:37:02<39:56,  1.63it/s]
 71%|███████   | 9413/13317 [1:37:02<39:56,  1.63it/s]
                                                      
{'loss': 0.1404, 'grad_norm': 0.5516318678855896, 'learning_rate': 2.397432051553601e-06, 'epoch': 2.12}

 71%|███████   | 9413/13317 [1:37:02<39:56,  1.63it/s]
 71%|███████   | 9414/13317 [1:37:03<39:55,  1.63it/s]
                                                      
{'loss': 0.1814, 'grad_norm': 0.6119487881660461, 'learning_rate': 2.3963130506482736e-06, 'epoch': 2.12}

 71%|███████   | 9414/13317 [1:37:03<39:55,  1.63it/s]
 71%|███████   | 9415/13317 [1:37:03<39:55,  1.63it/s]
                                                      
{'loss': 0.0966, 'grad_norm': 0.41101500391960144, 'learning_rate': 2.395194228643806e-06, 'epoch': 2.12}

 71%|███████   | 9415/13317 [1:37:03<39:55,  1.63it/s]
 71%|███████   | 9416/13317 [1:37:04<39:53,  1.63it/s]
                                                      
{'loss': 0.1477, 'grad_norm': 0.6061045527458191, 'learning_rate': 2.3940755856170744e-06, 'epoch': 2.12}

 71%|███████   | 9416/13317 [1:37:04<39:53,  1.63it/s]
 71%|███████   | 9417/13317 [1:37:04<39:53,  1.63it/s]
                                                      
{'loss': 0.1306, 'grad_norm': 0.5326359868049622, 'learning_rate': 2.3929571216449444e-06, 'epoch': 2.12}

 71%|███████   | 9417/13317 [1:37:05<39:53,  1.63it/s]
 71%|███████   | 9418/13317 [1:37:05<39:51,  1.63it/s]
                                                      
{'loss': 0.1577, 'grad_norm': 0.7155811190605164, 'learning_rate': 2.3918388368042633e-06, 'epoch': 2.12}

 71%|███████   | 9418/13317 [1:37:05<39:51,  1.63it/s]
 71%|███████   | 9419/13317 [1:37:06<39:50,  1.63it/s]
                                                      
{'loss': 0.1295, 'grad_norm': 0.4775228500366211, 'learning_rate': 2.3907207311718682e-06, 'epoch': 2.12}

 71%|███████   | 9419/13317 [1:37:06<39:50,  1.63it/s]
 71%|███████   | 9420/13317 [1:37:06<39:50,  1.63it/s]
                                                      
{'loss': 0.1307, 'grad_norm': 0.5219584107398987, 'learning_rate': 2.389602804824589e-06, 'epoch': 2.12}

 71%|███████   | 9420/13317 [1:37:06<39:50,  1.63it/s]
 71%|███████   | 9421/13317 [1:37:07<39:49,  1.63it/s]
                                                      
{'loss': 0.1531, 'grad_norm': 0.630628228187561, 'learning_rate': 2.388485057839234e-06, 'epoch': 2.12}

 71%|███████   | 9421/13317 [1:37:07<39:49,  1.63it/s]
 71%|███████   | 9422/13317 [1:37:08<39:50,  1.63it/s]
                                                      
{'loss': 0.1563, 'grad_norm': 0.5660696625709534, 'learning_rate': 2.387367490292608e-06, 'epoch': 2.12}

 71%|███████   | 9422/13317 [1:37:08<39:50,  1.63it/s]
 71%|███████   | 9423/13317 [1:37:08<39:48,  1.63it/s]
                                                      
{'loss': 0.1388, 'grad_norm': 0.48132020235061646, 'learning_rate': 2.386250102261497e-06, 'epoch': 2.12}

 71%|███████   | 9423/13317 [1:37:08<39:48,  1.63it/s]
 71%|███████   | 9424/13317 [1:37:09<39:49,  1.63it/s]
                                                      
{'loss': 0.1338, 'grad_norm': 0.5056582093238831, 'learning_rate': 2.3851328938226808e-06, 'epoch': 2.12}

 71%|███████   | 9424/13317 [1:37:09<39:49,  1.63it/s]
 71%|███████   | 9425/13317 [1:37:09<39:48,  1.63it/s]
                                                      
{'loss': 0.1443, 'grad_norm': 0.5367896556854248, 'learning_rate': 2.3840158650529216e-06, 'epoch': 2.12}

 71%|███████   | 9425/13317 [1:37:10<39:48,  1.63it/s]
 71%|███████   | 9426/13317 [1:37:10<39:47,  1.63it/s]
                                                      
{'loss': 0.1226, 'grad_norm': 0.5124879479408264, 'learning_rate': 2.382899016028969e-06, 'epoch': 2.12}

 71%|███████   | 9426/13317 [1:37:10<39:47,  1.63it/s]
 71%|███████   | 9427/13317 [1:37:11<39:46,  1.63it/s]
                                                      
{'loss': 0.1318, 'grad_norm': 0.5916809439659119, 'learning_rate': 2.381782346827566e-06, 'epoch': 2.12}

 71%|███████   | 9427/13317 [1:37:11<39:46,  1.63it/s]
 71%|███████   | 9428/13317 [1:37:11<39:46,  1.63it/s]
                                                      
{'loss': 0.1295, 'grad_norm': 0.5043729543685913, 'learning_rate': 2.380665857525436e-06, 'epoch': 2.12}

 71%|███████   | 9428/13317 [1:37:11<39:46,  1.63it/s]
 71%|███████   | 9429/13317 [1:37:12<39:45,  1.63it/s]
                                                      
{'loss': 0.1424, 'grad_norm': 0.5502538084983826, 'learning_rate': 2.3795495481992975e-06, 'epoch': 2.12}

 71%|███████   | 9429/13317 [1:37:12<39:45,  1.63it/s]
 71%|███████   | 9430/13317 [1:37:12<39:44,  1.63it/s]
                                                      
{'loss': 0.1472, 'grad_norm': 0.5621833801269531, 'learning_rate': 2.3784334189258505e-06, 'epoch': 2.12}

 71%|███████   | 9430/13317 [1:37:13<39:44,  1.63it/s]
 71%|███████   | 9431/13317 [1:37:13<39:44,  1.63it/s]
                                                      
{'loss': 0.1621, 'grad_norm': 0.6276944279670715, 'learning_rate': 2.377317469781784e-06, 'epoch': 2.12}

 71%|███████   | 9431/13317 [1:37:13<39:44,  1.63it/s]
 71%|███████   | 9432/13317 [1:37:14<39:43,  1.63it/s]
                                                      
{'loss': 0.1531, 'grad_norm': 0.5521619319915771, 'learning_rate': 2.3762017008437783e-06, 'epoch': 2.12}

 71%|███████   | 9432/13317 [1:37:14<39:43,  1.63it/s]
 71%|███████   | 9433/13317 [1:37:14<39:42,  1.63it/s]
                                                      
{'loss': 0.1363, 'grad_norm': 0.49627310037612915, 'learning_rate': 2.375086112188495e-06, 'epoch': 2.13}

 71%|███████   | 9433/13317 [1:37:14<39:42,  1.63it/s]
 71%|███████   | 9434/13317 [1:37:15<39:42,  1.63it/s]
                                                      
{'loss': 0.1595, 'grad_norm': 0.53859943151474, 'learning_rate': 2.373970703892591e-06, 'epoch': 2.13}

 71%|███████   | 9434/13317 [1:37:15<39:42,  1.63it/s]
 71%|███████   | 9435/13317 [1:37:16<39:41,  1.63it/s]
                                                      
{'loss': 0.1404, 'grad_norm': 0.5898581743240356, 'learning_rate': 2.3728554760327023e-06, 'epoch': 2.13}

 71%|███████   | 9435/13317 [1:37:16<39:41,  1.63it/s]
 71%|███████   | 9436/13317 [1:37:16<39:40,  1.63it/s]
                                                      
{'loss': 0.1688, 'grad_norm': 0.6235234141349792, 'learning_rate': 2.371740428685461e-06, 'epoch': 2.13}

 71%|███████   | 9436/13317 [1:37:16<39:40,  1.63it/s]
 71%|███████   | 9437/13317 [1:37:17<39:40,  1.63it/s]
                                                      
{'loss': 0.1496, 'grad_norm': 0.5801730751991272, 'learning_rate': 2.3706255619274804e-06, 'epoch': 2.13}

 71%|███████   | 9437/13317 [1:37:17<39:40,  1.63it/s]
 71%|███████   | 9438/13317 [1:37:17<39:40,  1.63it/s]
                                                      
{'loss': 0.1349, 'grad_norm': 0.5208654999732971, 'learning_rate': 2.3695108758353626e-06, 'epoch': 2.13}

 71%|███████   | 9438/13317 [1:37:18<39:40,  1.63it/s]
 71%|███████   | 9439/13317 [1:37:18<39:39,  1.63it/s]
                                                      
{'loss': 0.1483, 'grad_norm': 0.5116142630577087, 'learning_rate': 2.3683963704857016e-06, 'epoch': 2.13}

 71%|███████   | 9439/13317 [1:37:18<39:39,  1.63it/s]
 71%|███████   | 9440/13317 [1:37:19<39:39,  1.63it/s]
                                                      
{'loss': 0.1294, 'grad_norm': 0.4829523265361786, 'learning_rate': 2.3672820459550734e-06, 'epoch': 2.13}

 71%|███████   | 9440/13317 [1:37:19<39:39,  1.63it/s]
 71%|███████   | 9441/13317 [1:37:19<39:38,  1.63it/s]
                                                      
{'loss': 0.1533, 'grad_norm': 0.5657406449317932, 'learning_rate': 2.3661679023200422e-06, 'epoch': 2.13}

 71%|███████   | 9441/13317 [1:37:19<39:38,  1.63it/s]
 71%|███████   | 9442/13317 [1:37:20<39:37,  1.63it/s]
                                                      
{'loss': 0.1307, 'grad_norm': 0.575853168964386, 'learning_rate': 2.365053939657164e-06, 'epoch': 2.13}

 71%|███████   | 9442/13317 [1:37:20<39:37,  1.63it/s]
 71%|███████   | 9443/13317 [1:37:20<39:37,  1.63it/s]
                                                      
{'loss': 0.1607, 'grad_norm': 0.5846664905548096, 'learning_rate': 2.3639401580429815e-06, 'epoch': 2.13}

 71%|███████   | 9443/13317 [1:37:21<39:37,  1.63it/s]
 71%|███████   | 9444/13317 [1:37:21<39:36,  1.63it/s]
                                                      
{'loss': 0.2849, 'grad_norm': 0.8901411294937134, 'learning_rate': 2.362826557554021e-06, 'epoch': 2.13}

 71%|███████   | 9444/13317 [1:37:21<39:36,  1.63it/s]
 71%|███████   | 9445/13317 [1:37:22<39:35,  1.63it/s]
                                                      
{'loss': 0.1572, 'grad_norm': 0.5818642973899841, 'learning_rate': 2.361713138266797e-06, 'epoch': 2.13}

 71%|███████   | 9445/13317 [1:37:22<39:35,  1.63it/s]
 71%|███████   | 9446/13317 [1:37:22<39:35,  1.63it/s]
                                                      
{'loss': 0.1267, 'grad_norm': 0.4468002915382385, 'learning_rate': 2.360599900257815e-06, 'epoch': 2.13}

 71%|███████   | 9446/13317 [1:37:22<39:35,  1.63it/s]
 71%|███████   | 9447/13317 [1:37:23<39:35,  1.63it/s]
                                                      
{'loss': 0.1538, 'grad_norm': 0.5446751117706299, 'learning_rate': 2.35948684360357e-06, 'epoch': 2.13}

 71%|███████   | 9447/13317 [1:37:23<39:35,  1.63it/s]
 71%|███████   | 9448/13317 [1:37:24<39:36,  1.63it/s]
                                                      
{'loss': 0.1505, 'grad_norm': 0.5481831431388855, 'learning_rate': 2.358373968380533e-06, 'epoch': 2.13}

 71%|███████   | 9448/13317 [1:37:24<39:36,  1.63it/s]
 71%|███████   | 9449/13317 [1:37:24<39:39,  1.63it/s]
                                                      
{'loss': 0.1027, 'grad_norm': 0.4341077208518982, 'learning_rate': 2.3572612746651748e-06, 'epoch': 2.13}

 71%|███████   | 9449/13317 [1:37:24<39:39,  1.63it/s]
 71%|███████   | 9450/13317 [1:37:25<39:36,  1.63it/s]
                                                      
{'loss': 0.2102, 'grad_norm': 0.6699540615081787, 'learning_rate': 2.3561487625339503e-06, 'epoch': 2.13}

 71%|███████   | 9450/13317 [1:37:25<39:36,  1.63it/s]
 71%|███████   | 9451/13317 [1:37:25<39:34,  1.63it/s]
                                                      
{'loss': 0.1539, 'grad_norm': 0.6336162090301514, 'learning_rate': 2.3550364320632995e-06, 'epoch': 2.13}

 71%|███████   | 9451/13317 [1:37:25<39:34,  1.63it/s]
 71%|███████   | 9452/13317 [1:37:26<39:33,  1.63it/s]
                                                      
{'loss': 0.1408, 'grad_norm': 0.5749939680099487, 'learning_rate': 2.3539242833296493e-06, 'epoch': 2.13}

 71%|███████   | 9452/13317 [1:37:26<39:33,  1.63it/s]
 71%|███████   | 9453/13317 [1:37:27<39:32,  1.63it/s]
                                                      
{'loss': 0.1545, 'grad_norm': 0.6149581074714661, 'learning_rate': 2.352812316409418e-06, 'epoch': 2.13}

 71%|███████   | 9453/13317 [1:37:27<39:32,  1.63it/s]
 71%|███████   | 9454/13317 [1:37:27<39:32,  1.63it/s]
                                                      
{'loss': 0.1981, 'grad_norm': 0.6181749105453491, 'learning_rate': 2.351700531379013e-06, 'epoch': 2.13}

 71%|███████   | 9454/13317 [1:37:27<39:32,  1.63it/s]
 71%|███████   | 9455/13317 [1:37:28<39:32,  1.63it/s]
                                                      
{'loss': 0.1291, 'grad_norm': 0.5744105577468872, 'learning_rate': 2.3505889283148185e-06, 'epoch': 2.13}

 71%|███████   | 9455/13317 [1:37:28<39:32,  1.63it/s]
 71%|███████   | 9456/13317 [1:37:28<39:30,  1.63it/s]
                                                      
{'loss': 0.1618, 'grad_norm': 0.5793806314468384, 'learning_rate': 2.3494775072932173e-06, 'epoch': 2.13}

 71%|███████   | 9456/13317 [1:37:29<39:30,  1.63it/s]
 71%|███████   | 9457/13317 [1:37:29<39:29,  1.63it/s]
                                                      
{'loss': 0.1198, 'grad_norm': 0.45460042357444763, 'learning_rate': 2.3483662683905773e-06, 'epoch': 2.13}

 71%|███████   | 9457/13317 [1:37:29<39:29,  1.63it/s]
 71%|███████   | 9458/13317 [1:37:30<39:28,  1.63it/s]
                                                      
{'loss': 0.1055, 'grad_norm': 0.4397164583206177, 'learning_rate': 2.3472552116832505e-06, 'epoch': 2.13}

 71%|███████   | 9458/13317 [1:37:30<39:28,  1.63it/s]
 71%|███████   | 9459/13317 [1:37:30<39:30,  1.63it/s]
                                                      
{'loss': 0.1274, 'grad_norm': 0.5128952860832214, 'learning_rate': 2.3461443372475764e-06, 'epoch': 2.13}

 71%|███████   | 9459/13317 [1:37:30<39:30,  1.63it/s]
 71%|███████   | 9460/13317 [1:37:31<39:29,  1.63it/s]
                                                      
{'loss': 0.1597, 'grad_norm': 0.5919933319091797, 'learning_rate': 2.345033645159886e-06, 'epoch': 2.13}

 71%|███████   | 9460/13317 [1:37:31<39:29,  1.63it/s]
 71%|███████   | 9461/13317 [1:37:32<39:27,  1.63it/s]
                                                      
{'loss': 0.1321, 'grad_norm': 0.4939368963241577, 'learning_rate': 2.343923135496499e-06, 'epoch': 2.13}

 71%|███████   | 9461/13317 [1:37:32<39:27,  1.63it/s]
 71%|███████   | 9462/13317 [1:37:32<39:27,  1.63it/s]
                                                      
{'loss': 0.1309, 'grad_norm': 0.4476832449436188, 'learning_rate': 2.3428128083337116e-06, 'epoch': 2.13}

 71%|███████   | 9462/13317 [1:37:32<39:27,  1.63it/s]
 71%|███████   | 9463/13317 [1:37:33<41:27,  1.55it/s]
                                                      
{'loss': 0.1381, 'grad_norm': 0.49897244572639465, 'learning_rate': 2.341702663747819e-06, 'epoch': 2.13}

 71%|███████   | 9463/13317 [1:37:33<41:27,  1.55it/s]
 71%|███████   | 9464/13317 [1:37:33<40:54,  1.57it/s]
                                                      
{'loss': 0.158, 'grad_norm': 0.5565969347953796, 'learning_rate': 2.3405927018151025e-06, 'epoch': 2.13}

 71%|███████   | 9464/13317 [1:37:34<40:54,  1.57it/s]
 71%|███████   | 9465/13317 [1:37:34<40:30,  1.59it/s]
                                                      
{'loss': 0.1224, 'grad_norm': 0.4993838369846344, 'learning_rate': 2.339482922611825e-06, 'epoch': 2.13}

 71%|███████   | 9465/13317 [1:37:34<40:30,  1.59it/s]
 71%|███████   | 9466/13317 [1:37:35<40:09,  1.60it/s]
                                                      
{'loss': 0.2141, 'grad_norm': 0.6401931643486023, 'learning_rate': 2.3383733262142387e-06, 'epoch': 2.13}

 71%|███████   | 9466/13317 [1:37:35<40:09,  1.60it/s]
 71%|███████   | 9467/13317 [1:37:35<39:55,  1.61it/s]
                                                      
{'loss': 0.1526, 'grad_norm': 0.579030454158783, 'learning_rate': 2.3372639126985875e-06, 'epoch': 2.13}

 71%|███████   | 9467/13317 [1:37:35<39:55,  1.61it/s]
 71%|███████   | 9468/13317 [1:37:36<39:45,  1.61it/s]
                                                      
{'loss': 0.1364, 'grad_norm': 0.5149893164634705, 'learning_rate': 2.3361546821411003e-06, 'epoch': 2.13}

 71%|███████   | 9468/13317 [1:37:36<39:45,  1.61it/s]
 71%|███████   | 9469/13317 [1:37:37<39:41,  1.62it/s]
                                                      
{'loss': 0.1632, 'grad_norm': 0.5946924090385437, 'learning_rate': 2.3350456346179917e-06, 'epoch': 2.13}

 71%|███████   | 9469/13317 [1:37:37<39:41,  1.62it/s]
 71%|███████   | 9470/13317 [1:37:37<39:37,  1.62it/s]
                                                      
{'loss': 0.1554, 'grad_norm': 0.59336918592453, 'learning_rate': 2.333936770205463e-06, 'epoch': 2.13}

 71%|███████   | 9470/13317 [1:37:37<39:37,  1.62it/s]
 71%|███████   | 9471/13317 [1:37:38<39:31,  1.62it/s]
                                                      
{'loss': 0.2116, 'grad_norm': 0.6884808540344238, 'learning_rate': 2.332828088979709e-06, 'epoch': 2.13}

 71%|███████   | 9471/13317 [1:37:38<39:31,  1.62it/s]
 71%|███████   | 9472/13317 [1:37:38<39:27,  1.62it/s]
                                                      
{'loss': 0.155, 'grad_norm': 0.5866338014602661, 'learning_rate': 2.331719591016906e-06, 'epoch': 2.13}

 71%|███████   | 9472/13317 [1:37:39<39:27,  1.62it/s]
 71%|███████   | 9473/13317 [1:37:39<39:24,  1.63it/s]
                                                      
{'loss': 0.1547, 'grad_norm': 0.5751768350601196, 'learning_rate': 2.330611276393217e-06, 'epoch': 2.13}

 71%|███████   | 9473/13317 [1:37:39<39:24,  1.63it/s]
 71%|███████   | 9474/13317 [1:37:40<39:26,  1.62it/s]
                                                      
{'loss': 0.1343, 'grad_norm': 0.47249457240104675, 'learning_rate': 2.3295031451847995e-06, 'epoch': 2.13}

 71%|███████   | 9474/13317 [1:37:40<39:26,  1.62it/s]
 71%|███████   | 9475/13317 [1:37:40<41:29,  1.54it/s]
                                                      
{'loss': 0.1751, 'grad_norm': 0.6289388537406921, 'learning_rate': 2.3283951974677895e-06, 'epoch': 2.13}

 71%|███████   | 9475/13317 [1:37:40<41:29,  1.54it/s]
 71%|███████   | 9476/13317 [1:37:41<40:50,  1.57it/s]
                                                      
{'loss': 0.1383, 'grad_norm': 0.4701226055622101, 'learning_rate': 2.327287433318318e-06, 'epoch': 2.13}

 71%|███████   | 9476/13317 [1:37:41<40:50,  1.57it/s]
 71%|███████   | 9477/13317 [1:37:42<40:23,  1.58it/s]
                                                      
{'loss': 0.1378, 'grad_norm': 0.5553456544876099, 'learning_rate': 2.3261798528124986e-06, 'epoch': 2.13}

 71%|███████   | 9477/13317 [1:37:42<40:23,  1.58it/s]
 71%|███████   | 9478/13317 [1:37:42<40:03,  1.60it/s]
                                                      
{'loss': 0.1624, 'grad_norm': 0.5438458323478699, 'learning_rate': 2.3250724560264346e-06, 'epoch': 2.14}

 71%|███████   | 9478/13317 [1:37:42<40:03,  1.60it/s]
 71%|███████   | 9479/13317 [1:37:43<39:48,  1.61it/s]
                                                      
{'loss': 0.1323, 'grad_norm': 0.49553534388542175, 'learning_rate': 2.3239652430362165e-06, 'epoch': 2.14}

 71%|███████   | 9479/13317 [1:37:43<39:48,  1.61it/s]
 71%|███████   | 9480/13317 [1:37:43<39:40,  1.61it/s]
                                                      
{'loss': 0.1264, 'grad_norm': 0.4579916298389435, 'learning_rate': 2.322858213917918e-06, 'epoch': 2.14}

 71%|███████   | 9480/13317 [1:37:44<39:40,  1.61it/s]
 71%|███████   | 9481/13317 [1:37:44<39:31,  1.62it/s]
                                                      
{'loss': 0.1715, 'grad_norm': 0.6873247623443604, 'learning_rate': 2.3217513687476084e-06, 'epoch': 2.14}

 71%|███████   | 9481/13317 [1:37:44<39:31,  1.62it/s]
 71%|███████   | 9482/13317 [1:37:45<39:24,  1.62it/s]
                                                      
{'loss': 0.1657, 'grad_norm': 0.5455465316772461, 'learning_rate': 2.3206447076013357e-06, 'epoch': 2.14}

 71%|███████   | 9482/13317 [1:37:45<39:24,  1.62it/s]
 71%|███████   | 9483/13317 [1:37:45<39:21,  1.62it/s]
                                                      
{'loss': 0.1599, 'grad_norm': 0.594218909740448, 'learning_rate': 2.319538230555143e-06, 'epoch': 2.14}

 71%|███████   | 9483/13317 [1:37:45<39:21,  1.62it/s]
 71%|███████   | 9484/13317 [1:37:46<39:16,  1.63it/s]
                                                      
{'loss': 0.1487, 'grad_norm': 0.5299713611602783, 'learning_rate': 2.318431937685053e-06, 'epoch': 2.14}

 71%|███████   | 9484/13317 [1:37:46<39:16,  1.63it/s]
 71%|███████   | 9485/13317 [1:37:46<39:14,  1.63it/s]
                                                      
{'loss': 0.1294, 'grad_norm': 0.4995766878128052, 'learning_rate': 2.3173258290670835e-06, 'epoch': 2.14}

 71%|███████   | 9485/13317 [1:37:47<39:14,  1.63it/s]
 71%|███████   | 9486/13317 [1:37:47<39:12,  1.63it/s]
                                                      
{'loss': 0.1546, 'grad_norm': 0.5356048941612244, 'learning_rate': 2.316219904777234e-06, 'epoch': 2.14}

 71%|███████   | 9486/13317 [1:37:47<39:12,  1.63it/s]
 71%|███████   | 9487/13317 [1:37:48<39:12,  1.63it/s]
                                                      
{'loss': 0.1824, 'grad_norm': 0.5868626832962036, 'learning_rate': 2.3151141648914922e-06, 'epoch': 2.14}

 71%|███████   | 9487/13317 [1:37:48<39:12,  1.63it/s]
 71%|███████   | 9488/13317 [1:37:48<39:11,  1.63it/s]
                                                      
{'loss': 0.164, 'grad_norm': 0.548800528049469, 'learning_rate': 2.3140086094858365e-06, 'epoch': 2.14}

 71%|███████   | 9488/13317 [1:37:48<39:11,  1.63it/s]
 71%|███████▏  | 9489/13317 [1:37:49<39:11,  1.63it/s]
                                                      
{'loss': 0.1666, 'grad_norm': 0.5916743874549866, 'learning_rate': 2.3129032386362276e-06, 'epoch': 2.14}

 71%|███████▏  | 9489/13317 [1:37:49<39:11,  1.63it/s]
 71%|███████▏  | 9490/13317 [1:37:50<39:11,  1.63it/s]
                                                      
{'loss': 0.1709, 'grad_norm': 0.6505721211433411, 'learning_rate': 2.3117980524186195e-06, 'epoch': 2.14}

 71%|███████▏  | 9490/13317 [1:37:50<39:11,  1.63it/s]
 71%|███████▏  | 9491/13317 [1:37:50<39:10,  1.63it/s]
                                                      
{'loss': 0.1474, 'grad_norm': 0.8375946283340454, 'learning_rate': 2.3106930509089482e-06, 'epoch': 2.14}

 71%|███████▏  | 9491/13317 [1:37:50<39:10,  1.63it/s]
 71%|███████▏  | 9492/13317 [1:37:51<39:09,  1.63it/s]
                                                      
{'loss': 0.1351, 'grad_norm': 0.5454652309417725, 'learning_rate': 2.309588234183137e-06, 'epoch': 2.14}

 71%|███████▏  | 9492/13317 [1:37:51<39:09,  1.63it/s]
 71%|███████▏  | 9493/13317 [1:37:51<39:07,  1.63it/s]
                                                      
{'loss': 0.1782, 'grad_norm': 0.5185303688049316, 'learning_rate': 2.308483602317103e-06, 'epoch': 2.14}

 71%|███████▏  | 9493/13317 [1:37:52<39:07,  1.63it/s]
 71%|███████▏  | 9494/13317 [1:37:52<39:08,  1.63it/s]
                                                      
{'loss': 0.2043, 'grad_norm': 0.6138970851898193, 'learning_rate': 2.3073791553867413e-06, 'epoch': 2.14}

 71%|███████▏  | 9494/13317 [1:37:52<39:08,  1.63it/s]
 71%|███████▏  | 9495/13317 [1:37:53<39:08,  1.63it/s]
                                                      
{'loss': 0.1613, 'grad_norm': 0.5762250423431396, 'learning_rate': 2.306274893467944e-06, 'epoch': 2.14}

 71%|███████▏  | 9495/13317 [1:37:53<39:08,  1.63it/s]
 71%|███████▏  | 9496/13317 [1:37:53<39:06,  1.63it/s]
                                                      
{'loss': 0.1993, 'grad_norm': 0.6836298108100891, 'learning_rate': 2.3051708166365812e-06, 'epoch': 2.14}

 71%|███████▏  | 9496/13317 [1:37:53<39:06,  1.63it/s]
 71%|███████▏  | 9497/13317 [1:37:54<39:04,  1.63it/s]
                                                      
{'loss': 0.1222, 'grad_norm': 0.4405120015144348, 'learning_rate': 2.304066924968518e-06, 'epoch': 2.14}

 71%|███████▏  | 9497/13317 [1:37:54<39:04,  1.63it/s]
 71%|███████▏  | 9498/13317 [1:37:54<39:04,  1.63it/s]
                                                      
{'loss': 0.1826, 'grad_norm': 0.5730105638504028, 'learning_rate': 2.3029632185396028e-06, 'epoch': 2.14}

 71%|███████▏  | 9498/13317 [1:37:55<39:04,  1.63it/s]
 71%|███████▏  | 9499/13317 [1:37:55<39:03,  1.63it/s]
                                                      
{'loss': 0.1649, 'grad_norm': 0.5577694773674011, 'learning_rate': 2.3018596974256685e-06, 'epoch': 2.14}

 71%|███████▏  | 9499/13317 [1:37:55<39:03,  1.63it/s]
 71%|███████▏  | 9500/13317 [1:37:56<39:03,  1.63it/s]
                                                      
{'loss': 0.135, 'grad_norm': 0.5918098092079163, 'learning_rate': 2.300756361702542e-06, 'epoch': 2.14}

 71%|███████▏  | 9500/13317 [1:37:56<39:03,  1.63it/s]

  0%|          | 0/13 [00:00<?, ?it/s]

 15%|█▌        | 2/13 [00:00<00:00, 13.36it/s]

 31%|███       | 4/13 [00:00<00:01,  8.39it/s]

 38%|███▊      | 5/13 [00:00<00:01,  7.79it/s]

 46%|████▌     | 6/13 [00:00<00:00,  7.41it/s]

 54%|█████▍    | 7/13 [00:00<00:00,  7.17it/s]

 62%|██████▏   | 8/13 [00:01<00:00,  7.01it/s]

 69%|██████▉   | 9/13 [00:01<00:00,  6.91it/s]

 77%|███████▋  | 10/13 [00:01<00:00,  6.83it/s]

 85%|████████▍ | 11/13 [00:01<00:00,  6.78it/s]

 92%|█████████▏| 12/13 [00:01<00:00,  6.75it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]
                                                      

                                               
{'eval_loss': 0.5490559935569763, 'eval_runtime': 1.9536, 'eval_samples_per_second': 102.375, 'eval_steps_per_second': 6.654, 'epoch': 2.14}

 71%|███████▏  | 9500/13317 [1:37:58<39:03,  1.63it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]

                                               
 71%|███████▏  | 9501/13317 [1:37:58<1:16:26,  1.20s/it]
                                                        
{'loss': 0.1387, 'grad_norm': 0.5648056864738464, 'learning_rate': 2.299653211446037e-06, 'epoch': 2.14}

 71%|███████▏  | 9501/13317 [1:37:58<1:16:26,  1.20s/it]
 71%|███████▏  | 9502/13317 [1:37:59<1:05:13,  1.03s/it]
                                                        
{'loss': 0.1179, 'grad_norm': 0.41321712732315063, 'learning_rate': 2.298550246731944e-06, 'epoch': 2.14}

 71%|███████▏  | 9502/13317 [1:37:59<1:05:13,  1.03s/it]
 71%|███████▏  | 9503/13317 [1:37:59<57:20,  1.11it/s]  
                                                      
{'loss': 0.1594, 'grad_norm': 0.5590125322341919, 'learning_rate': 2.2974474676360533e-06, 'epoch': 2.14}

 71%|███████▏  | 9503/13317 [1:38:00<57:20,  1.11it/s]
 71%|███████▏  | 9504/13317 [1:38:00<51:50,  1.23it/s]
                                                      
{'loss': 0.1414, 'grad_norm': 0.5150036215782166, 'learning_rate': 2.2963448742341376e-06, 'epoch': 2.14}

 71%|███████▏  | 9504/13317 [1:38:00<51:50,  1.23it/s]
 71%|███████▏  | 9505/13317 [1:38:01<47:57,  1.32it/s]
                                                      
{'loss': 0.1292, 'grad_norm': 0.48461800813674927, 'learning_rate': 2.2952424666019564e-06, 'epoch': 2.14}

 71%|███████▏  | 9505/13317 [1:38:01<47:57,  1.32it/s]
 71%|███████▏  | 9506/13317 [1:38:01<45:14,  1.40it/s]
                                                      
{'loss': 0.1423, 'grad_norm': 0.48838144540786743, 'learning_rate': 2.2941402448152545e-06, 'epoch': 2.14}

 71%|███████▏  | 9506/13317 [1:38:01<45:14,  1.40it/s]
 71%|███████▏  | 9507/13317 [1:38:02<43:21,  1.46it/s]
                                                      
{'loss': 0.1594, 'grad_norm': 0.5903033018112183, 'learning_rate': 2.293038208949767e-06, 'epoch': 2.14}

 71%|███████▏  | 9507/13317 [1:38:02<43:21,  1.46it/s]
 71%|███████▏  | 9508/13317 [1:38:03<42:01,  1.51it/s]
                                                      
{'loss': 0.1593, 'grad_norm': 0.5410141944885254, 'learning_rate': 2.291936359081221e-06, 'epoch': 2.14}

 71%|███████▏  | 9508/13317 [1:38:03<42:01,  1.51it/s]
 71%|███████▏  | 9509/13317 [1:38:03<41:05,  1.54it/s]
                                                      
{'loss': 0.1433, 'grad_norm': 0.5610045790672302, 'learning_rate': 2.290834695285316e-06, 'epoch': 2.14}

 71%|███████▏  | 9509/13317 [1:38:03<41:05,  1.54it/s]
 71%|███████▏  | 9510/13317 [1:38:04<40:25,  1.57it/s]
                                                      
{'loss': 0.1481, 'grad_norm': 0.5857846140861511, 'learning_rate': 2.289733217637753e-06, 'epoch': 2.14}

 71%|███████▏  | 9510/13317 [1:38:04<40:25,  1.57it/s]
 71%|███████▏  | 9511/13317 [1:38:04<39:57,  1.59it/s]
                                                      
{'loss': 0.1385, 'grad_norm': 0.5155845880508423, 'learning_rate': 2.2886319262142163e-06, 'epoch': 2.14}

 71%|███████▏  | 9511/13317 [1:38:05<39:57,  1.59it/s]
 71%|███████▏  | 9512/13317 [1:38:05<39:37,  1.60it/s]
                                                      
{'loss': 0.1448, 'grad_norm': 0.4914042055606842, 'learning_rate': 2.287530821090375e-06, 'epoch': 2.14}

 71%|███████▏  | 9512/13317 [1:38:05<39:37,  1.60it/s]
 71%|███████▏  | 9513/13317 [1:38:06<39:24,  1.61it/s]
                                                      
{'loss': 0.1344, 'grad_norm': 0.4867173731327057, 'learning_rate': 2.2864299023418846e-06, 'epoch': 2.14}

 71%|███████▏  | 9513/13317 [1:38:06<39:24,  1.61it/s]
 71%|███████▏  | 9514/13317 [1:38:06<39:14,  1.62it/s]
                                                      
{'loss': 0.151, 'grad_norm': 0.5276265144348145, 'learning_rate': 2.2853291700443914e-06, 'epoch': 2.14}

 71%|███████▏  | 9514/13317 [1:38:06<39:14,  1.62it/s]
 71%|███████▏  | 9515/13317 [1:38:07<39:06,  1.62it/s]
                                                      
{'loss': 0.133, 'grad_norm': 0.5099972486495972, 'learning_rate': 2.2842286242735295e-06, 'epoch': 2.14}

 71%|███████▏  | 9515/13317 [1:38:07<39:06,  1.62it/s]
 71%|███████▏  | 9516/13317 [1:38:07<39:02,  1.62it/s]
                                                      
{'loss': 0.1651, 'grad_norm': 0.5538008213043213, 'learning_rate': 2.283128265104916e-06, 'epoch': 2.14}

 71%|███████▏  | 9516/13317 [1:38:08<39:02,  1.62it/s]
 71%|███████▏  | 9517/13317 [1:38:08<38:58,  1.63it/s]
                                                      
{'loss': 0.1652, 'grad_norm': 0.652894139289856, 'learning_rate': 2.2820280926141562e-06, 'epoch': 2.14}

 71%|███████▏  | 9517/13317 [1:38:08<38:58,  1.63it/s]
 71%|███████▏  | 9518/13317 [1:38:09<38:55,  1.63it/s]
                                                      
{'loss': 0.1708, 'grad_norm': 0.6002802848815918, 'learning_rate': 2.280928106876846e-06, 'epoch': 2.14}

 71%|███████▏  | 9518/13317 [1:38:09<38:55,  1.63it/s]
 71%|███████▏  | 9519/13317 [1:38:09<38:53,  1.63it/s]
                                                      
{'loss': 0.1344, 'grad_norm': 0.49421295523643494, 'learning_rate': 2.2798283079685653e-06, 'epoch': 2.14}

 71%|███████▏  | 9519/13317 [1:38:09<38:53,  1.63it/s]
 71%|███████▏  | 9520/13317 [1:38:10<38:51,  1.63it/s]
                                                      
{'loss': 0.1109, 'grad_norm': 0.4712539613246918, 'learning_rate': 2.2787286959648793e-06, 'epoch': 2.14}

 71%|███████▏  | 9520/13317 [1:38:10<38:51,  1.63it/s]
 71%|███████▏  | 9521/13317 [1:38:11<38:49,  1.63it/s]
                                                      
{'loss': 0.1361, 'grad_norm': 0.4938821792602539, 'learning_rate': 2.277629270941345e-06, 'epoch': 2.14}

 71%|███████▏  | 9521/13317 [1:38:11<38:49,  1.63it/s]
 72%|███████▏  | 9522/13317 [1:38:11<38:49,  1.63it/s]
                                                      
{'loss': 0.1932, 'grad_norm': 0.6912675499916077, 'learning_rate': 2.2765300329735066e-06, 'epoch': 2.15}

 72%|███████▏  | 9522/13317 [1:38:11<38:49,  1.63it/s]
 72%|███████▏  | 9523/13317 [1:38:12<38:49,  1.63it/s]
                                                      
{'loss': 0.1447, 'grad_norm': 0.5566692352294922, 'learning_rate': 2.2754309821368918e-06, 'epoch': 2.15}

 72%|███████▏  | 9523/13317 [1:38:12<38:49,  1.63it/s]
 72%|███████▏  | 9524/13317 [1:38:12<38:47,  1.63it/s]
                                                      
{'loss': 0.159, 'grad_norm': 0.5679302215576172, 'learning_rate': 2.274332118507015e-06, 'epoch': 2.15}

 72%|███████▏  | 9524/13317 [1:38:12<38:47,  1.63it/s]
 72%|███████▏  | 9525/13317 [1:38:13<38:47,  1.63it/s]
                                                      
{'loss': 0.1815, 'grad_norm': 0.6947090029716492, 'learning_rate': 2.2732334421593833e-06, 'epoch': 2.15}

 72%|███████▏  | 9525/13317 [1:38:13<38:47,  1.63it/s]
 72%|███████▏  | 9526/13317 [1:38:14<38:45,  1.63it/s]
                                                      
{'loss': 0.1645, 'grad_norm': 0.5526297092437744, 'learning_rate': 2.2721349531694852e-06, 'epoch': 2.15}

 72%|███████▏  | 9526/13317 [1:38:14<38:45,  1.63it/s]
 72%|███████▏  | 9527/13317 [1:38:14<38:44,  1.63it/s]
                                                      
{'loss': 0.1451, 'grad_norm': 0.5510047078132629, 'learning_rate': 2.2710366516127974e-06, 'epoch': 2.15}

 72%|███████▏  | 9527/13317 [1:38:14<38:44,  1.63it/s]
 72%|███████▏  | 9528/13317 [1:38:15<38:45,  1.63it/s]
                                                      
{'loss': 0.1444, 'grad_norm': 0.5307261943817139, 'learning_rate': 2.269938537564786e-06, 'epoch': 2.15}

 72%|███████▏  | 9528/13317 [1:38:15<38:45,  1.63it/s]
 72%|███████▏  | 9529/13317 [1:38:15<38:46,  1.63it/s]
                                                      
{'loss': 0.1329, 'grad_norm': 0.48110970854759216, 'learning_rate': 2.268840611100906e-06, 'epoch': 2.15}

 72%|███████▏  | 9529/13317 [1:38:16<38:46,  1.63it/s]
 72%|███████▏  | 9530/13317 [1:38:16<38:45,  1.63it/s]
                                                      
{'loss': 0.1767, 'grad_norm': 0.5438015460968018, 'learning_rate': 2.2677428722965932e-06, 'epoch': 2.15}

 72%|███████▏  | 9530/13317 [1:38:16<38:45,  1.63it/s]
 72%|███████▏  | 9531/13317 [1:38:17<38:45,  1.63it/s]
                                                      
{'loss': 0.1347, 'grad_norm': 0.623134195804596, 'learning_rate': 2.2666453212272733e-06, 'epoch': 2.15}

 72%|███████▏  | 9531/13317 [1:38:17<38:45,  1.63it/s]
 72%|███████▏  | 9532/13317 [1:38:17<38:45,  1.63it/s]
                                                      
{'loss': 0.1408, 'grad_norm': 0.5937651991844177, 'learning_rate': 2.2655479579683627e-06, 'epoch': 2.15}

 72%|███████▏  | 9532/13317 [1:38:17<38:45,  1.63it/s]
 72%|███████▏  | 9533/13317 [1:38:18<38:44,  1.63it/s]
                                                      
{'loss': 0.1597, 'grad_norm': 0.6380331516265869, 'learning_rate': 2.26445078259526e-06, 'epoch': 2.15}

 72%|███████▏  | 9533/13317 [1:38:18<38:44,  1.63it/s]
 72%|███████▏  | 9534/13317 [1:38:19<38:42,  1.63it/s]
                                                      
{'loss': 0.1192, 'grad_norm': 0.5048477649688721, 'learning_rate': 2.2633537951833513e-06, 'epoch': 2.15}

 72%|███████▏  | 9534/13317 [1:38:19<38:42,  1.63it/s]
 72%|███████▏  | 9535/13317 [1:38:19<38:41,  1.63it/s]
                                                      
{'loss': 0.1481, 'grad_norm': 0.5153228640556335, 'learning_rate': 2.2622569958080147e-06, 'epoch': 2.15}

 72%|███████▏  | 9535/13317 [1:38:19<38:41,  1.63it/s]
 72%|███████▏  | 9536/13317 [1:38:20<38:41,  1.63it/s]
                                                      
{'loss': 0.1675, 'grad_norm': 0.6157910227775574, 'learning_rate': 2.261160384544609e-06, 'epoch': 2.15}

 72%|███████▏  | 9536/13317 [1:38:20<38:41,  1.63it/s]
 72%|███████▏  | 9537/13317 [1:38:20<38:40,  1.63it/s]
                                                      
{'loss': 0.1491, 'grad_norm': 0.5580548048019409, 'learning_rate': 2.2600639614684857e-06, 'epoch': 2.15}

 72%|███████▏  | 9537/13317 [1:38:20<38:40,  1.63it/s]
 72%|███████▏  | 9538/13317 [1:38:21<38:40,  1.63it/s]
                                                      
{'loss': 0.1196, 'grad_norm': 0.463003933429718, 'learning_rate': 2.258967726654977e-06, 'epoch': 2.15}

 72%|███████▏  | 9538/13317 [1:38:21<38:40,  1.63it/s]
 72%|███████▏  | 9539/13317 [1:38:22<38:38,  1.63it/s]
                                                      
{'loss': 0.1187, 'grad_norm': 0.4592392146587372, 'learning_rate': 2.25787168017941e-06, 'epoch': 2.15}

 72%|███████▏  | 9539/13317 [1:38:22<38:38,  1.63it/s]
 72%|███████▏  | 9540/13317 [1:38:22<38:38,  1.63it/s]
                                                      
{'loss': 0.1355, 'grad_norm': 0.5075501203536987, 'learning_rate': 2.256775822117093e-06, 'epoch': 2.15}

 72%|███████▏  | 9540/13317 [1:38:22<38:38,  1.63it/s]
 72%|███████▏  | 9541/13317 [1:38:23<38:37,  1.63it/s]
                                                      
{'loss': 0.1325, 'grad_norm': 0.533390998840332, 'learning_rate': 2.2556801525433202e-06, 'epoch': 2.15}

 72%|███████▏  | 9541/13317 [1:38:23<38:37,  1.63it/s]
 72%|███████▏  | 9542/13317 [1:38:23<38:35,  1.63it/s]
                                                      
{'loss': 0.1424, 'grad_norm': 0.5140653848648071, 'learning_rate': 2.2545846715333804e-06, 'epoch': 2.15}

 72%|███████▏  | 9542/13317 [1:38:24<38:35,  1.63it/s]
 72%|███████▏  | 9543/13317 [1:38:24<38:35,  1.63it/s]
                                                      
{'loss': 0.1583, 'grad_norm': 0.5486966967582703, 'learning_rate': 2.2534893791625408e-06, 'epoch': 2.15}

 72%|███████▏  | 9543/13317 [1:38:24<38:35,  1.63it/s]
 72%|███████▏  | 9544/13317 [1:38:25<38:35,  1.63it/s]
                                                      
{'loss': 0.139, 'grad_norm': 0.47871753573417664, 'learning_rate': 2.2523942755060636e-06, 'epoch': 2.15}

 72%|███████▏  | 9544/13317 [1:38:25<38:35,  1.63it/s]
 72%|███████▏  | 9545/13317 [1:38:25<38:35,  1.63it/s]
                                                      
{'loss': 0.1881, 'grad_norm': 0.8429793119430542, 'learning_rate': 2.251299360639189e-06, 'epoch': 2.15}

 72%|███████▏  | 9545/13317 [1:38:25<38:35,  1.63it/s]
 72%|███████▏  | 9546/13317 [1:38:26<38:34,  1.63it/s]
                                                      
{'loss': 0.1535, 'grad_norm': 0.5156388282775879, 'learning_rate': 2.250204634637155e-06, 'epoch': 2.15}

 72%|███████▏  | 9546/13317 [1:38:26<38:34,  1.63it/s]
 72%|███████▏  | 9547/13317 [1:38:26<38:32,  1.63it/s]
                                                      
{'loss': 0.1481, 'grad_norm': 0.5509892702102661, 'learning_rate': 2.249110097575175e-06, 'epoch': 2.15}

 72%|███████▏  | 9547/13317 [1:38:27<38:32,  1.63it/s]
 72%|███████▏  | 9548/13317 [1:38:27<38:31,  1.63it/s]
                                                      
{'loss': 0.1628, 'grad_norm': 0.5536291599273682, 'learning_rate': 2.248015749528461e-06, 'epoch': 2.15}

 72%|███████▏  | 9548/13317 [1:38:27<38:31,  1.63it/s]
 72%|███████▏  | 9549/13317 [1:38:28<38:30,  1.63it/s]
                                                      
{'loss': 0.1171, 'grad_norm': 0.4781899154186249, 'learning_rate': 2.246921590572203e-06, 'epoch': 2.15}

 72%|███████▏  | 9549/13317 [1:38:28<38:30,  1.63it/s]
 72%|███████▏  | 9550/13317 [1:38:28<38:29,  1.63it/s]
                                                      
{'loss': 0.1592, 'grad_norm': 0.5661247372627258, 'learning_rate': 2.2458276207815797e-06, 'epoch': 2.15}

 72%|███████▏  | 9550/13317 [1:38:28<38:29,  1.63it/s]
 72%|███████▏  | 9551/13317 [1:38:29<38:29,  1.63it/s]
                                                      
{'loss': 0.168, 'grad_norm': 0.6259070634841919, 'learning_rate': 2.2447338402317624e-06, 'epoch': 2.15}

 72%|███████▏  | 9551/13317 [1:38:29<38:29,  1.63it/s]
 72%|███████▏  | 9552/13317 [1:38:30<38:28,  1.63it/s]
                                                      
{'loss': 0.1542, 'grad_norm': 0.5465496778488159, 'learning_rate': 2.2436402489979042e-06, 'epoch': 2.15}

 72%|███████▏  | 9552/13317 [1:38:30<38:28,  1.63it/s]
 72%|███████▏  | 9553/13317 [1:38:30<38:27,  1.63it/s]
                                                      
{'loss': 0.1207, 'grad_norm': 0.40798068046569824, 'learning_rate': 2.2425468471551433e-06, 'epoch': 2.15}

 72%|███████▏  | 9553/13317 [1:38:30<38:27,  1.63it/s]
 72%|███████▏  | 9554/13317 [1:38:31<38:27,  1.63it/s]
                                                      
{'loss': 0.1744, 'grad_norm': 0.5549492239952087, 'learning_rate': 2.2414536347786103e-06, 'epoch': 2.15}

 72%|███████▏  | 9554/13317 [1:38:31<38:27,  1.63it/s]
 72%|███████▏  | 9555/13317 [1:38:31<38:26,  1.63it/s]
                                                      
{'loss': 0.168, 'grad_norm': 0.6264676451683044, 'learning_rate': 2.240360611943423e-06, 'epoch': 2.15}

 72%|███████▏  | 9555/13317 [1:38:32<38:26,  1.63it/s]
 72%|███████▏  | 9556/13317 [1:38:32<38:25,  1.63it/s]
                                                      
{'loss': 0.14, 'grad_norm': 0.5023837685585022, 'learning_rate': 2.2392677787246804e-06, 'epoch': 2.15}

 72%|███████▏  | 9556/13317 [1:38:32<38:25,  1.63it/s]
 72%|███████▏  | 9557/13317 [1:38:33<38:26,  1.63it/s]
                                                      
{'loss': 0.137, 'grad_norm': 0.5306767821311951, 'learning_rate': 2.238175135197471e-06, 'epoch': 2.15}

 72%|███████▏  | 9557/13317 [1:38:33<38:26,  1.63it/s]
 72%|███████▏  | 9558/13317 [1:38:33<38:25,  1.63it/s]
                                                      
{'loss': 0.1721, 'grad_norm': 0.5886092185974121, 'learning_rate': 2.2370826814368745e-06, 'epoch': 2.15}

 72%|███████▏  | 9558/13317 [1:38:33<38:25,  1.63it/s]
 72%|███████▏  | 9559/13317 [1:38:34<38:24,  1.63it/s]
                                                      
{'loss': 0.1319, 'grad_norm': 0.48273637890815735, 'learning_rate': 2.2359904175179515e-06, 'epoch': 2.15}

 72%|███████▏  | 9559/13317 [1:38:34<38:24,  1.63it/s]
 72%|███████▏  | 9560/13317 [1:38:34<38:24,  1.63it/s]
                                                      
{'loss': 0.1243, 'grad_norm': 0.47814038395881653, 'learning_rate': 2.234898343515751e-06, 'epoch': 2.15}

 72%|███████▏  | 9560/13317 [1:38:35<38:24,  1.63it/s]
 72%|███████▏  | 9561/13317 [1:38:35<38:23,  1.63it/s]
                                                      
{'loss': 0.1901, 'grad_norm': 0.6735563278198242, 'learning_rate': 2.2338064595053123e-06, 'epoch': 2.15}

 72%|███████▏  | 9561/13317 [1:38:35<38:23,  1.63it/s]
 72%|███████▏  | 9562/13317 [1:38:36<38:22,  1.63it/s]
                                                      
{'loss': 0.1516, 'grad_norm': 0.5637577772140503, 'learning_rate': 2.2327147655616615e-06, 'epoch': 2.15}

 72%|███████▏  | 9562/13317 [1:38:36<38:22,  1.63it/s]
 72%|███████▏  | 9563/13317 [1:38:36<38:23,  1.63it/s]
                                                      
{'loss': 0.1597, 'grad_norm': 0.5105282068252563, 'learning_rate': 2.2316232617598037e-06, 'epoch': 2.15}

 72%|███████▏  | 9563/13317 [1:38:36<38:23,  1.63it/s]
 72%|███████▏  | 9564/13317 [1:38:37<38:22,  1.63it/s]
                                                      
{'loss': 0.1439, 'grad_norm': 0.5562664866447449, 'learning_rate': 2.2305319481747394e-06, 'epoch': 2.15}

 72%|███████▏  | 9564/13317 [1:38:37<38:22,  1.63it/s]
 72%|███████▏  | 9565/13317 [1:38:38<38:22,  1.63it/s]
                                                      
{'loss': 0.1448, 'grad_norm': 0.5235214829444885, 'learning_rate': 2.2294408248814564e-06, 'epoch': 2.15}

 72%|███████▏  | 9565/13317 [1:38:38<38:22,  1.63it/s]
 72%|███████▏  | 9566/13317 [1:38:38<38:21,  1.63it/s]
                                                      
{'loss': 0.1896, 'grad_norm': 0.5919169187545776, 'learning_rate': 2.2283498919549235e-06, 'epoch': 2.15}

 72%|███████▏  | 9566/13317 [1:38:38<38:21,  1.63it/s]
 72%|███████▏  | 9567/13317 [1:38:39<38:20,  1.63it/s]
                                                      
{'loss': 0.1487, 'grad_norm': 0.5886659026145935, 'learning_rate': 2.227259149470098e-06, 'epoch': 2.16}

 72%|███████▏  | 9567/13317 [1:38:39<38:20,  1.63it/s]
 72%|███████▏  | 9568/13317 [1:38:39<38:19,  1.63it/s]
                                                      
{'loss': 0.1379, 'grad_norm': 0.4941645860671997, 'learning_rate': 2.226168597501927e-06, 'epoch': 2.16}

 72%|███████▏  | 9568/13317 [1:38:39<38:19,  1.63it/s]
 72%|███████▏  | 9569/13317 [1:38:40<39:14,  1.59it/s]
                                                      
{'loss': 0.1587, 'grad_norm': 0.599963366985321, 'learning_rate': 2.2250782361253447e-06, 'epoch': 2.16}

 72%|███████▏  | 9569/13317 [1:38:40<39:14,  1.59it/s]
 72%|███████▏  | 9570/13317 [1:38:41<39:11,  1.59it/s]
                                                      
{'loss': 0.1351, 'grad_norm': 0.5043509602546692, 'learning_rate': 2.2239880654152698e-06, 'epoch': 2.16}

 72%|███████▏  | 9570/13317 [1:38:41<39:11,  1.59it/s]
 72%|███████▏  | 9571/13317 [1:38:41<38:57,  1.60it/s]
                                                      
{'loss': 0.1477, 'grad_norm': 0.5587493181228638, 'learning_rate': 2.222898085446605e-06, 'epoch': 2.16}

 72%|███████▏  | 9571/13317 [1:38:41<38:57,  1.60it/s]
 72%|███████▏  | 9572/13317 [1:38:42<38:46,  1.61it/s]
                                                      
{'loss': 0.1346, 'grad_norm': 0.4679979979991913, 'learning_rate': 2.221808296294246e-06, 'epoch': 2.16}

 72%|███████▏  | 9572/13317 [1:38:42<38:46,  1.61it/s]
 72%|███████▏  | 9573/13317 [1:38:42<38:41,  1.61it/s]
                                                      
{'loss': 0.1227, 'grad_norm': 0.4111804664134979, 'learning_rate': 2.2207186980330775e-06, 'epoch': 2.16}

 72%|███████▏  | 9573/13317 [1:38:43<38:41,  1.61it/s]
 72%|███████▏  | 9574/13317 [1:38:43<38:35,  1.62it/s]
                                                      
{'loss': 0.1353, 'grad_norm': 0.4900304079055786, 'learning_rate': 2.219629290737958e-06, 'epoch': 2.16}

 72%|███████▏  | 9574/13317 [1:38:43<38:35,  1.62it/s]
 72%|███████▏  | 9575/13317 [1:38:44<38:33,  1.62it/s]
                                                      
{'loss': 0.1386, 'grad_norm': 0.5183537602424622, 'learning_rate': 2.2185400744837447e-06, 'epoch': 2.16}

 72%|███████▏  | 9575/13317 [1:38:44<38:33,  1.62it/s]
 72%|███████▏  | 9576/13317 [1:38:44<38:27,  1.62it/s]
                                                      
{'loss': 0.1721, 'grad_norm': 0.701962947845459, 'learning_rate': 2.2174510493452817e-06, 'epoch': 2.16}

 72%|███████▏  | 9576/13317 [1:38:44<38:27,  1.62it/s]
 72%|███████▏  | 9577/13317 [1:38:45<38:34,  1.62it/s]
                                                      
{'loss': 0.1654, 'grad_norm': 0.5604267716407776, 'learning_rate': 2.216362215397393e-06, 'epoch': 2.16}

 72%|███████▏  | 9577/13317 [1:38:45<38:34,  1.62it/s]
 72%|███████▏  | 9578/13317 [1:38:46<38:28,  1.62it/s]
                                                      
{'loss': 0.1358, 'grad_norm': 0.5261970162391663, 'learning_rate': 2.215273572714892e-06, 'epoch': 2.16}

 72%|███████▏  | 9578/13317 [1:38:46<38:28,  1.62it/s]
 72%|███████▏  | 9579/13317 [1:38:46<38:22,  1.62it/s]
                                                      
{'loss': 0.1715, 'grad_norm': 0.5669317245483398, 'learning_rate': 2.214185121372584e-06, 'epoch': 2.16}

 72%|███████▏  | 9579/13317 [1:38:46<38:22,  1.62it/s]
 72%|███████▏  | 9580/13317 [1:38:47<38:22,  1.62it/s]
                                                      
{'loss': 0.1713, 'grad_norm': 0.5866199135780334, 'learning_rate': 2.2130968614452543e-06, 'epoch': 2.16}

 72%|███████▏  | 9580/13317 [1:38:47<38:22,  1.62it/s]
 72%|███████▏  | 9581/13317 [1:38:47<38:17,  1.63it/s]
                                                      
{'loss': 0.1522, 'grad_norm': 0.5404232144355774, 'learning_rate': 2.212008793007677e-06, 'epoch': 2.16}

 72%|███████▏  | 9581/13317 [1:38:48<38:17,  1.63it/s]
 72%|███████▏  | 9582/13317 [1:38:48<38:27,  1.62it/s]
                                                      
{'loss': 0.1485, 'grad_norm': 0.5565872192382812, 'learning_rate': 2.2109209161346146e-06, 'epoch': 2.16}

 72%|███████▏  | 9582/13317 [1:38:48<38:27,  1.62it/s]
 72%|███████▏  | 9583/13317 [1:38:49<38:22,  1.62it/s]
                                                      
{'loss': 0.1326, 'grad_norm': 0.545772910118103, 'learning_rate': 2.209833230900819e-06, 'epoch': 2.16}

 72%|███████▏  | 9583/13317 [1:38:49<38:22,  1.62it/s]
 72%|███████▏  | 9584/13317 [1:38:49<38:18,  1.62it/s]
                                                      
{'loss': 0.1458, 'grad_norm': 0.5716257095336914, 'learning_rate': 2.208745737381023e-06, 'epoch': 2.16}

 72%|███████▏  | 9584/13317 [1:38:49<38:18,  1.62it/s]
 72%|███████▏  | 9585/13317 [1:38:50<38:17,  1.62it/s]
                                                      
{'loss': 0.1668, 'grad_norm': 0.5383652448654175, 'learning_rate': 2.2076584356499476e-06, 'epoch': 2.16}

 72%|███████▏  | 9585/13317 [1:38:50<38:17,  1.62it/s]
 72%|███████▏  | 9586/13317 [1:38:51<38:15,  1.63it/s]
                                                      
{'loss': 0.154, 'grad_norm': 0.5514011979103088, 'learning_rate': 2.2065713257823046e-06, 'epoch': 2.16}

 72%|███████▏  | 9586/13317 [1:38:51<38:15,  1.63it/s]
 72%|███████▏  | 9587/13317 [1:38:51<38:17,  1.62it/s]
                                                      
{'loss': 0.1801, 'grad_norm': 0.7175381183624268, 'learning_rate': 2.205484407852789e-06, 'epoch': 2.16}

 72%|███████▏  | 9587/13317 [1:38:51<38:17,  1.62it/s]
 72%|███████▏  | 9588/13317 [1:38:52<38:17,  1.62it/s]
                                                      
{'loss': 0.1475, 'grad_norm': 0.7204063534736633, 'learning_rate': 2.2043976819360823e-06, 'epoch': 2.16}

 72%|███████▏  | 9588/13317 [1:38:52<38:17,  1.62it/s]
 72%|███████▏  | 9589/13317 [1:38:52<38:13,  1.63it/s]
                                                      
{'loss': 0.1432, 'grad_norm': 0.5656189322471619, 'learning_rate': 2.203311148106855e-06, 'epoch': 2.16}

 72%|███████▏  | 9589/13317 [1:38:52<38:13,  1.63it/s]
 72%|███████▏  | 9590/13317 [1:38:53<38:22,  1.62it/s]
                                                      
{'loss': 0.1407, 'grad_norm': 0.5499799847602844, 'learning_rate': 2.2022248064397645e-06, 'epoch': 2.16}

 72%|███████▏  | 9590/13317 [1:38:53<38:22,  1.62it/s]
 72%|███████▏  | 9591/13317 [1:38:54<38:15,  1.62it/s]
                                                      
{'loss': 0.1498, 'grad_norm': 0.4695211946964264, 'learning_rate': 2.201138657009454e-06, 'epoch': 2.16}

 72%|███████▏  | 9591/13317 [1:38:54<38:15,  1.62it/s]
 72%|███████▏  | 9592/13317 [1:38:54<38:14,  1.62it/s]
                                                      
{'loss': 0.2061, 'grad_norm': 0.727699875831604, 'learning_rate': 2.20005269989055e-06, 'epoch': 2.16}

 72%|███████▏  | 9592/13317 [1:38:54<38:14,  1.62it/s]
 72%|███████▏  | 9593/13317 [1:38:55<38:13,  1.62it/s]
                                                      
{'loss': 0.1595, 'grad_norm': 0.5573151111602783, 'learning_rate': 2.1989669351576744e-06, 'epoch': 2.16}

 72%|███████▏  | 9593/13317 [1:38:55<38:13,  1.62it/s]
 72%|███████▏  | 9594/13317 [1:38:55<38:09,  1.63it/s]
                                                      
{'loss': 0.1479, 'grad_norm': 0.5074721574783325, 'learning_rate': 2.197881362885426e-06, 'epoch': 2.16}

 72%|███████▏  | 9594/13317 [1:38:56<38:09,  1.63it/s]
 72%|███████▏  | 9595/13317 [1:38:56<38:09,  1.63it/s]
                                                      
{'loss': 0.1372, 'grad_norm': 0.47631919384002686, 'learning_rate': 2.196795983148399e-06, 'epoch': 2.16}

 72%|███████▏  | 9595/13317 [1:38:56<38:09,  1.63it/s]
 72%|███████▏  | 9596/13317 [1:38:57<38:06,  1.63it/s]
                                                      
{'loss': 0.1496, 'grad_norm': 0.5462974309921265, 'learning_rate': 2.1957107960211692e-06, 'epoch': 2.16}

 72%|███████▏  | 9596/13317 [1:38:57<38:06,  1.63it/s]
 72%|███████▏  | 9597/13317 [1:38:57<38:10,  1.62it/s]
                                                      
{'loss': 0.1341, 'grad_norm': 0.5037379264831543, 'learning_rate': 2.194625801578298e-06, 'epoch': 2.16}

 72%|███████▏  | 9597/13317 [1:38:57<38:10,  1.62it/s]
 72%|███████▏  | 9598/13317 [1:38:58<38:10,  1.62it/s]
                                                      
{'loss': 0.2014, 'grad_norm': 0.6616451740264893, 'learning_rate': 2.1935409998943397e-06, 'epoch': 2.16}

 72%|███████▏  | 9598/13317 [1:38:58<38:10,  1.62it/s]
 72%|███████▏  | 9599/13317 [1:38:59<38:06,  1.63it/s]
                                                      
{'loss': 0.1298, 'grad_norm': 0.5126449465751648, 'learning_rate': 2.1924563910438285e-06, 'epoch': 2.16}

 72%|███████▏  | 9599/13317 [1:38:59<38:06,  1.63it/s]
 72%|███████▏  | 9600/13317 [1:38:59<38:05,  1.63it/s]
                                                      
{'loss': 0.1898, 'grad_norm': 0.6559743285179138, 'learning_rate': 2.191371975101291e-06, 'epoch': 2.16}

 72%|███████▏  | 9600/13317 [1:38:59<38:05,  1.63it/s]
 72%|███████▏  | 9601/13317 [1:39:00<38:04,  1.63it/s]
                                                      
{'loss': 0.1296, 'grad_norm': 0.5293241143226624, 'learning_rate': 2.1902877521412356e-06, 'epoch': 2.16}

 72%|███████▏  | 9601/13317 [1:39:00<38:04,  1.63it/s]
 72%|███████▏  | 9602/13317 [1:39:00<38:03,  1.63it/s]
                                                      
{'loss': 0.1303, 'grad_norm': 0.5213539004325867, 'learning_rate': 2.1892037222381635e-06, 'epoch': 2.16}

 72%|███████▏  | 9602/13317 [1:39:00<38:03,  1.63it/s]
 72%|███████▏  | 9603/13317 [1:39:01<38:02,  1.63it/s]
                                                      
{'loss': 0.1288, 'grad_norm': 0.5282312035560608, 'learning_rate': 2.188119885466557e-06, 'epoch': 2.16}

 72%|███████▏  | 9603/13317 [1:39:01<38:02,  1.63it/s]
 72%|███████▏  | 9604/13317 [1:39:02<38:00,  1.63it/s]
                                                      
{'loss': 0.1391, 'grad_norm': 0.4760388731956482, 'learning_rate': 2.1870362419008844e-06, 'epoch': 2.16}

 72%|███████▏  | 9604/13317 [1:39:02<38:00,  1.63it/s]
 72%|███████▏  | 9605/13317 [1:39:02<38:01,  1.63it/s]
                                                      
{'loss': 0.1601, 'grad_norm': 0.5683869123458862, 'learning_rate': 2.1859527916156086e-06, 'epoch': 2.16}

 72%|███████▏  | 9605/13317 [1:39:02<38:01,  1.63it/s]
 72%|███████▏  | 9606/13317 [1:39:03<37:59,  1.63it/s]
                                                      
{'loss': 0.164, 'grad_norm': 0.6541790962219238, 'learning_rate': 2.1848695346851716e-06, 'epoch': 2.16}

 72%|███████▏  | 9606/13317 [1:39:03<37:59,  1.63it/s]
 72%|███████▏  | 9607/13317 [1:39:03<37:58,  1.63it/s]
                                                      
{'loss': 0.1842, 'grad_norm': 0.6907607913017273, 'learning_rate': 2.183786471184002e-06, 'epoch': 2.16}

 72%|███████▏  | 9607/13317 [1:39:04<37:58,  1.63it/s]
 72%|███████▏  | 9608/13317 [1:39:04<38:00,  1.63it/s]
                                                      
{'loss': 0.1851, 'grad_norm': 0.5999835729598999, 'learning_rate': 2.1827036011865214e-06, 'epoch': 2.16}

 72%|███████▏  | 9608/13317 [1:39:04<38:00,  1.63it/s]
 72%|███████▏  | 9609/13317 [1:39:05<37:58,  1.63it/s]
                                                      
{'loss': 0.1357, 'grad_norm': 0.5360778570175171, 'learning_rate': 2.181620924767135e-06, 'epoch': 2.16}

 72%|███████▏  | 9609/13317 [1:39:05<37:58,  1.63it/s]
 72%|███████▏  | 9610/13317 [1:39:05<38:07,  1.62it/s]
                                                      
{'loss': 0.1683, 'grad_norm': 0.6166915893554688, 'learning_rate': 2.1805384420002325e-06, 'epoch': 2.16}

 72%|███████▏  | 9610/13317 [1:39:05<38:07,  1.62it/s]
 72%|███████▏  | 9611/13317 [1:39:06<38:03,  1.62it/s]
                                                      
{'loss': 0.1363, 'grad_norm': 0.4573975205421448, 'learning_rate': 2.17945615296019e-06, 'epoch': 2.17}

 72%|███████▏  | 9611/13317 [1:39:06<38:03,  1.62it/s]
 72%|███████▏  | 9612/13317 [1:39:07<38:02,  1.62it/s]
                                                      
{'loss': 0.129, 'grad_norm': 0.47311434149742126, 'learning_rate': 2.1783740577213764e-06, 'epoch': 2.17}

 72%|███████▏  | 9612/13317 [1:39:07<38:02,  1.62it/s]
 72%|███████▏  | 9613/13317 [1:39:07<38:00,  1.62it/s]
                                                      
{'loss': 0.1506, 'grad_norm': 0.5829309225082397, 'learning_rate': 2.17729215635814e-06, 'epoch': 2.17}

 72%|███████▏  | 9613/13317 [1:39:07<38:00,  1.62it/s]
 72%|███████▏  | 9614/13317 [1:39:08<37:59,  1.62it/s]
                                                      
{'loss': 0.183, 'grad_norm': 0.5785174369812012, 'learning_rate': 2.176210448944819e-06, 'epoch': 2.17}

 72%|███████▏  | 9614/13317 [1:39:08<37:59,  1.62it/s]
 72%|███████▏  | 9615/13317 [1:39:08<37:59,  1.62it/s]
                                                      
{'loss': 0.1667, 'grad_norm': 0.5961935520172119, 'learning_rate': 2.175128935555738e-06, 'epoch': 2.17}

 72%|███████▏  | 9615/13317 [1:39:08<37:59,  1.62it/s]
 72%|███████▏  | 9616/13317 [1:39:09<37:57,  1.63it/s]
                                                      
{'loss': 0.1599, 'grad_norm': 0.5922106504440308, 'learning_rate': 2.1740476162652117e-06, 'epoch': 2.17}

 72%|███████▏  | 9616/13317 [1:39:09<37:57,  1.63it/s]
 72%|███████▏  | 9617/13317 [1:39:10<37:57,  1.62it/s]
                                                      
{'loss': 0.1503, 'grad_norm': 0.5718266367912292, 'learning_rate': 2.1729664911475357e-06, 'epoch': 2.17}

 72%|███████▏  | 9617/13317 [1:39:10<37:57,  1.62it/s]
 72%|███████▏  | 9618/13317 [1:39:10<37:57,  1.62it/s]
                                                      
{'loss': 0.1561, 'grad_norm': 0.5947166085243225, 'learning_rate': 2.1718855602769934e-06, 'epoch': 2.17}

 72%|███████▏  | 9618/13317 [1:39:10<37:57,  1.62it/s]
 72%|███████▏  | 9619/13317 [1:39:11<37:53,  1.63it/s]
                                                      
{'loss': 0.1447, 'grad_norm': 0.5315902233123779, 'learning_rate': 2.170804823727859e-06, 'epoch': 2.17}

 72%|███████▏  | 9619/13317 [1:39:11<37:53,  1.63it/s]
 72%|███████▏  | 9620/13317 [1:39:11<37:54,  1.63it/s]
                                                      
{'loss': 0.2251, 'grad_norm': 0.633913516998291, 'learning_rate': 2.169724281574389e-06, 'epoch': 2.17}

 72%|███████▏  | 9620/13317 [1:39:12<37:54,  1.63it/s]
 72%|███████▏  | 9621/13317 [1:39:12<37:50,  1.63it/s]
                                                      
{'loss': 0.1422, 'grad_norm': 0.5004397630691528, 'learning_rate': 2.168643933890827e-06, 'epoch': 2.17}

 72%|███████▏  | 9621/13317 [1:39:12<37:50,  1.63it/s]
 72%|███████▏  | 9622/13317 [1:39:13<37:52,  1.63it/s]
                                                      
{'loss': 0.18, 'grad_norm': 0.7092124819755554, 'learning_rate': 2.1675637807514056e-06, 'epoch': 2.17}

 72%|███████▏  | 9622/13317 [1:39:13<37:52,  1.63it/s]
 72%|███████▏  | 9623/13317 [1:39:13<37:51,  1.63it/s]
                                                      
{'loss': 0.1605, 'grad_norm': 0.5747087001800537, 'learning_rate': 2.1664838222303446e-06, 'epoch': 2.17}

 72%|███████▏  | 9623/13317 [1:39:13<37:51,  1.63it/s]
 72%|███████▏  | 9624/13317 [1:39:14<37:50,  1.63it/s]
                                                      
{'loss': 0.1804, 'grad_norm': 0.5966758131980896, 'learning_rate': 2.1654040584018466e-06, 'epoch': 2.17}

 72%|███████▏  | 9624/13317 [1:39:14<37:50,  1.63it/s]
 72%|███████▏  | 9625/13317 [1:39:14<37:47,  1.63it/s]
                                                      
{'loss': 0.1611, 'grad_norm': 0.5760729908943176, 'learning_rate': 2.164324489340101e-06, 'epoch': 2.17}

 72%|███████▏  | 9625/13317 [1:39:15<37:47,  1.63it/s]
 72%|███████▏  | 9626/13317 [1:39:15<37:47,  1.63it/s]
                                                      
{'loss': 0.139, 'grad_norm': 0.5742717981338501, 'learning_rate': 2.163245115119288e-06, 'epoch': 2.17}

 72%|███████▏  | 9626/13317 [1:39:15<37:47,  1.63it/s]
 72%|███████▏  | 9627/13317 [1:39:16<37:48,  1.63it/s]
                                                      
{'loss': 0.1912, 'grad_norm': 0.6967107653617859, 'learning_rate': 2.1621659358135747e-06, 'epoch': 2.17}

 72%|███████▏  | 9627/13317 [1:39:16<37:48,  1.63it/s]
 72%|███████▏  | 9628/13317 [1:39:16<37:56,  1.62it/s]
                                                      
{'loss': 0.1687, 'grad_norm': 0.6530493497848511, 'learning_rate': 2.161086951497106e-06, 'epoch': 2.17}

 72%|███████▏  | 9628/13317 [1:39:16<37:56,  1.62it/s]
 72%|███████▏  | 9629/13317 [1:39:17<37:52,  1.62it/s]
                                                      
{'loss': 0.1493, 'grad_norm': 0.5763394832611084, 'learning_rate': 2.1600081622440224e-06, 'epoch': 2.17}

 72%|███████▏  | 9629/13317 [1:39:17<37:52,  1.62it/s]
 72%|███████▏  | 9630/13317 [1:39:18<37:51,  1.62it/s]
                                                      
{'loss': 0.1532, 'grad_norm': 0.5418107509613037, 'learning_rate': 2.15892956812845e-06, 'epoch': 2.17}

 72%|███████▏  | 9630/13317 [1:39:18<37:51,  1.62it/s]
 72%|███████▏  | 9631/13317 [1:39:18<37:48,  1.62it/s]
                                                      
{'loss': 0.1688, 'grad_norm': 0.5210307240486145, 'learning_rate': 2.1578511692244984e-06, 'epoch': 2.17}

 72%|███████▏  | 9631/13317 [1:39:18<37:48,  1.62it/s]
 72%|███████▏  | 9632/13317 [1:39:19<37:48,  1.62it/s]
                                                      
{'loss': 0.1438, 'grad_norm': 0.6798591017723083, 'learning_rate': 2.1567729656062625e-06, 'epoch': 2.17}

 72%|███████▏  | 9632/13317 [1:39:19<37:48,  1.62it/s]
 72%|███████▏  | 9633/13317 [1:39:19<37:47,  1.62it/s]
                                                      
{'loss': 0.136, 'grad_norm': 0.6007913947105408, 'learning_rate': 2.1556949573478288e-06, 'epoch': 2.17}

 72%|███████▏  | 9633/13317 [1:39:20<37:47,  1.62it/s]
 72%|███████▏  | 9634/13317 [1:39:20<37:44,  1.63it/s]
                                                      
{'loss': 0.1626, 'grad_norm': 0.5722194314002991, 'learning_rate': 2.1546171445232707e-06, 'epoch': 2.17}

 72%|███████▏  | 9634/13317 [1:39:20<37:44,  1.63it/s]
 72%|███████▏  | 9635/13317 [1:39:21<37:42,  1.63it/s]
                                                      
{'loss': 0.1398, 'grad_norm': 0.5252208113670349, 'learning_rate': 2.1535395272066385e-06, 'epoch': 2.17}

 72%|███████▏  | 9635/13317 [1:39:21<37:42,  1.63it/s]
 72%|███████▏  | 9636/13317 [1:39:21<37:41,  1.63it/s]
                                                      
{'loss': 0.1273, 'grad_norm': 0.48416656255722046, 'learning_rate': 2.15246210547198e-06, 'epoch': 2.17}

 72%|███████▏  | 9636/13317 [1:39:21<37:41,  1.63it/s]
 72%|███████▏  | 9637/13317 [1:39:22<37:39,  1.63it/s]
                                                      
{'loss': 0.1623, 'grad_norm': 0.5141414999961853, 'learning_rate': 2.1513848793933267e-06, 'epoch': 2.17}

 72%|███████▏  | 9637/13317 [1:39:22<37:39,  1.63it/s]
 72%|███████▏  | 9638/13317 [1:39:22<37:41,  1.63it/s]
                                                      
{'loss': 0.165, 'grad_norm': 0.6570124626159668, 'learning_rate': 2.1503078490446937e-06, 'epoch': 2.17}

 72%|███████▏  | 9638/13317 [1:39:23<37:41,  1.63it/s]
 72%|███████▏  | 9639/13317 [1:39:23<37:39,  1.63it/s]
                                                      
{'loss': 0.1184, 'grad_norm': 0.43206509947776794, 'learning_rate': 2.149231014500083e-06, 'epoch': 2.17}

 72%|███████▏  | 9639/13317 [1:39:23<37:39,  1.63it/s]
 72%|███████▏  | 9640/13317 [1:39:24<37:36,  1.63it/s]
                                                      
{'loss': 0.1355, 'grad_norm': 0.4816090166568756, 'learning_rate': 2.148154375833487e-06, 'epoch': 2.17}

 72%|███████▏  | 9640/13317 [1:39:24<37:36,  1.63it/s]
 72%|███████▏  | 9641/13317 [1:39:24<37:36,  1.63it/s]
                                                      
{'loss': 0.1813, 'grad_norm': 0.6276383996009827, 'learning_rate': 2.1470779331188824e-06, 'epoch': 2.17}

 72%|███████▏  | 9641/13317 [1:39:24<37:36,  1.63it/s]
 72%|███████▏  | 9642/13317 [1:39:25<37:34,  1.63it/s]
                                                      
{'loss': 0.1893, 'grad_norm': 0.6404988765716553, 'learning_rate': 2.1460016864302286e-06, 'epoch': 2.17}

 72%|███████▏  | 9642/13317 [1:39:25<37:34,  1.63it/s]
 72%|███████▏  | 9643/13317 [1:39:26<37:36,  1.63it/s]
                                                      
{'loss': 0.1677, 'grad_norm': 0.5805224180221558, 'learning_rate': 2.1449256358414776e-06, 'epoch': 2.17}

 72%|███████▏  | 9643/13317 [1:39:26<37:36,  1.63it/s]
 72%|███████▏  | 9644/13317 [1:39:26<37:36,  1.63it/s]
                                                      
{'loss': 0.1731, 'grad_norm': 0.5794992446899414, 'learning_rate': 2.1438497814265672e-06, 'epoch': 2.17}

 72%|███████▏  | 9644/13317 [1:39:26<37:36,  1.63it/s]
 72%|███████▏  | 9645/13317 [1:39:27<37:36,  1.63it/s]
                                                      
{'loss': 0.197, 'grad_norm': 0.54282546043396, 'learning_rate': 2.1427741232594185e-06, 'epoch': 2.17}

 72%|███████▏  | 9645/13317 [1:39:27<37:36,  1.63it/s]
 72%|███████▏  | 9646/13317 [1:39:27<37:34,  1.63it/s]
                                                      
{'loss': 0.1172, 'grad_norm': 0.6286143660545349, 'learning_rate': 2.141698661413938e-06, 'epoch': 2.17}

 72%|███████▏  | 9646/13317 [1:39:28<37:34,  1.63it/s]
 72%|███████▏  | 9647/13317 [1:39:28<37:33,  1.63it/s]
                                                      
{'loss': 0.1622, 'grad_norm': 0.634152889251709, 'learning_rate': 2.1406233959640253e-06, 'epoch': 2.17}

 72%|███████▏  | 9647/13317 [1:39:28<37:33,  1.63it/s]
 72%|███████▏  | 9648/13317 [1:39:29<37:36,  1.63it/s]
                                                      
{'loss': 0.1168, 'grad_norm': 0.4280552566051483, 'learning_rate': 2.13954832698356e-06, 'epoch': 2.17}

 72%|███████▏  | 9648/13317 [1:39:29<37:36,  1.63it/s]
 72%|███████▏  | 9649/13317 [1:39:29<37:33,  1.63it/s]
                                                      
{'loss': 0.1282, 'grad_norm': 0.5297620892524719, 'learning_rate': 2.1384734545464125e-06, 'epoch': 2.17}

 72%|███████▏  | 9649/13317 [1:39:29<37:33,  1.63it/s]
 72%|███████▏  | 9650/13317 [1:39:30<37:32,  1.63it/s]
                                                      
{'loss': 0.1595, 'grad_norm': 0.7536245584487915, 'learning_rate': 2.1373987787264354e-06, 'epoch': 2.17}

 72%|███████▏  | 9650/13317 [1:39:30<37:32,  1.63it/s]
 72%|███████▏  | 9651/13317 [1:39:30<37:31,  1.63it/s]
                                                      
{'loss': 0.1277, 'grad_norm': 0.5300168991088867, 'learning_rate': 2.136324299597474e-06, 'epoch': 2.17}

 72%|███████▏  | 9651/13317 [1:39:31<37:31,  1.63it/s]
 72%|███████▏  | 9652/13317 [1:39:31<37:28,  1.63it/s]
                                                      
{'loss': 0.1557, 'grad_norm': 0.5953387022018433, 'learning_rate': 2.1352500172333546e-06, 'epoch': 2.17}

 72%|███████▏  | 9652/13317 [1:39:31<37:28,  1.63it/s]
 72%|███████▏  | 9653/13317 [1:39:32<37:30,  1.63it/s]
                                                      
{'loss': 0.1327, 'grad_norm': 0.4997396171092987, 'learning_rate': 2.134175931707889e-06, 'epoch': 2.17}

 72%|███████▏  | 9653/13317 [1:39:32<37:30,  1.63it/s]
 72%|███████▏  | 9654/13317 [1:39:32<37:28,  1.63it/s]
                                                      
{'loss': 0.1291, 'grad_norm': 0.5032700300216675, 'learning_rate': 2.1331020430948828e-06, 'epoch': 2.17}

 72%|███████▏  | 9654/13317 [1:39:32<37:28,  1.63it/s]
 73%|███████▎  | 9655/13317 [1:39:33<37:27,  1.63it/s]
                                                      
{'loss': 0.1827, 'grad_norm': 0.7101200819015503, 'learning_rate': 2.1320283514681197e-06, 'epoch': 2.18}

 73%|███████▎  | 9655/13317 [1:39:33<37:27,  1.63it/s]
 73%|███████▎  | 9656/13317 [1:39:34<37:26,  1.63it/s]
                                                      
{'loss': 0.1486, 'grad_norm': 0.5751850008964539, 'learning_rate': 2.130954856901377e-06, 'epoch': 2.18}

 73%|███████▎  | 9656/13317 [1:39:34<37:26,  1.63it/s]
 73%|███████▎  | 9657/13317 [1:39:34<37:26,  1.63it/s]
                                                      
{'loss': 0.1505, 'grad_norm': 0.5838367342948914, 'learning_rate': 2.1298815594684137e-06, 'epoch': 2.18}

 73%|███████▎  | 9657/13317 [1:39:34<37:26,  1.63it/s]
 73%|███████▎  | 9658/13317 [1:39:35<37:27,  1.63it/s]
                                                      
{'loss': 0.1459, 'grad_norm': 0.4756169617176056, 'learning_rate': 2.128808459242975e-06, 'epoch': 2.18}

 73%|███████▎  | 9658/13317 [1:39:35<37:27,  1.63it/s]
 73%|███████▎  | 9659/13317 [1:39:35<37:27,  1.63it/s]
                                                      
{'loss': 0.1694, 'grad_norm': 0.5906301736831665, 'learning_rate': 2.1277355562987974e-06, 'epoch': 2.18}

 73%|███████▎  | 9659/13317 [1:39:36<37:27,  1.63it/s]
 73%|███████▎  | 9660/13317 [1:39:36<37:26,  1.63it/s]
                                                      
{'loss': 0.1578, 'grad_norm': 0.6927595734596252, 'learning_rate': 2.1266628507095972e-06, 'epoch': 2.18}

 73%|███████▎  | 9660/13317 [1:39:36<37:26,  1.63it/s]
 73%|███████▎  | 9661/13317 [1:39:37<37:25,  1.63it/s]
                                                      
{'loss': 0.1649, 'grad_norm': 0.5076952576637268, 'learning_rate': 2.1255903425490844e-06, 'epoch': 2.18}

 73%|███████▎  | 9661/13317 [1:39:37<37:25,  1.63it/s]
 73%|███████▎  | 9662/13317 [1:39:37<37:22,  1.63it/s]
                                                      
{'loss': 0.1374, 'grad_norm': 0.5091433525085449, 'learning_rate': 2.1245180318909482e-06, 'epoch': 2.18}

 73%|███████▎  | 9662/13317 [1:39:37<37:22,  1.63it/s]
 73%|███████▎  | 9663/13317 [1:39:38<37:24,  1.63it/s]
                                                      
{'loss': 0.1401, 'grad_norm': 0.4939037561416626, 'learning_rate': 2.123445918808871e-06, 'epoch': 2.18}

 73%|███████▎  | 9663/13317 [1:39:38<37:24,  1.63it/s]
 73%|███████▎  | 9664/13317 [1:39:38<37:23,  1.63it/s]
                                                      
{'loss': 0.1762, 'grad_norm': 0.5856211185455322, 'learning_rate': 2.122374003376516e-06, 'epoch': 2.18}

 73%|███████▎  | 9664/13317 [1:39:39<37:23,  1.63it/s]
 73%|███████▎  | 9665/13317 [1:39:39<37:20,  1.63it/s]
                                                      
{'loss': 0.1328, 'grad_norm': 0.4680151343345642, 'learning_rate': 2.121302285667535e-06, 'epoch': 2.18}

 73%|███████▎  | 9665/13317 [1:39:39<37:20,  1.63it/s]
 73%|███████▎  | 9666/13317 [1:39:40<37:21,  1.63it/s]
                                                      
{'loss': 0.1069, 'grad_norm': 0.43010255694389343, 'learning_rate': 2.1202307657555684e-06, 'epoch': 2.18}

 73%|███████▎  | 9666/13317 [1:39:40<37:21,  1.63it/s]
 73%|███████▎  | 9667/13317 [1:39:40<37:19,  1.63it/s]
                                                      
{'loss': 0.1276, 'grad_norm': 0.5720033049583435, 'learning_rate': 2.1191594437142393e-06, 'epoch': 2.18}

 73%|███████▎  | 9667/13317 [1:39:40<37:19,  1.63it/s]
 73%|███████▎  | 9668/13317 [1:39:41<37:21,  1.63it/s]
                                                      
{'loss': 0.1861, 'grad_norm': 0.6461279392242432, 'learning_rate': 2.118088319617158e-06, 'epoch': 2.18}

 73%|███████▎  | 9668/13317 [1:39:41<37:21,  1.63it/s]
 73%|███████▎  | 9669/13317 [1:39:42<37:21,  1.63it/s]
                                                      
{'loss': 0.1642, 'grad_norm': 0.635292112827301, 'learning_rate': 2.1170173935379236e-06, 'epoch': 2.18}

 73%|███████▎  | 9669/13317 [1:39:42<37:21,  1.63it/s]
 73%|███████▎  | 9670/13317 [1:39:42<37:19,  1.63it/s]
                                                      
{'loss': 0.1556, 'grad_norm': 0.5882689952850342, 'learning_rate': 2.115946665550121e-06, 'epoch': 2.18}

 73%|███████▎  | 9670/13317 [1:39:42<37:19,  1.63it/s]
 73%|███████▎  | 9671/13317 [1:39:43<37:17,  1.63it/s]
                                                      
{'loss': 0.1854, 'grad_norm': 0.6377984285354614, 'learning_rate': 2.114876135727319e-06, 'epoch': 2.18}

 73%|███████▎  | 9671/13317 [1:39:43<37:17,  1.63it/s]
 73%|███████▎  | 9672/13317 [1:39:43<37:15,  1.63it/s]
                                                      
{'loss': 0.1571, 'grad_norm': 0.5638749599456787, 'learning_rate': 2.1138058041430735e-06, 'epoch': 2.18}

 73%|███████▎  | 9672/13317 [1:39:43<37:15,  1.63it/s]
 73%|███████▎  | 9673/13317 [1:39:44<37:18,  1.63it/s]
                                                      
{'loss': 0.1867, 'grad_norm': 0.655656099319458, 'learning_rate': 2.1127356708709283e-06, 'epoch': 2.18}

 73%|███████▎  | 9673/13317 [1:39:44<37:18,  1.63it/s]
 73%|███████▎  | 9674/13317 [1:39:45<37:16,  1.63it/s]
                                                      
{'loss': 0.1876, 'grad_norm': 0.5818018317222595, 'learning_rate': 2.1116657359844174e-06, 'epoch': 2.18}

 73%|███████▎  | 9674/13317 [1:39:45<37:16,  1.63it/s]
 73%|███████▎  | 9675/13317 [1:39:45<37:14,  1.63it/s]
                                                      
{'loss': 0.1885, 'grad_norm': 0.6687139868736267, 'learning_rate': 2.1105959995570493e-06, 'epoch': 2.18}

 73%|███████▎  | 9675/13317 [1:39:45<37:14,  1.63it/s]
 73%|███████▎  | 9676/13317 [1:39:46<37:13,  1.63it/s]
                                                      
{'loss': 0.1681, 'grad_norm': 0.48028939962387085, 'learning_rate': 2.109526461662329e-06, 'epoch': 2.18}

 73%|███████▎  | 9676/13317 [1:39:46<37:13,  1.63it/s]
 73%|███████▎  | 9677/13317 [1:39:46<37:13,  1.63it/s]
                                                      
{'loss': 0.142, 'grad_norm': 0.5116896629333496, 'learning_rate': 2.1084571223737473e-06, 'epoch': 2.18}

 73%|███████▎  | 9677/13317 [1:39:47<37:13,  1.63it/s]
 73%|███████▎  | 9678/13317 [1:39:47<37:14,  1.63it/s]
                                                      
{'loss': 0.1339, 'grad_norm': 0.5141571760177612, 'learning_rate': 2.1073879817647774e-06, 'epoch': 2.18}

 73%|███████▎  | 9678/13317 [1:39:47<37:14,  1.63it/s]
 73%|███████▎  | 9679/13317 [1:39:48<37:14,  1.63it/s]
                                                      
{'loss': 0.1058, 'grad_norm': 0.4500645399093628, 'learning_rate': 2.106319039908879e-06, 'epoch': 2.18}

 73%|███████▎  | 9679/13317 [1:39:48<37:14,  1.63it/s]
 73%|███████▎  | 9680/13317 [1:39:48<37:12,  1.63it/s]
                                                      
{'loss': 0.1852, 'grad_norm': 0.5815144181251526, 'learning_rate': 2.1052502968795015e-06, 'epoch': 2.18}

 73%|███████▎  | 9680/13317 [1:39:48<37:12,  1.63it/s]
 73%|███████▎  | 9681/13317 [1:39:49<37:13,  1.63it/s]
                                                      
{'loss': 0.1919, 'grad_norm': 0.7950619459152222, 'learning_rate': 2.1041817527500824e-06, 'epoch': 2.18}

 73%|███████▎  | 9681/13317 [1:39:49<37:13,  1.63it/s]
 73%|███████▎  | 9682/13317 [1:39:50<37:13,  1.63it/s]
                                                      
{'loss': 0.179, 'grad_norm': 0.6429020762443542, 'learning_rate': 2.1031134075940346e-06, 'epoch': 2.18}

 73%|███████▎  | 9682/13317 [1:39:50<37:13,  1.63it/s]
 73%|███████▎  | 9683/13317 [1:39:50<37:14,  1.63it/s]
                                                      
{'loss': 0.1791, 'grad_norm': 0.5847098231315613, 'learning_rate': 2.102045261484768e-06, 'epoch': 2.18}

 73%|███████▎  | 9683/13317 [1:39:50<37:14,  1.63it/s]
 73%|███████▎  | 9684/13317 [1:39:51<37:13,  1.63it/s]
                                                      
{'loss': 0.1294, 'grad_norm': 0.5668554306030273, 'learning_rate': 2.100977314495678e-06, 'epoch': 2.18}

 73%|███████▎  | 9684/13317 [1:39:51<37:13,  1.63it/s]
 73%|███████▎  | 9685/13317 [1:39:51<37:12,  1.63it/s]
                                                      
{'loss': 0.1945, 'grad_norm': 0.7442306280136108, 'learning_rate': 2.0999095667001424e-06, 'epoch': 2.18}

 73%|███████▎  | 9685/13317 [1:39:51<37:12,  1.63it/s]
 73%|███████▎  | 9686/13317 [1:39:52<37:16,  1.62it/s]
                                                      
{'loss': 0.1801, 'grad_norm': 0.65650475025177, 'learning_rate': 2.0988420181715236e-06, 'epoch': 2.18}

 73%|███████▎  | 9686/13317 [1:39:52<37:16,  1.62it/s]
 73%|███████▎  | 9687/13317 [1:39:53<37:21,  1.62it/s]
                                                      
{'loss': 0.1646, 'grad_norm': 0.5535340905189514, 'learning_rate': 2.097774668983176e-06, 'epoch': 2.18}

 73%|███████▎  | 9687/13317 [1:39:53<37:21,  1.62it/s]
 73%|███████▎  | 9688/13317 [1:39:53<37:20,  1.62it/s]
                                                      
{'loss': 0.1645, 'grad_norm': 0.6138601303100586, 'learning_rate': 2.0967075192084426e-06, 'epoch': 2.18}

 73%|███████▎  | 9688/13317 [1:39:53<37:20,  1.62it/s]
 73%|███████▎  | 9689/13317 [1:39:54<37:16,  1.62it/s]
                                                      
{'loss': 0.1433, 'grad_norm': 0.5549393892288208, 'learning_rate': 2.095640568920639e-06, 'epoch': 2.18}

 73%|███████▎  | 9689/13317 [1:39:54<37:16,  1.62it/s]
 73%|███████▎  | 9690/13317 [1:39:54<37:15,  1.62it/s]
                                                      
{'loss': 0.236, 'grad_norm': 0.5342336893081665, 'learning_rate': 2.09457381819308e-06, 'epoch': 2.18}

 73%|███████▎  | 9690/13317 [1:39:55<37:15,  1.62it/s]
 73%|███████▎  | 9691/13317 [1:39:55<37:13,  1.62it/s]
                                                      
{'loss': 0.1647, 'grad_norm': 0.5740393400192261, 'learning_rate': 2.093507267099065e-06, 'epoch': 2.18}

 73%|███████▎  | 9691/13317 [1:39:55<37:13,  1.62it/s]
 73%|███████▎  | 9692/13317 [1:39:56<37:14,  1.62it/s]
                                                      
{'loss': 0.1229, 'grad_norm': 0.48150748014450073, 'learning_rate': 2.092440915711875e-06, 'epoch': 2.18}

 73%|███████▎  | 9692/13317 [1:39:56<37:14,  1.62it/s]
 73%|███████▎  | 9693/13317 [1:39:56<37:13,  1.62it/s]
                                                      
{'loss': 0.1399, 'grad_norm': 0.5335787534713745, 'learning_rate': 2.0913747641047775e-06, 'epoch': 2.18}

 73%|███████▎  | 9693/13317 [1:39:56<37:13,  1.62it/s]
 73%|███████▎  | 9694/13317 [1:39:57<37:11,  1.62it/s]
                                                      
{'loss': 0.1664, 'grad_norm': 0.5631445646286011, 'learning_rate': 2.090308812351031e-06, 'epoch': 2.18}

 73%|███████▎  | 9694/13317 [1:39:57<37:11,  1.62it/s]
 73%|███████▎  | 9695/13317 [1:39:58<37:21,  1.62it/s]
                                                      
{'loss': 0.1398, 'grad_norm': 0.48925042152404785, 'learning_rate': 2.0892430605238795e-06, 'epoch': 2.18}

 73%|███████▎  | 9695/13317 [1:39:58<37:21,  1.62it/s]
 73%|███████▎  | 9696/13317 [1:39:58<37:17,  1.62it/s]
                                                      
{'loss': 0.1319, 'grad_norm': 0.49755531549453735, 'learning_rate': 2.0881775086965494e-06, 'epoch': 2.18}

 73%|███████▎  | 9696/13317 [1:39:58<37:17,  1.62it/s]
 73%|███████▎  | 9697/13317 [1:39:59<37:15,  1.62it/s]
                                                      
{'loss': 0.1321, 'grad_norm': 0.47099462151527405, 'learning_rate': 2.087112156942253e-06, 'epoch': 2.18}

 73%|███████▎  | 9697/13317 [1:39:59<37:15,  1.62it/s]
 73%|███████▎  | 9698/13317 [1:39:59<37:14,  1.62it/s]
                                                      
{'loss': 0.145, 'grad_norm': 0.5293557047843933, 'learning_rate': 2.0860470053341957e-06, 'epoch': 2.18}

 73%|███████▎  | 9698/13317 [1:40:00<37:14,  1.62it/s]
 73%|███████▎  | 9699/13317 [1:40:00<37:12,  1.62it/s]
                                                      
{'loss': 0.1271, 'grad_norm': 0.5912387371063232, 'learning_rate': 2.0849820539455627e-06, 'epoch': 2.18}

 73%|███████▎  | 9699/13317 [1:40:00<37:12,  1.62it/s]
 73%|███████▎  | 9700/13317 [1:40:01<37:11,  1.62it/s]
                                                      
{'loss': 0.1989, 'grad_norm': 0.605200469493866, 'learning_rate': 2.0839173028495253e-06, 'epoch': 2.19}

 73%|███████▎  | 9700/13317 [1:40:01<37:11,  1.62it/s]
 73%|███████▎  | 9701/13317 [1:40:01<37:08,  1.62it/s]
                                                      
{'loss': 0.1331, 'grad_norm': 0.4711941182613373, 'learning_rate': 2.082852752119247e-06, 'epoch': 2.19}

 73%|███████▎  | 9701/13317 [1:40:01<37:08,  1.62it/s]
 73%|███████▎  | 9702/13317 [1:40:02<37:09,  1.62it/s]
                                                      
{'loss': 0.1359, 'grad_norm': 0.525636613368988, 'learning_rate': 2.08178840182787e-06, 'epoch': 2.19}

 73%|███████▎  | 9702/13317 [1:40:02<37:09,  1.62it/s]
 73%|███████▎  | 9703/13317 [1:40:02<37:07,  1.62it/s]
                                                      
{'loss': 0.191, 'grad_norm': 0.5582669377326965, 'learning_rate': 2.0807242520485305e-06, 'epoch': 2.19}

 73%|███████▎  | 9703/13317 [1:40:03<37:07,  1.62it/s]
 73%|███████▎  | 9704/13317 [1:40:03<37:04,  1.62it/s]
                                                      
{'loss': 0.1367, 'grad_norm': 0.5218127369880676, 'learning_rate': 2.079660302854343e-06, 'epoch': 2.19}

 73%|███████▎  | 9704/13317 [1:40:03<37:04,  1.62it/s]
 73%|███████▎  | 9705/13317 [1:40:04<37:05,  1.62it/s]
                                                      
{'loss': 0.1481, 'grad_norm': 0.5231180787086487, 'learning_rate': 2.0785965543184144e-06, 'epoch': 2.19}

 73%|███████▎  | 9705/13317 [1:40:04<37:05,  1.62it/s]
 73%|███████▎  | 9706/13317 [1:40:04<37:03,  1.62it/s]
                                                      
{'loss': 0.1745, 'grad_norm': 0.5740811228752136, 'learning_rate': 2.0775330065138357e-06, 'epoch': 2.19}

 73%|███████▎  | 9706/13317 [1:40:04<37:03,  1.62it/s]
 73%|███████▎  | 9707/13317 [1:40:05<37:11,  1.62it/s]
                                                      
{'loss': 0.1511, 'grad_norm': 0.5521074533462524, 'learning_rate': 2.076469659513681e-06, 'epoch': 2.19}

 73%|███████▎  | 9707/13317 [1:40:05<37:11,  1.62it/s]
 73%|███████▎  | 9708/13317 [1:40:06<37:04,  1.62it/s]
                                                      
{'loss': 0.1656, 'grad_norm': 0.5921932458877563, 'learning_rate': 2.0754065133910174e-06, 'epoch': 2.19}

 73%|███████▎  | 9708/13317 [1:40:06<37:04,  1.62it/s]
 73%|███████▎  | 9709/13317 [1:40:06<37:01,  1.62it/s]
                                                      
{'loss': 0.1569, 'grad_norm': 0.6202285289764404, 'learning_rate': 2.0743435682188905e-06, 'epoch': 2.19}

 73%|███████▎  | 9709/13317 [1:40:06<37:01,  1.62it/s]
 73%|███████▎  | 9710/13317 [1:40:07<37:02,  1.62it/s]
                                                      
{'loss': 0.1573, 'grad_norm': 0.5686677694320679, 'learning_rate': 2.07328082407034e-06, 'epoch': 2.19}

 73%|███████▎  | 9710/13317 [1:40:07<37:02,  1.62it/s]
 73%|███████▎  | 9711/13317 [1:40:07<36:59,  1.62it/s]
                                                      
{'loss': 0.1478, 'grad_norm': 0.5483866333961487, 'learning_rate': 2.0722182810183857e-06, 'epoch': 2.19}

 73%|███████▎  | 9711/13317 [1:40:08<36:59,  1.62it/s]
 73%|███████▎  | 9712/13317 [1:40:08<36:58,  1.63it/s]
                                                      
{'loss': 0.1581, 'grad_norm': 0.5140905380249023, 'learning_rate': 2.0711559391360323e-06, 'epoch': 2.19}

 73%|███████▎  | 9712/13317 [1:40:08<36:58,  1.63it/s]
 73%|███████▎  | 9713/13317 [1:40:09<36:58,  1.62it/s]
                                                      
{'loss': 0.1489, 'grad_norm': 0.6175867319107056, 'learning_rate': 2.07009379849628e-06, 'epoch': 2.19}

 73%|███████▎  | 9713/13317 [1:40:09<36:58,  1.62it/s]
 73%|███████▎  | 9714/13317 [1:40:09<36:55,  1.63it/s]
                                                      
{'loss': 0.1459, 'grad_norm': 0.5633026361465454, 'learning_rate': 2.0690318591721036e-06, 'epoch': 2.19}

 73%|███████▎  | 9714/13317 [1:40:09<36:55,  1.63it/s]
 73%|███████▎  | 9715/13317 [1:40:10<36:55,  1.63it/s]
                                                      
{'loss': 0.1081, 'grad_norm': 0.4419114887714386, 'learning_rate': 2.0679701212364745e-06, 'epoch': 2.19}

 73%|███████▎  | 9715/13317 [1:40:10<36:55,  1.63it/s]
 73%|███████▎  | 9716/13317 [1:40:10<36:53,  1.63it/s]
                                                      
{'loss': 0.1351, 'grad_norm': 0.5040185451507568, 'learning_rate': 2.0669085847623405e-06, 'epoch': 2.19}

 73%|███████▎  | 9716/13317 [1:40:11<36:53,  1.63it/s]
 73%|███████▎  | 9717/13317 [1:40:11<36:52,  1.63it/s]
                                                      
{'loss': 0.1273, 'grad_norm': 0.4385397732257843, 'learning_rate': 2.065847249822645e-06, 'epoch': 2.19}

 73%|███████▎  | 9717/13317 [1:40:11<36:52,  1.63it/s]
 73%|███████▎  | 9718/13317 [1:40:12<36:52,  1.63it/s]
                                                      
{'loss': 0.1599, 'grad_norm': 0.6015409231185913, 'learning_rate': 2.0647861164903103e-06, 'epoch': 2.19}

 73%|███████▎  | 9718/13317 [1:40:12<36:52,  1.63it/s]
 73%|███████▎  | 9719/13317 [1:40:12<36:50,  1.63it/s]
                                                      
{'loss': 0.1149, 'grad_norm': 0.4509018361568451, 'learning_rate': 2.063725184838247e-06, 'epoch': 2.19}

 73%|███████▎  | 9719/13317 [1:40:12<36:50,  1.63it/s]
 73%|███████▎  | 9720/13317 [1:40:13<36:49,  1.63it/s]
                                                      
{'loss': 0.141, 'grad_norm': 0.5445764660835266, 'learning_rate': 2.062664454939354e-06, 'epoch': 2.19}

 73%|███████▎  | 9720/13317 [1:40:13<36:49,  1.63it/s]
 73%|███████▎  | 9721/13317 [1:40:14<36:50,  1.63it/s]
                                                      
{'loss': 0.1256, 'grad_norm': 0.4432627558708191, 'learning_rate': 2.061603926866513e-06, 'epoch': 2.19}

 73%|███████▎  | 9721/13317 [1:40:14<36:50,  1.63it/s]
 73%|███████▎  | 9722/13317 [1:40:14<36:50,  1.63it/s]
                                                      
{'loss': 0.1294, 'grad_norm': 0.5200178623199463, 'learning_rate': 2.0605436006925953e-06, 'epoch': 2.19}

 73%|███████▎  | 9722/13317 [1:40:14<36:50,  1.63it/s]
 73%|███████▎  | 9723/13317 [1:40:15<36:49,  1.63it/s]
                                                      
{'loss': 0.1507, 'grad_norm': 0.5754102468490601, 'learning_rate': 2.0594834764904547e-06, 'epoch': 2.19}

 73%|███████▎  | 9723/13317 [1:40:15<36:49,  1.63it/s]
 73%|███████▎  | 9724/13317 [1:40:15<36:48,  1.63it/s]
                                                      
{'loss': 0.1302, 'grad_norm': 0.5116443634033203, 'learning_rate': 2.0584235543329358e-06, 'epoch': 2.19}

 73%|███████▎  | 9724/13317 [1:40:16<36:48,  1.63it/s]
 73%|███████▎  | 9725/13317 [1:40:16<36:48,  1.63it/s]
                                                      
{'loss': 0.1617, 'grad_norm': 0.5298305153846741, 'learning_rate': 2.057363834292864e-06, 'epoch': 2.19}

 73%|███████▎  | 9725/13317 [1:40:16<36:48,  1.63it/s]
 73%|███████▎  | 9726/13317 [1:40:17<36:46,  1.63it/s]
                                                      
{'loss': 0.1226, 'grad_norm': 0.4933488965034485, 'learning_rate': 2.0563043164430525e-06, 'epoch': 2.19}

 73%|███████▎  | 9726/13317 [1:40:17<36:46,  1.63it/s]
 73%|███████▎  | 9727/13317 [1:40:17<36:47,  1.63it/s]
                                                      
{'loss': 0.1731, 'grad_norm': 0.5826593637466431, 'learning_rate': 2.055245000856303e-06, 'epoch': 2.19}

 73%|███████▎  | 9727/13317 [1:40:17<36:47,  1.63it/s]
 73%|███████▎  | 9728/13317 [1:40:18<36:48,  1.62it/s]
                                                      
{'loss': 0.153, 'grad_norm': 0.5140443444252014, 'learning_rate': 2.0541858876054044e-06, 'epoch': 2.19}

 73%|███████▎  | 9728/13317 [1:40:18<36:48,  1.62it/s]
 73%|███████▎  | 9729/13317 [1:40:18<36:46,  1.63it/s]
                                                      
{'loss': 0.1303, 'grad_norm': 0.48593586683273315, 'learning_rate': 2.0531269767631236e-06, 'epoch': 2.19}

 73%|███████▎  | 9729/13317 [1:40:19<36:46,  1.63it/s]
 73%|███████▎  | 9730/13317 [1:40:19<36:47,  1.62it/s]
                                                      
{'loss': 0.1632, 'grad_norm': 0.5986367464065552, 'learning_rate': 2.0520682684022203e-06, 'epoch': 2.19}

 73%|███████▎  | 9730/13317 [1:40:19<36:47,  1.62it/s]
 73%|███████▎  | 9731/13317 [1:40:20<36:44,  1.63it/s]
                                                      
{'loss': 0.1896, 'grad_norm': 0.6735404133796692, 'learning_rate': 2.051009762595443e-06, 'epoch': 2.19}

 73%|███████▎  | 9731/13317 [1:40:20<36:44,  1.63it/s]
 73%|███████▎  | 9732/13317 [1:40:20<36:55,  1.62it/s]
                                                      
{'loss': 0.1521, 'grad_norm': 0.5114848017692566, 'learning_rate': 2.049951459415519e-06, 'epoch': 2.19}

 73%|███████▎  | 9732/13317 [1:40:20<36:55,  1.62it/s]
 73%|███████▎  | 9733/13317 [1:40:21<36:53,  1.62it/s]
                                                      
{'loss': 0.1672, 'grad_norm': 0.5965266227722168, 'learning_rate': 2.0488933589351636e-06, 'epoch': 2.19}

 73%|███████▎  | 9733/13317 [1:40:21<36:53,  1.62it/s]
 73%|███████▎  | 9734/13317 [1:40:22<36:49,  1.62it/s]
                                                      
{'loss': 0.1652, 'grad_norm': 0.5967751145362854, 'learning_rate': 2.0478354612270805e-06, 'epoch': 2.19}

 73%|███████▎  | 9734/13317 [1:40:22<36:49,  1.62it/s]
 73%|███████▎  | 9735/13317 [1:40:22<36:46,  1.62it/s]
                                                      
{'loss': 0.17, 'grad_norm': 0.5805821418762207, 'learning_rate': 2.0467777663639638e-06, 'epoch': 2.19}

 73%|███████▎  | 9735/13317 [1:40:22<36:46,  1.62it/s]
 73%|███████▎  | 9736/13317 [1:40:23<36:44,  1.62it/s]
                                                      
{'loss': 0.1457, 'grad_norm': 0.5074613690376282, 'learning_rate': 2.0457202744184802e-06, 'epoch': 2.19}

 73%|███████▎  | 9736/13317 [1:40:23<36:44,  1.62it/s]
 73%|███████▎  | 9737/13317 [1:40:23<36:46,  1.62it/s]
                                                      
{'loss': 0.167, 'grad_norm': 0.5685589909553528, 'learning_rate': 2.044662985463294e-06, 'epoch': 2.19}

 73%|███████▎  | 9737/13317 [1:40:24<36:46,  1.62it/s]
 73%|███████▎  | 9738/13317 [1:40:24<36:43,  1.62it/s]
                                                      
{'loss': 0.1821, 'grad_norm': 0.6373994946479797, 'learning_rate': 2.0436058995710532e-06, 'epoch': 2.19}

 73%|███████▎  | 9738/13317 [1:40:24<36:43,  1.62it/s]
 73%|███████▎  | 9739/13317 [1:40:25<36:40,  1.63it/s]
                                                      
{'loss': 0.154, 'grad_norm': 0.5770812630653381, 'learning_rate': 2.0425490168143907e-06, 'epoch': 2.19}

 73%|███████▎  | 9739/13317 [1:40:25<36:40,  1.63it/s]
 73%|███████▎  | 9740/13317 [1:40:25<36:40,  1.63it/s]
                                                      
{'loss': 0.1542, 'grad_norm': 0.5192003846168518, 'learning_rate': 2.041492337265923e-06, 'epoch': 2.19}

 73%|███████▎  | 9740/13317 [1:40:25<36:40,  1.63it/s]
 73%|███████▎  | 9741/13317 [1:40:26<36:37,  1.63it/s]
                                                      
{'loss': 0.1495, 'grad_norm': 0.5776494741439819, 'learning_rate': 2.0404358609982567e-06, 'epoch': 2.19}

 73%|███████▎  | 9741/13317 [1:40:26<36:37,  1.63it/s]
 73%|███████▎  | 9742/13317 [1:40:26<36:34,  1.63it/s]
                                                      
{'loss': 0.1316, 'grad_norm': 0.5085136294364929, 'learning_rate': 2.0393795880839867e-06, 'epoch': 2.19}

 73%|███████▎  | 9742/13317 [1:40:27<36:34,  1.63it/s]
 73%|███████▎  | 9743/13317 [1:40:27<36:34,  1.63it/s]
                                                      
{'loss': 0.138, 'grad_norm': 0.5315972566604614, 'learning_rate': 2.0383235185956836e-06, 'epoch': 2.19}

 73%|███████▎  | 9743/13317 [1:40:27<36:34,  1.63it/s]
 73%|███████▎  | 9744/13317 [1:40:28<36:34,  1.63it/s]
                                                      
{'loss': 0.1559, 'grad_norm': 0.5245732665061951, 'learning_rate': 2.0372676526059127e-06, 'epoch': 2.2}

 73%|███████▎  | 9744/13317 [1:40:28<36:34,  1.63it/s]
 73%|███████▎  | 9745/13317 [1:40:28<36:37,  1.63it/s]
                                                      
{'loss': 0.1431, 'grad_norm': 0.5538181662559509, 'learning_rate': 2.0362119901872262e-06, 'epoch': 2.2}

 73%|███████▎  | 9745/13317 [1:40:28<36:37,  1.63it/s]
 73%|███████▎  | 9746/13317 [1:40:29<36:35,  1.63it/s]
                                                      
{'loss': 0.1322, 'grad_norm': 0.48446914553642273, 'learning_rate': 2.035156531412157e-06, 'epoch': 2.2}

 73%|███████▎  | 9746/13317 [1:40:29<36:35,  1.63it/s]
 73%|███████▎  | 9747/13317 [1:40:30<36:35,  1.63it/s]
                                                      
{'loss': 0.1462, 'grad_norm': 0.519862949848175, 'learning_rate': 2.0341012763532243e-06, 'epoch': 2.2}

 73%|███████▎  | 9747/13317 [1:40:30<36:35,  1.63it/s]
 73%|███████▎  | 9748/13317 [1:40:30<36:36,  1.63it/s]
                                                      
{'loss': 0.1637, 'grad_norm': 0.5610224604606628, 'learning_rate': 2.0330462250829374e-06, 'epoch': 2.2}

 73%|███████▎  | 9748/13317 [1:40:30<36:36,  1.63it/s]
 73%|███████▎  | 9749/13317 [1:40:31<36:34,  1.63it/s]
                                                      
{'loss': 0.1156, 'grad_norm': 0.48338189721107483, 'learning_rate': 2.0319913776737916e-06, 'epoch': 2.2}

 73%|███████▎  | 9749/13317 [1:40:31<36:34,  1.63it/s]
 73%|███████▎  | 9750/13317 [1:40:31<36:35,  1.62it/s]
                                                      
{'loss': 0.1572, 'grad_norm': 0.6011286973953247, 'learning_rate': 2.0309367341982633e-06, 'epoch': 2.2}

 73%|███████▎  | 9750/13317 [1:40:32<36:35,  1.62it/s]
 73%|███████▎  | 9751/13317 [1:40:32<37:43,  1.58it/s]
                                                      
{'loss': 0.1816, 'grad_norm': 0.6121290922164917, 'learning_rate': 2.029882294728817e-06, 'epoch': 2.2}

 73%|███████▎  | 9751/13317 [1:40:32<37:43,  1.58it/s]
 73%|███████▎  | 9752/13317 [1:40:33<38:43,  1.53it/s]
                                                      
{'loss': 0.2325, 'grad_norm': 0.7781095504760742, 'learning_rate': 2.0288280593379075e-06, 'epoch': 2.2}

 73%|███████▎  | 9752/13317 [1:40:33<38:43,  1.53it/s]
 73%|███████▎  | 9753/13317 [1:40:33<38:58,  1.52it/s]
                                                      
{'loss': 0.1407, 'grad_norm': 0.549557089805603, 'learning_rate': 2.0277740280979684e-06, 'epoch': 2.2}

 73%|███████▎  | 9753/13317 [1:40:34<38:58,  1.52it/s]
 73%|███████▎  | 9754/13317 [1:40:34<38:54,  1.53it/s]
                                                      
{'loss': 0.1338, 'grad_norm': 0.49380820989608765, 'learning_rate': 2.026720201081423e-06, 'epoch': 2.2}

 73%|███████▎  | 9754/13317 [1:40:34<38:54,  1.53it/s]
 73%|███████▎  | 9755/13317 [1:40:35<38:11,  1.55it/s]
                                                      
{'loss': 0.1077, 'grad_norm': 0.44169068336486816, 'learning_rate': 2.025666578360681e-06, 'epoch': 2.2}

 73%|███████▎  | 9755/13317 [1:40:35<38:11,  1.55it/s]
 73%|███████▎  | 9756/13317 [1:40:35<37:48,  1.57it/s]
                                                      
{'loss': 0.135, 'grad_norm': 0.5000378489494324, 'learning_rate': 2.02461316000814e-06, 'epoch': 2.2}

 73%|███████▎  | 9756/13317 [1:40:35<37:48,  1.57it/s]
 73%|███████▎  | 9757/13317 [1:40:36<37:22,  1.59it/s]
                                                      
{'loss': 0.2284, 'grad_norm': 0.8811785578727722, 'learning_rate': 2.0235599460961786e-06, 'epoch': 2.2}

 73%|███████▎  | 9757/13317 [1:40:36<37:22,  1.59it/s]
 73%|███████▎  | 9758/13317 [1:40:37<37:04,  1.60it/s]
                                                      
{'loss': 0.1351, 'grad_norm': 0.4773678183555603, 'learning_rate': 2.0225069366971618e-06, 'epoch': 2.2}

 73%|███████▎  | 9758/13317 [1:40:37<37:04,  1.60it/s]
 73%|███████▎  | 9759/13317 [1:40:37<36:51,  1.61it/s]
                                                      
{'loss': 0.1689, 'grad_norm': 0.565007209777832, 'learning_rate': 2.0214541318834467e-06, 'epoch': 2.2}

 73%|███████▎  | 9759/13317 [1:40:37<36:51,  1.61it/s]
 73%|███████▎  | 9760/13317 [1:40:38<36:53,  1.61it/s]
                                                      
{'loss': 0.155, 'grad_norm': 0.5955063700675964, 'learning_rate': 2.02040153172737e-06, 'epoch': 2.2}

 73%|███████▎  | 9760/13317 [1:40:38<36:53,  1.61it/s]
 73%|███████▎  | 9761/13317 [1:40:38<36:44,  1.61it/s]
                                                      
{'loss': 0.1559, 'grad_norm': 0.5533097982406616, 'learning_rate': 2.019349136301255e-06, 'epoch': 2.2}

 73%|███████▎  | 9761/13317 [1:40:39<36:44,  1.61it/s]
 73%|███████▎  | 9762/13317 [1:40:39<36:37,  1.62it/s]
                                                      
{'loss': 0.1575, 'grad_norm': 0.5807695984840393, 'learning_rate': 2.018296945677416e-06, 'epoch': 2.2}

 73%|███████▎  | 9762/13317 [1:40:39<36:37,  1.62it/s]
 73%|███████▎  | 9763/13317 [1:40:40<36:33,  1.62it/s]
                                                      
{'loss': 0.1518, 'grad_norm': 0.6427449584007263, 'learning_rate': 2.0172449599281457e-06, 'epoch': 2.2}

 73%|███████▎  | 9763/13317 [1:40:40<36:33,  1.62it/s]
 73%|███████▎  | 9764/13317 [1:40:40<36:29,  1.62it/s]
                                                      
{'loss': 0.1815, 'grad_norm': 0.755397617816925, 'learning_rate': 2.01619317912573e-06, 'epoch': 2.2}

 73%|███████▎  | 9764/13317 [1:40:40<36:29,  1.62it/s]
 73%|███████▎  | 9765/13317 [1:40:41<36:29,  1.62it/s]
                                                      
{'loss': 0.1822, 'grad_norm': 0.6349624991416931, 'learning_rate': 2.015141603342434e-06, 'epoch': 2.2}

 73%|███████▎  | 9765/13317 [1:40:41<36:29,  1.62it/s]
 73%|███████▎  | 9766/13317 [1:40:41<36:26,  1.62it/s]
                                                      
{'loss': 0.134, 'grad_norm': 0.5277715921401978, 'learning_rate': 2.0140902326505163e-06, 'epoch': 2.2}

 73%|███████▎  | 9766/13317 [1:40:42<36:26,  1.62it/s]
 73%|███████▎  | 9767/13317 [1:40:42<36:23,  1.63it/s]
                                                      
{'loss': 0.1281, 'grad_norm': 0.5396840572357178, 'learning_rate': 2.013039067122215e-06, 'epoch': 2.2}

 73%|███████▎  | 9767/13317 [1:40:42<36:23,  1.63it/s]
 73%|███████▎  | 9768/13317 [1:40:43<36:20,  1.63it/s]
                                                      
{'loss': 0.1472, 'grad_norm': 0.5464089512825012, 'learning_rate': 2.011988106829754e-06, 'epoch': 2.2}

 73%|███████▎  | 9768/13317 [1:40:43<36:20,  1.63it/s]
 73%|███████▎  | 9769/13317 [1:40:43<36:19,  1.63it/s]
                                                      
{'loss': 0.125, 'grad_norm': 0.45624494552612305, 'learning_rate': 2.01093735184535e-06, 'epoch': 2.2}

 73%|███████▎  | 9769/13317 [1:40:43<36:19,  1.63it/s]
 73%|███████▎  | 9770/13317 [1:40:44<36:19,  1.63it/s]
                                                      
{'loss': 0.1576, 'grad_norm': 0.5531331896781921, 'learning_rate': 2.0098868022411967e-06, 'epoch': 2.2}

 73%|███████▎  | 9770/13317 [1:40:44<36:19,  1.63it/s]
 73%|███████▎  | 9771/13317 [1:40:45<36:16,  1.63it/s]
                                                      
{'loss': 0.15, 'grad_norm': 0.5333548784255981, 'learning_rate': 2.008836458089481e-06, 'epoch': 2.2}

 73%|███████▎  | 9771/13317 [1:40:45<36:16,  1.63it/s]
 73%|███████▎  | 9772/13317 [1:40:45<36:15,  1.63it/s]
                                                      
{'loss': 0.1353, 'grad_norm': 0.4811815619468689, 'learning_rate': 2.007786319462372e-06, 'epoch': 2.2}

 73%|███████▎  | 9772/13317 [1:40:45<36:15,  1.63it/s]
 73%|███████▎  | 9773/13317 [1:40:46<36:14,  1.63it/s]
                                                      
{'loss': 0.1517, 'grad_norm': 0.5260938405990601, 'learning_rate': 2.0067363864320235e-06, 'epoch': 2.2}

 73%|███████▎  | 9773/13317 [1:40:46<36:14,  1.63it/s]
 73%|███████▎  | 9774/13317 [1:40:46<36:14,  1.63it/s]
                                                      
{'loss': 0.1536, 'grad_norm': 0.5394342541694641, 'learning_rate': 2.0056866590705783e-06, 'epoch': 2.2}

 73%|███████▎  | 9774/13317 [1:40:46<36:14,  1.63it/s]
 73%|███████▎  | 9775/13317 [1:40:47<36:13,  1.63it/s]
                                                      
{'loss': 0.1599, 'grad_norm': 0.5355023145675659, 'learning_rate': 2.0046371374501656e-06, 'epoch': 2.2}

 73%|███████▎  | 9775/13317 [1:40:47<36:13,  1.63it/s]
 73%|███████▎  | 9776/13317 [1:40:48<36:12,  1.63it/s]
                                                      
{'loss': 0.1477, 'grad_norm': 0.5214089155197144, 'learning_rate': 2.0035878216428967e-06, 'epoch': 2.2}

 73%|███████▎  | 9776/13317 [1:40:48<36:12,  1.63it/s]
 73%|███████▎  | 9777/13317 [1:40:48<36:10,  1.63it/s]
                                                      
{'loss': 0.1492, 'grad_norm': 0.5205944776535034, 'learning_rate': 2.0025387117208693e-06, 'epoch': 2.2}

 73%|███████▎  | 9777/13317 [1:40:48<36:10,  1.63it/s]
 73%|███████▎  | 9778/13317 [1:40:49<36:10,  1.63it/s]
                                                      
{'loss': 0.1825, 'grad_norm': 0.5615561008453369, 'learning_rate': 2.0014898077561724e-06, 'epoch': 2.2}

 73%|███████▎  | 9778/13317 [1:40:49<36:10,  1.63it/s]
 73%|███████▎  | 9779/13317 [1:40:49<36:09,  1.63it/s]
                                                      
{'loss': 0.133, 'grad_norm': 0.5436609983444214, 'learning_rate': 2.0004411098208735e-06, 'epoch': 2.2}

 73%|███████▎  | 9779/13317 [1:40:50<36:09,  1.63it/s]
 73%|███████▎  | 9780/13317 [1:40:50<36:09,  1.63it/s]
                                                      
{'loss': 0.1508, 'grad_norm': 0.5268656015396118, 'learning_rate': 1.9993926179870294e-06, 'epoch': 2.2}

 73%|███████▎  | 9780/13317 [1:40:50<36:09,  1.63it/s]
 73%|███████▎  | 9781/13317 [1:40:51<36:08,  1.63it/s]
                                                      
{'loss': 0.1669, 'grad_norm': 0.5590872168540955, 'learning_rate': 1.9983443323266827e-06, 'epoch': 2.2}

 73%|███████▎  | 9781/13317 [1:40:51<36:08,  1.63it/s]
 73%|███████▎  | 9782/13317 [1:40:51<36:07,  1.63it/s]
                                                      
{'loss': 0.1272, 'grad_norm': 0.4628641903400421, 'learning_rate': 1.9972962529118643e-06, 'epoch': 2.2}

 73%|███████▎  | 9782/13317 [1:40:51<36:07,  1.63it/s]
 73%|███████▎  | 9783/13317 [1:40:52<36:06,  1.63it/s]
                                                      
{'loss': 0.1981, 'grad_norm': 0.6345274448394775, 'learning_rate': 1.9962483798145864e-06, 'epoch': 2.2}

 73%|███████▎  | 9783/13317 [1:40:52<36:06,  1.63it/s]
 73%|███████▎  | 9784/13317 [1:40:53<36:06,  1.63it/s]
                                                      
{'loss': 0.1219, 'grad_norm': 0.463902086019516, 'learning_rate': 1.995200713106847e-06, 'epoch': 2.2}

 73%|███████▎  | 9784/13317 [1:40:53<36:06,  1.63it/s]
 73%|███████▎  | 9785/13317 [1:40:53<36:07,  1.63it/s]
                                                      
{'loss': 0.1569, 'grad_norm': 0.5702755451202393, 'learning_rate': 1.9941532528606364e-06, 'epoch': 2.2}

 73%|███████▎  | 9785/13317 [1:40:53<36:07,  1.63it/s]
 73%|███████▎  | 9786/13317 [1:40:54<36:07,  1.63it/s]
                                                      
{'loss': 0.1227, 'grad_norm': 0.527252197265625, 'learning_rate': 1.9931059991479236e-06, 'epoch': 2.2}

 73%|███████▎  | 9786/13317 [1:40:54<36:07,  1.63it/s]
 73%|███████▎  | 9787/13317 [1:40:54<36:06,  1.63it/s]
                                                      
{'loss': 0.1414, 'grad_norm': 0.5819434523582458, 'learning_rate': 1.992058952040664e-06, 'epoch': 2.2}

 73%|███████▎  | 9787/13317 [1:40:54<36:06,  1.63it/s]
 74%|███████▎  | 9788/13317 [1:40:55<36:07,  1.63it/s]
                                                      
{'loss': 0.1355, 'grad_norm': 0.5085768103599548, 'learning_rate': 1.991012111610803e-06, 'epoch': 2.21}

 74%|███████▎  | 9788/13317 [1:40:55<36:07,  1.63it/s]
 74%|███████▎  | 9789/13317 [1:40:56<36:05,  1.63it/s]
                                                      
{'loss': 0.1545, 'grad_norm': 0.5722763538360596, 'learning_rate': 1.989965477930274e-06, 'epoch': 2.21}

 74%|███████▎  | 9789/13317 [1:40:56<36:05,  1.63it/s]
 74%|███████▎  | 9790/13317 [1:40:56<36:03,  1.63it/s]
                                                      
{'loss': 0.1583, 'grad_norm': 0.5580970048904419, 'learning_rate': 1.9889190510709834e-06, 'epoch': 2.21}

 74%|███████▎  | 9790/13317 [1:40:56<36:03,  1.63it/s]
 74%|███████▎  | 9791/13317 [1:40:57<36:03,  1.63it/s]
                                                      
{'loss': 0.2221, 'grad_norm': 0.6976808905601501, 'learning_rate': 1.987872831104835e-06, 'epoch': 2.21}

 74%|███████▎  | 9791/13317 [1:40:57<36:03,  1.63it/s]
 74%|███████▎  | 9792/13317 [1:40:57<36:01,  1.63it/s]
                                                      
{'loss': 0.1214, 'grad_norm': 0.44002965092658997, 'learning_rate': 1.9868268181037186e-06, 'epoch': 2.21}

 74%|███████▎  | 9792/13317 [1:40:58<36:01,  1.63it/s]
 74%|███████▎  | 9793/13317 [1:40:58<36:01,  1.63it/s]
                                                      
{'loss': 0.1606, 'grad_norm': 0.690721333026886, 'learning_rate': 1.9857810121395036e-06, 'epoch': 2.21}

 74%|███████▎  | 9793/13317 [1:40:58<36:01,  1.63it/s]
 74%|███████▎  | 9794/13317 [1:40:59<36:00,  1.63it/s]
                                                      
{'loss': 0.1341, 'grad_norm': 0.516491711139679, 'learning_rate': 1.984735413284047e-06, 'epoch': 2.21}

 74%|███████▎  | 9794/13317 [1:40:59<36:00,  1.63it/s]
 74%|███████▎  | 9795/13317 [1:40:59<35:59,  1.63it/s]
                                                      
{'loss': 0.1479, 'grad_norm': 0.5431176424026489, 'learning_rate': 1.9836900216091927e-06, 'epoch': 2.21}

 74%|███████▎  | 9795/13317 [1:40:59<35:59,  1.63it/s]
 74%|███████▎  | 9796/13317 [1:41:00<36:00,  1.63it/s]
                                                      
{'loss': 0.1482, 'grad_norm': 0.5417271256446838, 'learning_rate': 1.9826448371867735e-06, 'epoch': 2.21}

 74%|███████▎  | 9796/13317 [1:41:00<36:00,  1.63it/s]
 74%|███████▎  | 9797/13317 [1:41:00<35:58,  1.63it/s]
                                                      
{'loss': 0.1586, 'grad_norm': 0.6279704570770264, 'learning_rate': 1.9815998600886015e-06, 'epoch': 2.21}

 74%|███████▎  | 9797/13317 [1:41:01<35:58,  1.63it/s]
 74%|███████▎  | 9798/13317 [1:41:01<35:57,  1.63it/s]
                                                      
{'loss': 0.1177, 'grad_norm': 0.49013176560401917, 'learning_rate': 1.9805550903864775e-06, 'epoch': 2.21}

 74%|███████▎  | 9798/13317 [1:41:01<35:57,  1.63it/s]
 74%|███████▎  | 9799/13317 [1:41:02<35:57,  1.63it/s]
                                                      
{'loss': 0.1277, 'grad_norm': 0.4429129362106323, 'learning_rate': 1.97951052815219e-06, 'epoch': 2.21}

 74%|███████▎  | 9799/13317 [1:41:02<35:57,  1.63it/s]
 74%|███████▎  | 9800/13317 [1:41:02<35:57,  1.63it/s]
                                                      
{'loss': 0.1792, 'grad_norm': 0.6428928971290588, 'learning_rate': 1.9784661734575115e-06, 'epoch': 2.21}

 74%|███████▎  | 9800/13317 [1:41:02<35:57,  1.63it/s]
 74%|███████▎  | 9801/13317 [1:41:03<35:56,  1.63it/s]
                                                      
{'loss': 0.1925, 'grad_norm': 0.5733208656311035, 'learning_rate': 1.977422026374198e-06, 'epoch': 2.21}

 74%|███████▎  | 9801/13317 [1:41:03<35:56,  1.63it/s]
 74%|███████▎  | 9802/13317 [1:41:04<35:56,  1.63it/s]
                                                      
{'loss': 0.1834, 'grad_norm': 0.6162282824516296, 'learning_rate': 1.976378086973994e-06, 'epoch': 2.21}

 74%|███████▎  | 9802/13317 [1:41:04<35:56,  1.63it/s]
 74%|███████▎  | 9803/13317 [1:41:04<35:56,  1.63it/s]
                                                      
{'loss': 0.129, 'grad_norm': 0.4772280156612396, 'learning_rate': 1.975334355328632e-06, 'epoch': 2.21}

 74%|███████▎  | 9803/13317 [1:41:04<35:56,  1.63it/s]
 74%|███████▎  | 9804/13317 [1:41:05<35:54,  1.63it/s]
                                                      
{'loss': 0.1728, 'grad_norm': 0.5916772484779358, 'learning_rate': 1.974290831509826e-06, 'epoch': 2.21}

 74%|███████▎  | 9804/13317 [1:41:05<35:54,  1.63it/s]
 74%|███████▎  | 9805/13317 [1:41:05<35:54,  1.63it/s]
                                                      
{'loss': 0.1682, 'grad_norm': 0.5670992136001587, 'learning_rate': 1.9732475155892737e-06, 'epoch': 2.21}

 74%|███████▎  | 9805/13317 [1:41:06<35:54,  1.63it/s]
 74%|███████▎  | 9806/13317 [1:41:06<35:53,  1.63it/s]
                                                      
{'loss': 0.1296, 'grad_norm': 0.4799472391605377, 'learning_rate': 1.9722044076386674e-06, 'epoch': 2.21}

 74%|███████▎  | 9806/13317 [1:41:06<35:53,  1.63it/s]
 74%|███████▎  | 9807/13317 [1:41:07<35:52,  1.63it/s]
                                                      
{'loss': 0.1582, 'grad_norm': 0.540835976600647, 'learning_rate': 1.971161507729677e-06, 'epoch': 2.21}

 74%|███████▎  | 9807/13317 [1:41:07<35:52,  1.63it/s]
 74%|███████▎  | 9808/13317 [1:41:07<35:53,  1.63it/s]
                                                      
{'loss': 0.1429, 'grad_norm': 0.5069097280502319, 'learning_rate': 1.970118815933959e-06, 'epoch': 2.21}

 74%|███████▎  | 9808/13317 [1:41:07<35:53,  1.63it/s]
 74%|███████▎  | 9809/13317 [1:41:08<35:52,  1.63it/s]
                                                      
{'loss': 0.125, 'grad_norm': 0.4789312481880188, 'learning_rate': 1.96907633232316e-06, 'epoch': 2.21}

 74%|███████▎  | 9809/13317 [1:41:08<35:52,  1.63it/s]
 74%|███████▎  | 9810/13317 [1:41:08<35:51,  1.63it/s]
                                                      
{'loss': 0.1264, 'grad_norm': 0.5104882717132568, 'learning_rate': 1.9680340569689095e-06, 'epoch': 2.21}

 74%|███████▎  | 9810/13317 [1:41:09<35:51,  1.63it/s]
 74%|███████▎  | 9811/13317 [1:41:09<35:50,  1.63it/s]
                                                      
{'loss': 0.1308, 'grad_norm': 0.5165057182312012, 'learning_rate': 1.966991989942823e-06, 'epoch': 2.21}

 74%|███████▎  | 9811/13317 [1:41:09<35:50,  1.63it/s]
 74%|███████▎  | 9812/13317 [1:41:10<35:50,  1.63it/s]
                                                      
{'loss': 0.1612, 'grad_norm': 0.549923300743103, 'learning_rate': 1.9659501313164985e-06, 'epoch': 2.21}

 74%|███████▎  | 9812/13317 [1:41:10<35:50,  1.63it/s]
 74%|███████▎  | 9813/13317 [1:41:10<35:49,  1.63it/s]
                                                      
{'loss': 0.1608, 'grad_norm': 0.6197243928909302, 'learning_rate': 1.9649084811615273e-06, 'epoch': 2.21}

 74%|███████▎  | 9813/13317 [1:41:10<35:49,  1.63it/s]
 74%|███████▎  | 9814/13317 [1:41:11<35:48,  1.63it/s]
                                                      
{'loss': 0.1183, 'grad_norm': 0.5384746789932251, 'learning_rate': 1.963867039549479e-06, 'epoch': 2.21}

 74%|███████▎  | 9814/13317 [1:41:11<35:48,  1.63it/s]
 74%|███████▎  | 9815/13317 [1:41:12<35:50,  1.63it/s]
                                                      
{'loss': 0.157, 'grad_norm': 0.6095770001411438, 'learning_rate': 1.9628258065519113e-06, 'epoch': 2.21}

 74%|███████▎  | 9815/13317 [1:41:12<35:50,  1.63it/s]
 74%|███████▎  | 9816/13317 [1:41:12<35:49,  1.63it/s]
                                                      
{'loss': 0.1594, 'grad_norm': 0.595813512802124, 'learning_rate': 1.961784782240368e-06, 'epoch': 2.21}

 74%|███████▎  | 9816/13317 [1:41:12<35:49,  1.63it/s]
 74%|███████▎  | 9817/13317 [1:41:13<35:49,  1.63it/s]
                                                      
{'loss': 0.1389, 'grad_norm': 0.5048791766166687, 'learning_rate': 1.960743966686381e-06, 'epoch': 2.21}

 74%|███████▎  | 9817/13317 [1:41:13<35:49,  1.63it/s]
 74%|███████▎  | 9818/13317 [1:41:13<35:48,  1.63it/s]
                                                      
{'loss': 0.2044, 'grad_norm': 0.7061919569969177, 'learning_rate': 1.959703359961464e-06, 'epoch': 2.21}

 74%|███████▎  | 9818/13317 [1:41:13<35:48,  1.63it/s]
 74%|███████▎  | 9819/13317 [1:41:14<35:47,  1.63it/s]
                                                      
{'loss': 0.1206, 'grad_norm': 0.4922713041305542, 'learning_rate': 1.9586629621371157e-06, 'epoch': 2.21}

 74%|███████▎  | 9819/13317 [1:41:14<35:47,  1.63it/s]
 74%|███████▎  | 9820/13317 [1:41:15<35:46,  1.63it/s]
                                                      
{'loss': 0.1396, 'grad_norm': 0.5478674173355103, 'learning_rate': 1.957622773284826e-06, 'epoch': 2.21}

 74%|███████▎  | 9820/13317 [1:41:15<35:46,  1.63it/s]
 74%|███████▎  | 9821/13317 [1:41:15<35:46,  1.63it/s]
                                                      
{'loss': 0.1212, 'grad_norm': 0.4676460027694702, 'learning_rate': 1.956582793476064e-06, 'epoch': 2.21}

 74%|███████▎  | 9821/13317 [1:41:15<35:46,  1.63it/s]
 74%|███████▍  | 9822/13317 [1:41:16<35:46,  1.63it/s]
                                                      
{'loss': 0.1319, 'grad_norm': 0.5311133861541748, 'learning_rate': 1.9555430227822876e-06, 'epoch': 2.21}

 74%|███████▍  | 9822/13317 [1:41:16<35:46,  1.63it/s]
 74%|███████▍  | 9823/13317 [1:41:16<35:45,  1.63it/s]
                                                      
{'loss': 0.1349, 'grad_norm': 0.5834656953811646, 'learning_rate': 1.954503461274942e-06, 'epoch': 2.21}

 74%|███████▍  | 9823/13317 [1:41:17<35:45,  1.63it/s]
 74%|███████▍  | 9824/13317 [1:41:17<35:44,  1.63it/s]
                                                      
{'loss': 0.1541, 'grad_norm': 0.5657472610473633, 'learning_rate': 1.953464109025453e-06, 'epoch': 2.21}

 74%|███████▍  | 9824/13317 [1:41:17<35:44,  1.63it/s]
 74%|███████▍  | 9825/13317 [1:41:18<35:43,  1.63it/s]
                                                      
{'loss': 0.142, 'grad_norm': 0.6077244281768799, 'learning_rate': 1.952424966105239e-06, 'epoch': 2.21}

 74%|███████▍  | 9825/13317 [1:41:18<35:43,  1.63it/s]
 74%|███████▍  | 9826/13317 [1:41:18<35:42,  1.63it/s]
                                                      
{'loss': 0.1625, 'grad_norm': 0.6047590374946594, 'learning_rate': 1.951386032585696e-06, 'epoch': 2.21}

 74%|███████▍  | 9826/13317 [1:41:18<35:42,  1.63it/s]
 74%|███████▍  | 9827/13317 [1:41:19<35:41,  1.63it/s]
                                                      
{'loss': 0.1293, 'grad_norm': 0.509216845035553, 'learning_rate': 1.950347308538214e-06, 'epoch': 2.21}

 74%|███████▍  | 9827/13317 [1:41:19<35:41,  1.63it/s]
 74%|███████▍  | 9828/13317 [1:41:20<35:41,  1.63it/s]
                                                      
{'loss': 0.1198, 'grad_norm': 0.43687349557876587, 'learning_rate': 1.94930879403416e-06, 'epoch': 2.21}

 74%|███████▍  | 9828/13317 [1:41:20<35:41,  1.63it/s]
 74%|███████▍  | 9829/13317 [1:41:20<35:42,  1.63it/s]
                                                      
{'loss': 0.1704, 'grad_norm': 0.5905942916870117, 'learning_rate': 1.948270489144895e-06, 'epoch': 2.21}

 74%|███████▍  | 9829/13317 [1:41:20<35:42,  1.63it/s]
 74%|███████▍  | 9830/13317 [1:41:21<35:45,  1.63it/s]
                                                      
{'loss': 0.1595, 'grad_norm': 0.5546557903289795, 'learning_rate': 1.94723239394176e-06, 'epoch': 2.21}

 74%|███████▍  | 9830/13317 [1:41:21<35:45,  1.63it/s]
 74%|███████▍  | 9831/13317 [1:41:21<35:43,  1.63it/s]
                                                      
{'loss': 0.3012, 'grad_norm': 0.5216507315635681, 'learning_rate': 1.94619450849608e-06, 'epoch': 2.21}

 74%|███████▍  | 9831/13317 [1:41:21<35:43,  1.63it/s]
 74%|███████▍  | 9832/13317 [1:41:22<35:41,  1.63it/s]
                                                      
{'loss': 0.1221, 'grad_norm': 0.5543763637542725, 'learning_rate': 1.945156832879174e-06, 'epoch': 2.21}

 74%|███████▍  | 9832/13317 [1:41:22<35:41,  1.63it/s]
 74%|███████▍  | 9833/13317 [1:41:23<35:39,  1.63it/s]
                                                      
{'loss': 0.1272, 'grad_norm': 0.5411708354949951, 'learning_rate': 1.9441193671623387e-06, 'epoch': 2.22}

 74%|███████▍  | 9833/13317 [1:41:23<35:39,  1.63it/s]
 74%|███████▍  | 9834/13317 [1:41:23<35:38,  1.63it/s]
                                                      
{'loss': 0.1756, 'grad_norm': 0.7214133143424988, 'learning_rate': 1.9430821114168576e-06, 'epoch': 2.22}

 74%|███████▍  | 9834/13317 [1:41:23<35:38,  1.63it/s]
 74%|███████▍  | 9835/13317 [1:41:24<35:37,  1.63it/s]
                                                      
{'loss': 0.1557, 'grad_norm': 0.5264822244644165, 'learning_rate': 1.9420450657140023e-06, 'epoch': 2.22}

 74%|███████▍  | 9835/13317 [1:41:24<35:37,  1.63it/s]
 74%|███████▍  | 9836/13317 [1:41:24<35:36,  1.63it/s]
                                                      
{'loss': 0.1562, 'grad_norm': 0.50702965259552, 'learning_rate': 1.9410082301250307e-06, 'epoch': 2.22}

 74%|███████▍  | 9836/13317 [1:41:25<35:36,  1.63it/s]
 74%|███████▍  | 9837/13317 [1:41:25<35:35,  1.63it/s]
                                                      
{'loss': 0.1184, 'grad_norm': 0.4652031362056732, 'learning_rate': 1.939971604721183e-06, 'epoch': 2.22}

 74%|███████▍  | 9837/13317 [1:41:25<35:35,  1.63it/s]
 74%|███████▍  | 9838/13317 [1:41:26<35:34,  1.63it/s]
                                                      
{'loss': 0.1816, 'grad_norm': 0.6482715010643005, 'learning_rate': 1.9389351895736842e-06, 'epoch': 2.22}

 74%|███████▍  | 9838/13317 [1:41:26<35:34,  1.63it/s]
 74%|███████▍  | 9839/13317 [1:41:26<35:33,  1.63it/s]
                                                      
{'loss': 0.1405, 'grad_norm': 0.5434179306030273, 'learning_rate': 1.937898984753751e-06, 'epoch': 2.22}

 74%|███████▍  | 9839/13317 [1:41:26<35:33,  1.63it/s]
 74%|███████▍  | 9840/13317 [1:41:27<35:33,  1.63it/s]
                                                      
{'loss': 0.1635, 'grad_norm': 0.5767742395401001, 'learning_rate': 1.9368629903325793e-06, 'epoch': 2.22}

 74%|███████▍  | 9840/13317 [1:41:27<35:33,  1.63it/s]
 74%|███████▍  | 9841/13317 [1:41:27<35:32,  1.63it/s]
                                                      
{'loss': 0.1358, 'grad_norm': 0.4837337136268616, 'learning_rate': 1.9358272063813516e-06, 'epoch': 2.22}

 74%|███████▍  | 9841/13317 [1:41:28<35:32,  1.63it/s]
 74%|███████▍  | 9842/13317 [1:41:28<35:32,  1.63it/s]
                                                      
{'loss': 0.1499, 'grad_norm': 0.5374407172203064, 'learning_rate': 1.934791632971239e-06, 'epoch': 2.22}

 74%|███████▍  | 9842/13317 [1:41:28<35:32,  1.63it/s]
 74%|███████▍  | 9843/13317 [1:41:29<35:31,  1.63it/s]
                                                      
{'loss': 0.124, 'grad_norm': 0.4963296353816986, 'learning_rate': 1.9337562701733974e-06, 'epoch': 2.22}

 74%|███████▍  | 9843/13317 [1:41:29<35:31,  1.63it/s]
 74%|███████▍  | 9844/13317 [1:41:29<35:31,  1.63it/s]
                                                      
{'loss': 0.1656, 'grad_norm': 0.6890973448753357, 'learning_rate': 1.932721118058966e-06, 'epoch': 2.22}

 74%|███████▍  | 9844/13317 [1:41:29<35:31,  1.63it/s]
 74%|███████▍  | 9845/13317 [1:41:30<35:30,  1.63it/s]
                                                      
{'loss': 0.1287, 'grad_norm': 0.48914992809295654, 'learning_rate': 1.9316861766990697e-06, 'epoch': 2.22}

 74%|███████▍  | 9845/13317 [1:41:30<35:30,  1.63it/s]
 74%|███████▍  | 9846/13317 [1:41:31<35:30,  1.63it/s]
                                                      
{'loss': 0.1328, 'grad_norm': 0.5534414052963257, 'learning_rate': 1.930651446164822e-06, 'epoch': 2.22}

 74%|███████▍  | 9846/13317 [1:41:31<35:30,  1.63it/s]
 74%|███████▍  | 9847/13317 [1:41:31<35:29,  1.63it/s]
                                                      
{'loss': 0.1167, 'grad_norm': 0.47280412912368774, 'learning_rate': 1.9296169265273186e-06, 'epoch': 2.22}

 74%|███████▍  | 9847/13317 [1:41:31<35:29,  1.63it/s]
 74%|███████▍  | 9848/13317 [1:41:32<35:28,  1.63it/s]
                                                      
{'loss': 0.1625, 'grad_norm': 0.59489506483078, 'learning_rate': 1.9285826178576408e-06, 'epoch': 2.22}

 74%|███████▍  | 9848/13317 [1:41:32<35:28,  1.63it/s]
 74%|███████▍  | 9849/13317 [1:41:32<35:27,  1.63it/s]
                                                      
{'loss': 0.1564, 'grad_norm': 0.519128680229187, 'learning_rate': 1.9275485202268574e-06, 'epoch': 2.22}

 74%|███████▍  | 9849/13317 [1:41:33<35:27,  1.63it/s]
 74%|███████▍  | 9850/13317 [1:41:33<35:25,  1.63it/s]
                                                      
{'loss': 0.1445, 'grad_norm': 0.5348256826400757, 'learning_rate': 1.926514633706024e-06, 'epoch': 2.22}

 74%|███████▍  | 9850/13317 [1:41:33<35:25,  1.63it/s]
 74%|███████▍  | 9851/13317 [1:41:34<35:26,  1.63it/s]
                                                      
{'loss': 0.1212, 'grad_norm': 0.43375563621520996, 'learning_rate': 1.925480958366178e-06, 'epoch': 2.22}

 74%|███████▍  | 9851/13317 [1:41:34<35:26,  1.63it/s]
 74%|███████▍  | 9852/13317 [1:41:34<35:24,  1.63it/s]
                                                      
{'loss': 0.1366, 'grad_norm': 0.5101565718650818, 'learning_rate': 1.9244474942783425e-06, 'epoch': 2.22}

 74%|███████▍  | 9852/13317 [1:41:34<35:24,  1.63it/s]
 74%|███████▍  | 9853/13317 [1:41:35<35:23,  1.63it/s]
                                                      
{'loss': 0.1427, 'grad_norm': 0.5617245435714722, 'learning_rate': 1.923414241513528e-06, 'epoch': 2.22}

 74%|███████▍  | 9853/13317 [1:41:35<35:23,  1.63it/s]
 74%|███████▍  | 9854/13317 [1:41:35<35:24,  1.63it/s]
                                                      
{'loss': 0.1589, 'grad_norm': 0.5468166470527649, 'learning_rate': 1.9223812001427344e-06, 'epoch': 2.22}

 74%|███████▍  | 9854/13317 [1:41:36<35:24,  1.63it/s]
 74%|███████▍  | 9855/13317 [1:41:36<35:24,  1.63it/s]
                                                      
{'loss': 0.1422, 'grad_norm': 0.5320146679878235, 'learning_rate': 1.921348370236935e-06, 'epoch': 2.22}

 74%|███████▍  | 9855/13317 [1:41:36<35:24,  1.63it/s]
 74%|███████▍  | 9856/13317 [1:41:37<35:23,  1.63it/s]
                                                      
{'loss': 0.1265, 'grad_norm': 0.4918132424354553, 'learning_rate': 1.9203157518670997e-06, 'epoch': 2.22}

 74%|███████▍  | 9856/13317 [1:41:37<35:23,  1.63it/s]
 74%|███████▍  | 9857/13317 [1:41:37<35:21,  1.63it/s]
                                                      
{'loss': 0.1292, 'grad_norm': 0.47974470257759094, 'learning_rate': 1.919283345104182e-06, 'epoch': 2.22}

 74%|███████▍  | 9857/13317 [1:41:37<35:21,  1.63it/s]
 74%|███████▍  | 9858/13317 [1:41:38<35:22,  1.63it/s]
                                                      
{'loss': 0.1349, 'grad_norm': 0.5400959253311157, 'learning_rate': 1.918251150019118e-06, 'epoch': 2.22}

 74%|███████▍  | 9858/13317 [1:41:38<35:22,  1.63it/s]
 74%|███████▍  | 9859/13317 [1:41:39<35:22,  1.63it/s]
                                                      
{'loss': 0.1369, 'grad_norm': 0.5054704546928406, 'learning_rate': 1.9172191666828287e-06, 'epoch': 2.22}

 74%|███████▍  | 9859/13317 [1:41:39<35:22,  1.63it/s]
 74%|███████▍  | 9860/13317 [1:41:39<35:20,  1.63it/s]
                                                      
{'loss': 0.1594, 'grad_norm': 0.6195840239524841, 'learning_rate': 1.916187395166223e-06, 'epoch': 2.22}

 74%|███████▍  | 9860/13317 [1:41:39<35:20,  1.63it/s]
 74%|███████▍  | 9861/13317 [1:41:40<35:20,  1.63it/s]
                                                      
{'loss': 0.1765, 'grad_norm': 0.5924485921859741, 'learning_rate': 1.915155835540199e-06, 'epoch': 2.22}

 74%|███████▍  | 9861/13317 [1:41:40<35:20,  1.63it/s]
 74%|███████▍  | 9862/13317 [1:41:40<35:19,  1.63it/s]
                                                      
{'loss': 0.1349, 'grad_norm': 0.5380464196205139, 'learning_rate': 1.9141244878756284e-06, 'epoch': 2.22}

 74%|███████▍  | 9862/13317 [1:41:41<35:19,  1.63it/s]
 74%|███████▍  | 9863/13317 [1:41:41<35:18,  1.63it/s]
                                                      
{'loss': 0.1674, 'grad_norm': 0.6490223407745361, 'learning_rate': 1.9130933522433792e-06, 'epoch': 2.22}

 74%|███████▍  | 9863/13317 [1:41:41<35:18,  1.63it/s]
 74%|███████▍  | 9864/13317 [1:41:42<35:17,  1.63it/s]
                                                      
{'loss': 0.1515, 'grad_norm': 0.6233896613121033, 'learning_rate': 1.912062428714303e-06, 'epoch': 2.22}

 74%|███████▍  | 9864/13317 [1:41:42<35:17,  1.63it/s]
 74%|███████▍  | 9865/13317 [1:41:42<35:16,  1.63it/s]
                                                      
{'loss': 0.1722, 'grad_norm': 0.6235622763633728, 'learning_rate': 1.911031717359234e-06, 'epoch': 2.22}

 74%|███████▍  | 9865/13317 [1:41:42<35:16,  1.63it/s]
 74%|███████▍  | 9866/13317 [1:41:43<35:15,  1.63it/s]
                                                      
{'loss': 0.275, 'grad_norm': 0.8203191161155701, 'learning_rate': 1.9100012182489907e-06, 'epoch': 2.22}

 74%|███████▍  | 9866/13317 [1:41:43<35:15,  1.63it/s]
 74%|███████▍  | 9867/13317 [1:41:43<35:15,  1.63it/s]
                                                      
{'loss': 0.1363, 'grad_norm': 0.5068349242210388, 'learning_rate': 1.9089709314543823e-06, 'epoch': 2.22}

 74%|███████▍  | 9867/13317 [1:41:44<35:15,  1.63it/s]
 74%|███████▍  | 9868/13317 [1:41:44<35:14,  1.63it/s]
                                                      
{'loss': 0.1428, 'grad_norm': 0.6022433638572693, 'learning_rate': 1.9079408570462e-06, 'epoch': 2.22}

 74%|███████▍  | 9868/13317 [1:41:44<35:14,  1.63it/s]
 74%|███████▍  | 9869/13317 [1:41:45<35:14,  1.63it/s]
                                                      
{'loss': 0.1553, 'grad_norm': 0.5959136486053467, 'learning_rate': 1.906910995095218e-06, 'epoch': 2.22}

 74%|███████▍  | 9869/13317 [1:41:45<35:14,  1.63it/s]
 74%|███████▍  | 9870/13317 [1:41:45<35:13,  1.63it/s]
                                                      
{'loss': 0.1469, 'grad_norm': 0.5877795815467834, 'learning_rate': 1.905881345672201e-06, 'epoch': 2.22}

 74%|███████▍  | 9870/13317 [1:41:45<35:13,  1.63it/s]
 74%|███████▍  | 9871/13317 [1:41:46<35:16,  1.63it/s]
                                                      
{'loss': 0.141, 'grad_norm': 0.5009265542030334, 'learning_rate': 1.9048519088478978e-06, 'epoch': 2.22}

 74%|███████▍  | 9871/13317 [1:41:46<35:16,  1.63it/s]
 74%|███████▍  | 9872/13317 [1:41:47<35:15,  1.63it/s]
                                                      
{'loss': 0.1525, 'grad_norm': 0.5494838356971741, 'learning_rate': 1.9038226846930407e-06, 'epoch': 2.22}

 74%|███████▍  | 9872/13317 [1:41:47<35:15,  1.63it/s]
 74%|███████▍  | 9873/13317 [1:41:47<35:15,  1.63it/s]
                                                      
{'loss': 0.1475, 'grad_norm': 0.539676308631897, 'learning_rate': 1.9027936732783465e-06, 'epoch': 2.22}

 74%|███████▍  | 9873/13317 [1:41:47<35:15,  1.63it/s]
 74%|███████▍  | 9874/13317 [1:41:48<35:14,  1.63it/s]
                                                      
{'loss': 0.1349, 'grad_norm': 0.5025632977485657, 'learning_rate': 1.901764874674522e-06, 'epoch': 2.22}

 74%|███████▍  | 9874/13317 [1:41:48<35:14,  1.63it/s]
 74%|███████▍  | 9875/13317 [1:41:48<35:11,  1.63it/s]
                                                      
{'loss': 0.1577, 'grad_norm': 0.5416219830513, 'learning_rate': 1.9007362889522541e-06, 'epoch': 2.22}

 74%|███████▍  | 9875/13317 [1:41:48<35:11,  1.63it/s]
 74%|███████▍  | 9876/13317 [1:41:49<35:11,  1.63it/s]
                                                      
{'loss': 0.1068, 'grad_norm': 0.43358585238456726, 'learning_rate': 1.8997079161822202e-06, 'epoch': 2.22}

 74%|███████▍  | 9876/13317 [1:41:49<35:11,  1.63it/s]
 74%|███████▍  | 9877/13317 [1:41:50<35:11,  1.63it/s]
                                                      
{'loss': 0.159, 'grad_norm': 0.5713628530502319, 'learning_rate': 1.8986797564350784e-06, 'epoch': 2.23}

 74%|███████▍  | 9877/13317 [1:41:50<35:11,  1.63it/s]
 74%|███████▍  | 9878/13317 [1:41:50<35:10,  1.63it/s]
                                                      
{'loss': 0.1896, 'grad_norm': 0.5571338534355164, 'learning_rate': 1.8976518097814733e-06, 'epoch': 2.23}

 74%|███████▍  | 9878/13317 [1:41:50<35:10,  1.63it/s]
 74%|███████▍  | 9879/13317 [1:41:51<35:09,  1.63it/s]
                                                      
{'loss': 0.1716, 'grad_norm': 0.6145184636116028, 'learning_rate': 1.896624076292039e-06, 'epoch': 2.23}

 74%|███████▍  | 9879/13317 [1:41:51<35:09,  1.63it/s]
 74%|███████▍  | 9880/13317 [1:41:51<35:09,  1.63it/s]
                                                      
{'loss': 0.1549, 'grad_norm': 0.6941574215888977, 'learning_rate': 1.8955965560373873e-06, 'epoch': 2.23}

 74%|███████▍  | 9880/13317 [1:41:52<35:09,  1.63it/s]
 74%|███████▍  | 9881/13317 [1:41:52<35:09,  1.63it/s]
                                                      
{'loss': 0.1667, 'grad_norm': 0.6091398596763611, 'learning_rate': 1.8945692490881247e-06, 'epoch': 2.23}

 74%|███████▍  | 9881/13317 [1:41:52<35:09,  1.63it/s]
 74%|███████▍  | 9882/13317 [1:41:53<35:06,  1.63it/s]
                                                      
{'loss': 0.1464, 'grad_norm': 0.5788701176643372, 'learning_rate': 1.893542155514833e-06, 'epoch': 2.23}

 74%|███████▍  | 9882/13317 [1:41:53<35:06,  1.63it/s]
 74%|███████▍  | 9883/13317 [1:41:53<35:07,  1.63it/s]
                                                      
{'loss': 0.1239, 'grad_norm': 0.48600485920906067, 'learning_rate': 1.8925152753880892e-06, 'epoch': 2.23}

 74%|███████▍  | 9883/13317 [1:41:53<35:07,  1.63it/s]
 74%|███████▍  | 9884/13317 [1:41:54<35:06,  1.63it/s]
                                                      
{'loss': 0.1368, 'grad_norm': 0.5815292596817017, 'learning_rate': 1.8914886087784485e-06, 'epoch': 2.23}

 74%|███████▍  | 9884/13317 [1:41:54<35:06,  1.63it/s]
 74%|███████▍  | 9885/13317 [1:41:54<35:06,  1.63it/s]
                                                      
{'loss': 0.1461, 'grad_norm': 0.5738659501075745, 'learning_rate': 1.8904621557564517e-06, 'epoch': 2.23}

 74%|███████▍  | 9885/13317 [1:41:55<35:06,  1.63it/s]
 74%|███████▍  | 9886/13317 [1:41:55<35:06,  1.63it/s]
                                                      
{'loss': 0.1326, 'grad_norm': 0.5121580362319946, 'learning_rate': 1.8894359163926312e-06, 'epoch': 2.23}

 74%|███████▍  | 9886/13317 [1:41:55<35:06,  1.63it/s]
 74%|███████▍  | 9887/13317 [1:41:56<35:04,  1.63it/s]
                                                      
{'loss': 0.1559, 'grad_norm': 0.5678399801254272, 'learning_rate': 1.8884098907574972e-06, 'epoch': 2.23}

 74%|███████▍  | 9887/13317 [1:41:56<35:04,  1.63it/s]
 74%|███████▍  | 9888/13317 [1:41:56<35:04,  1.63it/s]
                                                      
{'loss': 0.1253, 'grad_norm': 0.5353760719299316, 'learning_rate': 1.8873840789215509e-06, 'epoch': 2.23}

 74%|███████▍  | 9888/13317 [1:41:56<35:04,  1.63it/s]
 74%|███████▍  | 9889/13317 [1:41:57<35:03,  1.63it/s]
                                                      
{'loss': 0.1447, 'grad_norm': 0.5018065571784973, 'learning_rate': 1.8863584809552732e-06, 'epoch': 2.23}

 74%|███████▍  | 9889/13317 [1:41:57<35:03,  1.63it/s]
 74%|███████▍  | 9890/13317 [1:41:58<35:03,  1.63it/s]
                                                      
{'loss': 0.1442, 'grad_norm': 0.5259147882461548, 'learning_rate': 1.8853330969291377e-06, 'epoch': 2.23}

 74%|███████▍  | 9890/13317 [1:41:58<35:03,  1.63it/s]
 74%|███████▍  | 9891/13317 [1:41:58<35:01,  1.63it/s]
                                                      
{'loss': 0.2179, 'grad_norm': 0.663378894329071, 'learning_rate': 1.8843079269135973e-06, 'epoch': 2.23}

 74%|███████▍  | 9891/13317 [1:41:58<35:01,  1.63it/s]
 74%|███████▍  | 9892/13317 [1:41:59<35:01,  1.63it/s]
                                                      
{'loss': 0.1735, 'grad_norm': 0.696153998374939, 'learning_rate': 1.8832829709790901e-06, 'epoch': 2.23}

 74%|███████▍  | 9892/13317 [1:41:59<35:01,  1.63it/s]
 74%|███████▍  | 9893/13317 [1:41:59<35:01,  1.63it/s]
                                                      
{'loss': 0.1638, 'grad_norm': 0.5899654030799866, 'learning_rate': 1.8822582291960445e-06, 'epoch': 2.23}

 74%|███████▍  | 9893/13317 [1:42:00<35:01,  1.63it/s]
 74%|███████▍  | 9894/13317 [1:42:00<35:00,  1.63it/s]
                                                      
{'loss': 0.2064, 'grad_norm': 0.6668601036071777, 'learning_rate': 1.8812337016348697e-06, 'epoch': 2.23}

 74%|███████▍  | 9894/13317 [1:42:00<35:00,  1.63it/s]
 74%|███████▍  | 9895/13317 [1:42:01<35:00,  1.63it/s]
                                                      
{'loss': 0.1452, 'grad_norm': 0.5474463105201721, 'learning_rate': 1.88020938836596e-06, 'epoch': 2.23}

 74%|███████▍  | 9895/13317 [1:42:01<35:00,  1.63it/s]
 74%|███████▍  | 9896/13317 [1:42:01<34:59,  1.63it/s]
                                                      
{'loss': 0.1815, 'grad_norm': 0.6261109709739685, 'learning_rate': 1.8791852894596984e-06, 'epoch': 2.23}

 74%|███████▍  | 9896/13317 [1:42:01<34:59,  1.63it/s]
 74%|███████▍  | 9897/13317 [1:42:02<34:57,  1.63it/s]
                                                      
{'loss': 0.128, 'grad_norm': 0.4857233464717865, 'learning_rate': 1.8781614049864527e-06, 'epoch': 2.23}

 74%|███████▍  | 9897/13317 [1:42:02<34:57,  1.63it/s]
 74%|███████▍  | 9898/13317 [1:42:02<34:56,  1.63it/s]
                                                      
{'loss': 0.1242, 'grad_norm': 0.44619935750961304, 'learning_rate': 1.8771377350165732e-06, 'epoch': 2.23}

 74%|███████▍  | 9898/13317 [1:42:03<34:56,  1.63it/s]
 74%|███████▍  | 9899/13317 [1:42:03<34:57,  1.63it/s]
                                                      
{'loss': 0.14, 'grad_norm': 0.5717452168464661, 'learning_rate': 1.8761142796203946e-06, 'epoch': 2.23}

 74%|███████▍  | 9899/13317 [1:42:03<34:57,  1.63it/s]
 74%|███████▍  | 9900/13317 [1:42:04<34:56,  1.63it/s]
                                                      
{'loss': 0.196, 'grad_norm': 0.6902079582214355, 'learning_rate': 1.875091038868243e-06, 'epoch': 2.23}

 74%|███████▍  | 9900/13317 [1:42:04<34:56,  1.63it/s]
 74%|███████▍  | 9901/13317 [1:42:04<34:56,  1.63it/s]
                                                      
{'loss': 0.1352, 'grad_norm': 0.5594089031219482, 'learning_rate': 1.8740680128304245e-06, 'epoch': 2.23}

 74%|███████▍  | 9901/13317 [1:42:04<34:56,  1.63it/s]
 74%|███████▍  | 9902/13317 [1:42:05<34:54,  1.63it/s]
                                                      
{'loss': 0.1937, 'grad_norm': 0.5318349003791809, 'learning_rate': 1.8730452015772293e-06, 'epoch': 2.23}

 74%|███████▍  | 9902/13317 [1:42:05<34:54,  1.63it/s]
 74%|███████▍  | 9903/13317 [1:42:06<34:53,  1.63it/s]
                                                      
{'loss': 0.1599, 'grad_norm': 0.6225523352622986, 'learning_rate': 1.8720226051789376e-06, 'epoch': 2.23}

 74%|███████▍  | 9903/13317 [1:42:06<34:53,  1.63it/s]
 74%|███████▍  | 9904/13317 [1:42:06<34:53,  1.63it/s]
                                                      
{'loss': 0.1189, 'grad_norm': 0.4738180935382843, 'learning_rate': 1.871000223705814e-06, 'epoch': 2.23}

 74%|███████▍  | 9904/13317 [1:42:06<34:53,  1.63it/s]
 74%|███████▍  | 9905/13317 [1:42:07<34:52,  1.63it/s]
                                                      
{'loss': 0.1235, 'grad_norm': 0.451150119304657, 'learning_rate': 1.8699780572281056e-06, 'epoch': 2.23}

 74%|███████▍  | 9905/13317 [1:42:07<34:52,  1.63it/s]
 74%|███████▍  | 9906/13317 [1:42:07<34:52,  1.63it/s]
                                                      
{'loss': 0.1475, 'grad_norm': 0.5044490694999695, 'learning_rate': 1.8689561058160444e-06, 'epoch': 2.23}

 74%|███████▍  | 9906/13317 [1:42:07<34:52,  1.63it/s]
 74%|███████▍  | 9907/13317 [1:42:08<34:50,  1.63it/s]
                                                      
{'loss': 0.1616, 'grad_norm': 0.5823085308074951, 'learning_rate': 1.8679343695398505e-06, 'epoch': 2.23}

 74%|███████▍  | 9907/13317 [1:42:08<34:50,  1.63it/s]
 74%|███████▍  | 9908/13317 [1:42:09<34:50,  1.63it/s]
                                                      
{'loss': 0.1719, 'grad_norm': 0.5962512493133545, 'learning_rate': 1.866912848469732e-06, 'epoch': 2.23}

 74%|███████▍  | 9908/13317 [1:42:09<34:50,  1.63it/s]
 74%|███████▍  | 9909/13317 [1:42:09<34:50,  1.63it/s]
                                                      
{'loss': 0.1229, 'grad_norm': 0.4896490275859833, 'learning_rate': 1.8658915426758705e-06, 'epoch': 2.23}

 74%|███████▍  | 9909/13317 [1:42:09<34:50,  1.63it/s]
 74%|███████▍  | 9910/13317 [1:42:10<34:49,  1.63it/s]
                                                      
{'loss': 0.1113, 'grad_norm': 0.4590355455875397, 'learning_rate': 1.8648704522284449e-06, 'epoch': 2.23}

 74%|███████▍  | 9910/13317 [1:42:10<34:49,  1.63it/s]
 74%|███████▍  | 9911/13317 [1:42:10<34:49,  1.63it/s]
                                                      
{'loss': 0.1579, 'grad_norm': 0.5958064794540405, 'learning_rate': 1.8638495771976162e-06, 'epoch': 2.23}

 74%|███████▍  | 9911/13317 [1:42:11<34:49,  1.63it/s]
 74%|███████▍  | 9912/13317 [1:42:11<34:49,  1.63it/s]
                                                      
{'loss': 0.1438, 'grad_norm': 0.5365381240844727, 'learning_rate': 1.8628289176535274e-06, 'epoch': 2.23}

 74%|███████▍  | 9912/13317 [1:42:11<34:49,  1.63it/s]
 74%|███████▍  | 9913/13317 [1:42:12<34:48,  1.63it/s]
                                                      
{'loss': 0.1249, 'grad_norm': 0.5089814066886902, 'learning_rate': 1.8618084736663073e-06, 'epoch': 2.23}

 74%|███████▍  | 9913/13317 [1:42:12<34:48,  1.63it/s]
 74%|███████▍  | 9914/13317 [1:42:12<34:48,  1.63it/s]
                                                      
{'loss': 0.1568, 'grad_norm': 0.57213294506073, 'learning_rate': 1.8607882453060727e-06, 'epoch': 2.23}

 74%|███████▍  | 9914/13317 [1:42:12<34:48,  1.63it/s]
 74%|███████▍  | 9915/13317 [1:42:13<34:48,  1.63it/s]
                                                      
{'loss': 0.2112, 'grad_norm': 0.7774919271469116, 'learning_rate': 1.8597682326429274e-06, 'epoch': 2.23}

 74%|███████▍  | 9915/13317 [1:42:13<34:48,  1.63it/s]
 74%|███████▍  | 9916/13317 [1:42:14<34:50,  1.63it/s]
                                                      
{'loss': 0.1419, 'grad_norm': 0.5485872030258179, 'learning_rate': 1.8587484357469504e-06, 'epoch': 2.23}

 74%|███████▍  | 9916/13317 [1:42:14<34:50,  1.63it/s]
 74%|███████▍  | 9917/13317 [1:42:14<34:49,  1.63it/s]
                                                      
{'loss': 0.1906, 'grad_norm': 0.5749278664588928, 'learning_rate': 1.8577288546882167e-06, 'epoch': 2.23}

 74%|███████▍  | 9917/13317 [1:42:14<34:49,  1.63it/s]
 74%|███████▍  | 9918/13317 [1:42:15<34:48,  1.63it/s]
                                                      
{'loss': 0.1328, 'grad_norm': 0.6418213844299316, 'learning_rate': 1.8567094895367827e-06, 'epoch': 2.23}

 74%|███████▍  | 9918/13317 [1:42:15<34:48,  1.63it/s]
 74%|███████▍  | 9919/13317 [1:42:15<34:45,  1.63it/s]
                                                      
{'loss': 0.1355, 'grad_norm': 0.5037322640419006, 'learning_rate': 1.8556903403626885e-06, 'epoch': 2.23}

 74%|███████▍  | 9919/13317 [1:42:15<34:45,  1.63it/s]
 74%|███████▍  | 9920/13317 [1:42:16<34:44,  1.63it/s]
                                                      
{'loss': 0.1281, 'grad_norm': 0.4888019263744354, 'learning_rate': 1.8546714072359583e-06, 'epoch': 2.23}

 74%|███████▍  | 9920/13317 [1:42:16<34:44,  1.63it/s]
 74%|███████▍  | 9921/13317 [1:42:17<34:43,  1.63it/s]
                                                      
{'loss': 0.1483, 'grad_norm': 0.5825597047805786, 'learning_rate': 1.8536526902266066e-06, 'epoch': 2.23}

 74%|███████▍  | 9921/13317 [1:42:17<34:43,  1.63it/s]
 75%|███████▍  | 9922/13317 [1:42:17<34:42,  1.63it/s]
                                                      
{'loss': 0.1569, 'grad_norm': 0.578382670879364, 'learning_rate': 1.8526341894046318e-06, 'epoch': 2.24}

 75%|███████▍  | 9922/13317 [1:42:17<34:42,  1.63it/s]
 75%|███████▍  | 9923/13317 [1:42:18<34:42,  1.63it/s]
                                                      
{'loss': 0.2805, 'grad_norm': 0.8126170635223389, 'learning_rate': 1.85161590484001e-06, 'epoch': 2.24}

 75%|███████▍  | 9923/13317 [1:42:18<34:42,  1.63it/s]
 75%|███████▍  | 9924/13317 [1:42:18<34:42,  1.63it/s]
                                                      
{'loss': 0.1513, 'grad_norm': 0.5693485140800476, 'learning_rate': 1.850597836602711e-06, 'epoch': 2.24}

 75%|███████▍  | 9924/13317 [1:42:19<34:42,  1.63it/s]
 75%|███████▍  | 9925/13317 [1:42:19<34:41,  1.63it/s]
                                                      
{'loss': 0.1261, 'grad_norm': 0.4573975205421448, 'learning_rate': 1.8495799847626889e-06, 'epoch': 2.24}

 75%|███████▍  | 9925/13317 [1:42:19<34:41,  1.63it/s]
 75%|███████▍  | 9926/13317 [1:42:20<34:40,  1.63it/s]
                                                      
{'loss': 0.1789, 'grad_norm': 0.7046648263931274, 'learning_rate': 1.8485623493898792e-06, 'epoch': 2.24}

 75%|███████▍  | 9926/13317 [1:42:20<34:40,  1.63it/s]
 75%|███████▍  | 9927/13317 [1:42:20<34:39,  1.63it/s]
                                                      
{'loss': 0.1132, 'grad_norm': 0.4644513428211212, 'learning_rate': 1.8475449305542014e-06, 'epoch': 2.24}

 75%|███████▍  | 9927/13317 [1:42:20<34:39,  1.63it/s]
 75%|███████▍  | 9928/13317 [1:42:21<34:38,  1.63it/s]
                                                      
{'loss': 0.1677, 'grad_norm': 0.5469604134559631, 'learning_rate': 1.8465277283255684e-06, 'epoch': 2.24}

 75%|███████▍  | 9928/13317 [1:42:21<34:38,  1.63it/s]
 75%|███████▍  | 9929/13317 [1:42:21<34:38,  1.63it/s]
                                                      
{'loss': 0.1651, 'grad_norm': 0.5758827924728394, 'learning_rate': 1.8455107427738677e-06, 'epoch': 2.24}

 75%|███████▍  | 9929/13317 [1:42:22<34:38,  1.63it/s]
 75%|███████▍  | 9930/13317 [1:42:22<34:37,  1.63it/s]
                                                      
{'loss': 0.1556, 'grad_norm': 0.5367944240570068, 'learning_rate': 1.844493973968981e-06, 'epoch': 2.24}

 75%|███████▍  | 9930/13317 [1:42:22<34:37,  1.63it/s]
 75%|███████▍  | 9931/13317 [1:42:23<34:36,  1.63it/s]
                                                      
{'loss': 0.1591, 'grad_norm': 0.5222906470298767, 'learning_rate': 1.8434774219807678e-06, 'epoch': 2.24}

 75%|███████▍  | 9931/13317 [1:42:23<34:36,  1.63it/s]
 75%|███████▍  | 9932/13317 [1:42:23<34:36,  1.63it/s]
                                                      
{'loss': 0.1643, 'grad_norm': 0.5889471769332886, 'learning_rate': 1.8424610868790794e-06, 'epoch': 2.24}

 75%|███████▍  | 9932/13317 [1:42:23<34:36,  1.63it/s]
 75%|███████▍  | 9933/13317 [1:42:24<34:37,  1.63it/s]
                                                      
{'loss': 0.1134, 'grad_norm': 0.45418018102645874, 'learning_rate': 1.8414449687337467e-06, 'epoch': 2.24}

 75%|███████▍  | 9933/13317 [1:42:24<34:37,  1.63it/s]
 75%|███████▍  | 9934/13317 [1:42:25<34:35,  1.63it/s]
                                                      
{'loss': 0.1153, 'grad_norm': 0.45200595259666443, 'learning_rate': 1.8404290676145858e-06, 'epoch': 2.24}

 75%|███████▍  | 9934/13317 [1:42:25<34:35,  1.63it/s]
 75%|███████▍  | 9935/13317 [1:42:25<34:35,  1.63it/s]
                                                      
{'loss': 0.1789, 'grad_norm': 0.6036788821220398, 'learning_rate': 1.839413383591404e-06, 'epoch': 2.24}

 75%|███████▍  | 9935/13317 [1:42:25<34:35,  1.63it/s]
 75%|███████▍  | 9936/13317 [1:42:26<34:35,  1.63it/s]
                                                      
{'loss': 0.1486, 'grad_norm': 0.5302656888961792, 'learning_rate': 1.8383979167339861e-06, 'epoch': 2.24}

 75%|███████▍  | 9936/13317 [1:42:26<34:35,  1.63it/s]
 75%|███████▍  | 9937/13317 [1:42:26<34:34,  1.63it/s]
                                                      
{'loss': 0.2007, 'grad_norm': 0.642711877822876, 'learning_rate': 1.837382667112108e-06, 'epoch': 2.24}

 75%|███████▍  | 9937/13317 [1:42:27<34:34,  1.63it/s]
 75%|███████▍  | 9938/13317 [1:42:27<34:33,  1.63it/s]
                                                      
{'loss': 0.1821, 'grad_norm': 0.5683667063713074, 'learning_rate': 1.8363676347955279e-06, 'epoch': 2.24}

 75%|███████▍  | 9938/13317 [1:42:27<34:33,  1.63it/s]
 75%|███████▍  | 9939/13317 [1:42:28<34:32,  1.63it/s]
                                                      
{'loss': 0.1691, 'grad_norm': 0.5956763029098511, 'learning_rate': 1.8353528198539861e-06, 'epoch': 2.24}

 75%|███████▍  | 9939/13317 [1:42:28<34:32,  1.63it/s]
 75%|███████▍  | 9940/13317 [1:42:28<34:32,  1.63it/s]
                                                      
{'loss': 0.1615, 'grad_norm': 0.6441787481307983, 'learning_rate': 1.8343382223572153e-06, 'epoch': 2.24}

 75%|███████▍  | 9940/13317 [1:42:28<34:32,  1.63it/s]
 75%|███████▍  | 9941/13317 [1:42:29<34:31,  1.63it/s]
                                                      
{'loss': 0.1434, 'grad_norm': 0.653739869594574, 'learning_rate': 1.8333238423749256e-06, 'epoch': 2.24}

 75%|███████▍  | 9941/13317 [1:42:29<34:31,  1.63it/s]
 75%|███████▍  | 9942/13317 [1:42:29<34:30,  1.63it/s]
                                                      
{'loss': 0.1363, 'grad_norm': 0.5030653476715088, 'learning_rate': 1.8323096799768186e-06, 'epoch': 2.24}

 75%|███████▍  | 9942/13317 [1:42:30<34:30,  1.63it/s]
 75%|███████▍  | 9943/13317 [1:42:30<34:29,  1.63it/s]
                                                      
{'loss': 0.1199, 'grad_norm': 0.48575231432914734, 'learning_rate': 1.831295735232575e-06, 'epoch': 2.24}

 75%|███████▍  | 9943/13317 [1:42:30<34:29,  1.63it/s]
 75%|███████▍  | 9944/13317 [1:42:31<34:28,  1.63it/s]
                                                      
{'loss': 0.1601, 'grad_norm': 0.726725697517395, 'learning_rate': 1.8302820082118671e-06, 'epoch': 2.24}

 75%|███████▍  | 9944/13317 [1:42:31<34:28,  1.63it/s]
 75%|███████▍  | 9945/13317 [1:42:31<34:28,  1.63it/s]
                                                      
{'loss': 0.1808, 'grad_norm': 0.8105412125587463, 'learning_rate': 1.8292684989843462e-06, 'epoch': 2.24}

 75%|███████▍  | 9945/13317 [1:42:31<34:28,  1.63it/s]
 75%|███████▍  | 9946/13317 [1:42:32<34:27,  1.63it/s]
                                                      
{'loss': 0.1228, 'grad_norm': 0.4960697889328003, 'learning_rate': 1.8282552076196507e-06, 'epoch': 2.24}

 75%|███████▍  | 9946/13317 [1:42:32<34:27,  1.63it/s]
 75%|███████▍  | 9947/13317 [1:42:33<34:26,  1.63it/s]
                                                      
{'loss': 0.121, 'grad_norm': 0.4353468120098114, 'learning_rate': 1.827242134187407e-06, 'epoch': 2.24}

 75%|███████▍  | 9947/13317 [1:42:33<34:26,  1.63it/s]
 75%|███████▍  | 9948/13317 [1:42:33<34:26,  1.63it/s]
                                                      
{'loss': 0.1611, 'grad_norm': 0.6567095518112183, 'learning_rate': 1.82622927875722e-06, 'epoch': 2.24}

 75%|███████▍  | 9948/13317 [1:42:33<34:26,  1.63it/s]
 75%|███████▍  | 9949/13317 [1:42:34<34:26,  1.63it/s]
                                                      
{'loss': 0.1506, 'grad_norm': 0.5387439131736755, 'learning_rate': 1.8252166413986883e-06, 'epoch': 2.24}

 75%|███████▍  | 9949/13317 [1:42:34<34:26,  1.63it/s]
 75%|███████▍  | 9950/13317 [1:42:34<34:24,  1.63it/s]
                                                      
{'loss': 0.1581, 'grad_norm': 0.48288995027542114, 'learning_rate': 1.8242042221813865e-06, 'epoch': 2.24}

 75%|███████▍  | 9950/13317 [1:42:34<34:24,  1.63it/s]
 75%|███████▍  | 9951/13317 [1:42:35<34:25,  1.63it/s]
                                                      
{'loss': 0.1418, 'grad_norm': 0.4763987064361572, 'learning_rate': 1.8231920211748822e-06, 'epoch': 2.24}

 75%|███████▍  | 9951/13317 [1:42:35<34:25,  1.63it/s]
 75%|███████▍  | 9952/13317 [1:42:36<34:24,  1.63it/s]
                                                      
{'loss': 0.1437, 'grad_norm': 0.49061354994773865, 'learning_rate': 1.8221800384487226e-06, 'epoch': 2.24}

 75%|███████▍  | 9952/13317 [1:42:36<34:24,  1.63it/s]
 75%|███████▍  | 9953/13317 [1:42:36<34:23,  1.63it/s]
                                                      
{'loss': 0.1889, 'grad_norm': 0.63301682472229, 'learning_rate': 1.82116827407244e-06, 'epoch': 2.24}

 75%|███████▍  | 9953/13317 [1:42:36<34:23,  1.63it/s]
 75%|███████▍  | 9954/13317 [1:42:37<34:23,  1.63it/s]
                                                      
{'loss': 0.1323, 'grad_norm': 0.45985350012779236, 'learning_rate': 1.8201567281155553e-06, 'epoch': 2.24}

 75%|███████▍  | 9954/13317 [1:42:37<34:23,  1.63it/s]
 75%|███████▍  | 9955/13317 [1:42:37<34:23,  1.63it/s]
                                                      
{'loss': 0.1534, 'grad_norm': 0.6121608018875122, 'learning_rate': 1.8191454006475751e-06, 'epoch': 2.24}

 75%|███████▍  | 9955/13317 [1:42:38<34:23,  1.63it/s]
 75%|███████▍  | 9956/13317 [1:42:38<34:22,  1.63it/s]
                                                      
{'loss': 0.1337, 'grad_norm': 0.5472672581672668, 'learning_rate': 1.8181342917379823e-06, 'epoch': 2.24}

 75%|███████▍  | 9956/13317 [1:42:38<34:22,  1.63it/s]
 75%|███████▍  | 9957/13317 [1:42:39<34:22,  1.63it/s]
                                                      
{'loss': 0.1427, 'grad_norm': 0.48716315627098083, 'learning_rate': 1.8171234014562532e-06, 'epoch': 2.24}

 75%|███████▍  | 9957/13317 [1:42:39<34:22,  1.63it/s]
 75%|███████▍  | 9958/13317 [1:42:39<34:21,  1.63it/s]
                                                      
{'loss': 0.1287, 'grad_norm': 0.49833542108535767, 'learning_rate': 1.8161127298718495e-06, 'epoch': 2.24}

 75%|███████▍  | 9958/13317 [1:42:39<34:21,  1.63it/s]
 75%|███████▍  | 9959/13317 [1:42:40<34:21,  1.63it/s]
                                                      
{'loss': 0.1505, 'grad_norm': 0.5620140433311462, 'learning_rate': 1.8151022770542125e-06, 'epoch': 2.24}

 75%|███████▍  | 9959/13317 [1:42:40<34:21,  1.63it/s]
 75%|███████▍  | 9960/13317 [1:42:41<34:21,  1.63it/s]
                                                      
{'loss': 0.1283, 'grad_norm': 0.5318852066993713, 'learning_rate': 1.8140920430727698e-06, 'epoch': 2.24}

 75%|███████▍  | 9960/13317 [1:42:41<34:21,  1.63it/s]
 75%|███████▍  | 9961/13317 [1:42:41<34:20,  1.63it/s]
                                                      
{'loss': 0.1627, 'grad_norm': 0.5759602785110474, 'learning_rate': 1.8130820279969368e-06, 'epoch': 2.24}

 75%|███████▍  | 9961/13317 [1:42:41<34:20,  1.63it/s]
 75%|███████▍  | 9962/13317 [1:42:42<34:19,  1.63it/s]
                                                      
{'loss': 0.1701, 'grad_norm': 0.6201345324516296, 'learning_rate': 1.8120722318961153e-06, 'epoch': 2.24}

 75%|███████▍  | 9962/13317 [1:42:42<34:19,  1.63it/s]
 75%|███████▍  | 9963/13317 [1:42:42<34:18,  1.63it/s]
                                                      
{'loss': 0.1451, 'grad_norm': 0.5179939866065979, 'learning_rate': 1.811062654839682e-06, 'epoch': 2.24}

 75%|███████▍  | 9963/13317 [1:42:42<34:18,  1.63it/s]
 75%|███████▍  | 9964/13317 [1:42:43<34:17,  1.63it/s]
                                                      
{'loss': 0.1464, 'grad_norm': 0.5060896277427673, 'learning_rate': 1.8100532968970103e-06, 'epoch': 2.24}

 75%|███████▍  | 9964/13317 [1:42:43<34:17,  1.63it/s]
 75%|███████▍  | 9965/13317 [1:42:44<34:17,  1.63it/s]
                                                      
{'loss': 0.1467, 'grad_norm': 0.5285196304321289, 'learning_rate': 1.8090441581374534e-06, 'epoch': 2.24}

 75%|███████▍  | 9965/13317 [1:42:44<34:17,  1.63it/s]
 75%|███████▍  | 9966/13317 [1:42:44<34:16,  1.63it/s]
                                                      
{'loss': 0.1674, 'grad_norm': 0.6664294600486755, 'learning_rate': 1.80803523863035e-06, 'epoch': 2.25}

 75%|███████▍  | 9966/13317 [1:42:44<34:16,  1.63it/s]
 75%|███████▍  | 9967/13317 [1:42:45<34:15,  1.63it/s]
                                                      
{'loss': 0.1383, 'grad_norm': 0.528551459312439, 'learning_rate': 1.8070265384450209e-06, 'epoch': 2.25}

 75%|███████▍  | 9967/13317 [1:42:45<34:15,  1.63it/s]
 75%|███████▍  | 9968/13317 [1:42:45<34:16,  1.63it/s]
                                                      
{'loss': 0.1694, 'grad_norm': 0.671309232711792, 'learning_rate': 1.8060180576507757e-06, 'epoch': 2.25}

 75%|███████▍  | 9968/13317 [1:42:46<34:16,  1.63it/s]
 75%|███████▍  | 9969/13317 [1:42:46<34:15,  1.63it/s]
                                                      
{'loss': 0.1327, 'grad_norm': 0.5238642692565918, 'learning_rate': 1.8050097963169128e-06, 'epoch': 2.25}

 75%|███████▍  | 9969/13317 [1:42:46<34:15,  1.63it/s]
 75%|███████▍  | 9970/13317 [1:42:47<34:14,  1.63it/s]
                                                      
{'loss': 0.1589, 'grad_norm': 0.592820405960083, 'learning_rate': 1.804001754512702e-06, 'epoch': 2.25}

 75%|███████▍  | 9970/13317 [1:42:47<34:14,  1.63it/s]
 75%|███████▍  | 9971/13317 [1:42:47<34:13,  1.63it/s]
                                                      
{'loss': 0.1533, 'grad_norm': 0.5244022607803345, 'learning_rate': 1.8029939323074102e-06, 'epoch': 2.25}

 75%|███████▍  | 9971/13317 [1:42:47<34:13,  1.63it/s]
 75%|███████▍  | 9972/13317 [1:42:48<34:13,  1.63it/s]
                                                      
{'loss': 0.1513, 'grad_norm': 0.5214933753013611, 'learning_rate': 1.8019863297702877e-06, 'epoch': 2.25}

 75%|███████▍  | 9972/13317 [1:42:48<34:13,  1.63it/s]
 75%|███████▍  | 9973/13317 [1:42:48<34:12,  1.63it/s]
                                                      
{'loss': 0.1515, 'grad_norm': 0.6009355187416077, 'learning_rate': 1.8009789469705658e-06, 'epoch': 2.25}

 75%|███████▍  | 9973/13317 [1:42:49<34:12,  1.63it/s]
 75%|███████▍  | 9974/13317 [1:42:49<34:11,  1.63it/s]
                                                      
{'loss': 0.1631, 'grad_norm': 0.5508453249931335, 'learning_rate': 1.7999717839774599e-06, 'epoch': 2.25}

 75%|███████▍  | 9974/13317 [1:42:49<34:11,  1.63it/s]
 75%|███████▍  | 9975/13317 [1:42:50<34:11,  1.63it/s]
                                                      
{'loss': 0.161, 'grad_norm': 0.524081826210022, 'learning_rate': 1.798964840860175e-06, 'epoch': 2.25}

 75%|███████▍  | 9975/13317 [1:42:50<34:11,  1.63it/s]
 75%|███████▍  | 9976/13317 [1:42:50<34:11,  1.63it/s]
                                                      
{'loss': 0.1522, 'grad_norm': 0.6486268639564514, 'learning_rate': 1.7979581176879002e-06, 'epoch': 2.25}

 75%|███████▍  | 9976/13317 [1:42:50<34:11,  1.63it/s]
 75%|███████▍  | 9977/13317 [1:42:51<34:09,  1.63it/s]
                                                      
{'loss': 0.1725, 'grad_norm': 0.6434136629104614, 'learning_rate': 1.7969516145298065e-06, 'epoch': 2.25}

 75%|███████▍  | 9977/13317 [1:42:51<34:09,  1.63it/s]
 75%|███████▍  | 9978/13317 [1:42:52<34:08,  1.63it/s]
                                                      
{'loss': 0.129, 'grad_norm': 0.4879458546638489, 'learning_rate': 1.79594533145505e-06, 'epoch': 2.25}

 75%|███████▍  | 9978/13317 [1:42:52<34:08,  1.63it/s]
 75%|███████▍  | 9979/13317 [1:42:52<34:08,  1.63it/s]
                                                      
{'loss': 0.1677, 'grad_norm': 0.5404034852981567, 'learning_rate': 1.7949392685327754e-06, 'epoch': 2.25}

 75%|███████▍  | 9979/13317 [1:42:52<34:08,  1.63it/s]
 75%|███████▍  | 9980/13317 [1:42:53<34:07,  1.63it/s]
                                                      
{'loss': 0.1235, 'grad_norm': 0.4722449779510498, 'learning_rate': 1.7939334258321094e-06, 'epoch': 2.25}

 75%|███████▍  | 9980/13317 [1:42:53<34:07,  1.63it/s]
 75%|███████▍  | 9981/13317 [1:42:53<34:06,  1.63it/s]
                                                      
{'loss': 0.1454, 'grad_norm': 0.6078458428382874, 'learning_rate': 1.7929278034221614e-06, 'epoch': 2.25}

 75%|███████▍  | 9981/13317 [1:42:54<34:06,  1.63it/s]
 75%|███████▍  | 9982/13317 [1:42:54<34:05,  1.63it/s]
                                                      
{'loss': 0.1344, 'grad_norm': 0.49783051013946533, 'learning_rate': 1.7919224013720321e-06, 'epoch': 2.25}

 75%|███████▍  | 9982/13317 [1:42:54<34:05,  1.63it/s]
 75%|███████▍  | 9983/13317 [1:42:55<34:06,  1.63it/s]
                                                      
{'loss': 0.1727, 'grad_norm': 0.6163423657417297, 'learning_rate': 1.7909172197507996e-06, 'epoch': 2.25}

 75%|███████▍  | 9983/13317 [1:42:55<34:06,  1.63it/s]
 75%|███████▍  | 9984/13317 [1:42:55<34:05,  1.63it/s]
                                                      
{'loss': 0.1516, 'grad_norm': 0.6107058525085449, 'learning_rate': 1.789912258627534e-06, 'epoch': 2.25}

 75%|███████▍  | 9984/13317 [1:42:55<34:05,  1.63it/s]
 75%|███████▍  | 9985/13317 [1:42:56<34:04,  1.63it/s]
                                                      
{'loss': 0.1519, 'grad_norm': 0.6385837197303772, 'learning_rate': 1.7889075180712838e-06, 'epoch': 2.25}

 75%|███████▍  | 9985/13317 [1:42:56<34:04,  1.63it/s]
 75%|███████▍  | 9986/13317 [1:42:56<34:03,  1.63it/s]
                                                      
{'loss': 0.1684, 'grad_norm': 0.581311047077179, 'learning_rate': 1.7879029981510888e-06, 'epoch': 2.25}

 75%|███████▍  | 9986/13317 [1:42:57<34:03,  1.63it/s]
 75%|███████▍  | 9987/13317 [1:42:57<34:01,  1.63it/s]
                                                      
{'loss': 0.1702, 'grad_norm': 0.6534162163734436, 'learning_rate': 1.786898698935967e-06, 'epoch': 2.25}

 75%|███████▍  | 9987/13317 [1:42:57<34:01,  1.63it/s]
 75%|███████▌  | 9988/13317 [1:42:58<34:02,  1.63it/s]
                                                      
{'loss': 0.2182, 'grad_norm': 0.5673636794090271, 'learning_rate': 1.7858946204949246e-06, 'epoch': 2.25}

 75%|███████▌  | 9988/13317 [1:42:58<34:02,  1.63it/s]
 75%|███████▌  | 9989/13317 [1:42:58<34:01,  1.63it/s]
                                                      
{'loss': 0.1853, 'grad_norm': 0.6479796767234802, 'learning_rate': 1.7848907628969547e-06, 'epoch': 2.25}

 75%|███████▌  | 9989/13317 [1:42:58<34:01,  1.63it/s]
 75%|███████▌  | 9990/13317 [1:42:59<34:01,  1.63it/s]
                                                      
{'loss': 0.1504, 'grad_norm': 0.6734256744384766, 'learning_rate': 1.7838871262110297e-06, 'epoch': 2.25}

 75%|███████▌  | 9990/13317 [1:42:59<34:01,  1.63it/s]
 75%|███████▌  | 9991/13317 [1:43:00<33:59,  1.63it/s]
                                                      
{'loss': 0.1575, 'grad_norm': 0.6267859935760498, 'learning_rate': 1.7828837105061137e-06, 'epoch': 2.25}

 75%|███████▌  | 9991/13317 [1:43:00<33:59,  1.63it/s]
 75%|███████▌  | 9992/13317 [1:43:00<33:58,  1.63it/s]
                                                      
{'loss': 0.1595, 'grad_norm': 0.5645995736122131, 'learning_rate': 1.7818805158511487e-06, 'epoch': 2.25}

 75%|███████▌  | 9992/13317 [1:43:00<33:58,  1.63it/s]
 75%|███████▌  | 9993/13317 [1:43:01<33:57,  1.63it/s]
                                                      
{'loss': 0.155, 'grad_norm': 0.6004151105880737, 'learning_rate': 1.7808775423150677e-06, 'epoch': 2.25}

 75%|███████▌  | 9993/13317 [1:43:01<33:57,  1.63it/s]
 75%|███████▌  | 9994/13317 [1:43:01<33:57,  1.63it/s]
                                                      
{'loss': 0.1371, 'grad_norm': 0.567622184753418, 'learning_rate': 1.779874789966784e-06, 'epoch': 2.25}

 75%|███████▌  | 9994/13317 [1:43:01<33:57,  1.63it/s]
 75%|███████▌  | 9995/13317 [1:43:02<33:57,  1.63it/s]
                                                      
{'loss': 0.1523, 'grad_norm': 0.5347889065742493, 'learning_rate': 1.7788722588751961e-06, 'epoch': 2.25}

 75%|███████▌  | 9995/13317 [1:43:02<33:57,  1.63it/s]
 75%|███████▌  | 9996/13317 [1:43:03<33:57,  1.63it/s]
                                                      
{'loss': 0.179, 'grad_norm': 0.5997388362884521, 'learning_rate': 1.7778699491091911e-06, 'epoch': 2.25}

 75%|███████▌  | 9996/13317 [1:43:03<33:57,  1.63it/s]
 75%|███████▌  | 9997/13317 [1:43:03<34:51,  1.59it/s]
                                                      
{'loss': 0.1379, 'grad_norm': 0.6709505915641785, 'learning_rate': 1.776867860737636e-06, 'epoch': 2.25}

 75%|███████▌  | 9997/13317 [1:43:03<34:51,  1.59it/s]
 75%|███████▌  | 9998/13317 [1:43:04<34:33,  1.60it/s]
                                                      
{'loss': 0.1649, 'grad_norm': 0.587907075881958, 'learning_rate': 1.7758659938293871e-06, 'epoch': 2.25}

 75%|███████▌  | 9998/13317 [1:43:04<34:33,  1.60it/s]
 75%|███████▌  | 9999/13317 [1:43:04<34:21,  1.61it/s]
                                                      
{'loss': 0.1374, 'grad_norm': 0.5165930986404419, 'learning_rate': 1.7748643484532823e-06, 'epoch': 2.25}

 75%|███████▌  | 9999/13317 [1:43:05<34:21,  1.61it/s]
 75%|███████▌  | 10000/13317 [1:43:05<34:16,  1.61it/s]
                                                       
{'loss': 0.1683, 'grad_norm': 0.5711153149604797, 'learning_rate': 1.7738629246781435e-06, 'epoch': 2.25}

 75%|███████▌  | 10000/13317 [1:43:05<34:16,  1.61it/s]

  0%|          | 0/13 [00:00<?, ?it/s]

 15%|█▌        | 2/13 [00:00<00:00, 13.33it/s]

 31%|███       | 4/13 [00:00<00:01,  8.39it/s]

 38%|███▊      | 5/13 [00:00<00:01,  7.79it/s]

 46%|████▌     | 6/13 [00:00<00:00,  7.42it/s]

 54%|█████▍    | 7/13 [00:00<00:00,  7.18it/s]

 62%|██████▏   | 8/13 [00:01<00:00,  7.01it/s]

 69%|██████▉   | 9/13 [00:01<00:00,  6.90it/s]

 77%|███████▋  | 10/13 [00:01<00:00,  6.83it/s]

 85%|████████▍ | 11/13 [00:01<00:00,  6.78it/s]

 92%|█████████▏| 12/13 [00:01<00:00,  6.74it/s]

100%|██████████| 13/13 [00:01<00:00,  6.75it/s]
                                                       

                                               
{'eval_loss': 0.5498818755149841, 'eval_runtime': 1.9546, 'eval_samples_per_second': 102.324, 'eval_steps_per_second': 6.651, 'epoch': 2.25}

 75%|███████▌  | 10000/13317 [1:43:07<34:16,  1.61it/s]

100%|██████████| 13/13 [00:01<00:00,  6.75it/s]

                                               
 75%|███████▌  | 10001/13317 [1:43:22<5:10:38,  5.62s/it]
                                                         
{'loss': 0.14, 'grad_norm': 0.5980305671691895, 'learning_rate': 1.7728617225727824e-06, 'epoch': 2.25}

 75%|███████▌  | 10001/13317 [1:43:23<5:10:38,  5.62s/it]
 75%|███████▌  | 10002/13317 [1:43:23<3:47:32,  4.12s/it]
                                                         
{'loss': 0.1709, 'grad_norm': 0.6841018199920654, 'learning_rate': 1.771860742205988e-06, 'epoch': 2.25}

 75%|███████▌  | 10002/13317 [1:43:23<3:47:32,  4.12s/it]
 75%|███████▌  | 10003/13317 [1:43:24<2:49:23,  3.07s/it]
                                                         
{'loss': 0.16, 'grad_norm': 0.6443012356758118, 'learning_rate': 1.7708599836465429e-06, 'epoch': 2.25}

 75%|███████▌  | 10003/13317 [1:43:24<2:49:23,  3.07s/it]
 75%|███████▌  | 10004/13317 [1:43:24<2:08:41,  2.33s/it]
                                                         
{'loss': 0.1433, 'grad_norm': 0.5336258411407471, 'learning_rate': 1.7698594469632063e-06, 'epoch': 2.25}

 75%|███████▌  | 10004/13317 [1:43:24<2:08:41,  2.33s/it]
 75%|███████▌  | 10005/13317 [1:43:25<1:40:13,  1.82s/it]
                                                         
{'loss': 0.1522, 'grad_norm': 0.5601409077644348, 'learning_rate': 1.7688591322247283e-06, 'epoch': 2.25}

 75%|███████▌  | 10005/13317 [1:43:25<1:40:13,  1.82s/it]
 75%|███████▌  | 10006/13317 [1:43:25<1:20:19,  1.46s/it]
                                                         
{'loss': 0.1472, 'grad_norm': 0.539909839630127, 'learning_rate': 1.7678590394998397e-06, 'epoch': 2.25}

 75%|███████▌  | 10006/13317 [1:43:26<1:20:19,  1.46s/it]
 75%|███████▌  | 10007/13317 [1:43:26<1:06:20,  1.20s/it]
                                                         
{'loss': 0.1511, 'grad_norm': 0.5790305733680725, 'learning_rate': 1.766859168857256e-06, 'epoch': 2.25}

 75%|███████▌  | 10007/13317 [1:43:26<1:06:20,  1.20s/it]
 75%|███████▌  | 10008/13317 [1:43:27<56:35,  1.03s/it]  
                                                       
{'loss': 0.1701, 'grad_norm': 0.5823768973350525, 'learning_rate': 1.7658595203656802e-06, 'epoch': 2.25}

 75%|███████▌  | 10008/13317 [1:43:27<56:35,  1.03s/it]
 75%|███████▌  | 10009/13317 [1:43:27<49:48,  1.11it/s]
                                                       
{'loss': 0.1413, 'grad_norm': 0.5464665293693542, 'learning_rate': 1.764860094093801e-06, 'epoch': 2.25}

 75%|███████▌  | 10009/13317 [1:43:27<49:48,  1.11it/s]
 75%|███████▌  | 10010/13317 [1:43:28<44:59,  1.23it/s]
                                                       
{'loss': 0.1508, 'grad_norm': 0.5437793135643005, 'learning_rate': 1.7638608901102873e-06, 'epoch': 2.26}

 75%|███████▌  | 10010/13317 [1:43:28<44:59,  1.23it/s]
 75%|███████▌  | 10011/13317 [1:43:29<41:38,  1.32it/s]
                                                       
{'loss': 0.1358, 'grad_norm': 0.5374390482902527, 'learning_rate': 1.7628619084837939e-06, 'epoch': 2.26}

 75%|███████▌  | 10011/13317 [1:43:29<41:38,  1.32it/s]
 75%|███████▌  | 10012/13317 [1:43:29<39:15,  1.40it/s]
                                                       
{'loss': 0.1521, 'grad_norm': 0.6010021567344666, 'learning_rate': 1.7618631492829641e-06, 'epoch': 2.26}

 75%|███████▌  | 10012/13317 [1:43:29<39:15,  1.40it/s]
 75%|███████▌  | 10013/13317 [1:43:30<37:36,  1.46it/s]
                                                       
{'loss': 0.1244, 'grad_norm': 0.4888395667076111, 'learning_rate': 1.7608646125764217e-06, 'epoch': 2.26}

 75%|███████▌  | 10013/13317 [1:43:30<37:36,  1.46it/s]
 75%|███████▌  | 10014/13317 [1:43:30<36:28,  1.51it/s]
                                                       
{'loss': 0.1373, 'grad_norm': 0.5025853514671326, 'learning_rate': 1.7598662984327758e-06, 'epoch': 2.26}

 75%|███████▌  | 10014/13317 [1:43:31<36:28,  1.51it/s]
 75%|███████▌  | 10015/13317 [1:43:31<35:39,  1.54it/s]
                                                       
{'loss': 0.1369, 'grad_norm': 0.5570069551467896, 'learning_rate': 1.758868206920622e-06, 'epoch': 2.26}

 75%|███████▌  | 10015/13317 [1:43:31<35:39,  1.54it/s]
 75%|███████▌  | 10016/13317 [1:43:32<35:05,  1.57it/s]
                                                       
{'loss': 0.1516, 'grad_norm': 0.575553834438324, 'learning_rate': 1.7578703381085432e-06, 'epoch': 2.26}

 75%|███████▌  | 10016/13317 [1:43:32<35:05,  1.57it/s]
 75%|███████▌  | 10017/13317 [1:43:32<34:41,  1.59it/s]
                                                       
{'loss': 0.1454, 'grad_norm': 0.5632189512252808, 'learning_rate': 1.7568726920650976e-06, 'epoch': 2.26}

 75%|███████▌  | 10017/13317 [1:43:32<34:41,  1.59it/s]
 75%|███████▌  | 10018/13317 [1:43:33<34:23,  1.60it/s]
                                                       
{'loss': 0.162, 'grad_norm': 0.5810579657554626, 'learning_rate': 1.7558752688588366e-06, 'epoch': 2.26}

 75%|███████▌  | 10018/13317 [1:43:33<34:23,  1.60it/s]
 75%|███████▌  | 10019/13317 [1:43:33<34:14,  1.61it/s]
                                                       
{'loss': 0.1782, 'grad_norm': 0.6223437190055847, 'learning_rate': 1.754878068558295e-06, 'epoch': 2.26}

 75%|███████▌  | 10019/13317 [1:43:34<34:14,  1.61it/s]
 75%|███████▌  | 10020/13317 [1:43:34<34:04,  1.61it/s]
                                                       
{'loss': 0.1577, 'grad_norm': 0.6313531994819641, 'learning_rate': 1.7538810912319909e-06, 'epoch': 2.26}

 75%|███████▌  | 10020/13317 [1:43:34<34:04,  1.61it/s]
 75%|███████▌  | 10021/13317 [1:43:35<34:00,  1.62it/s]
                                                       
{'loss': 0.1266, 'grad_norm': 0.502604067325592, 'learning_rate': 1.7528843369484238e-06, 'epoch': 2.26}

 75%|███████▌  | 10021/13317 [1:43:35<34:00,  1.62it/s]
 75%|███████▌  | 10022/13317 [1:43:35<33:53,  1.62it/s]
                                                       
{'loss': 0.1741, 'grad_norm': 0.5844166874885559, 'learning_rate': 1.7518878057760847e-06, 'epoch': 2.26}

 75%|███████▌  | 10022/13317 [1:43:35<33:53,  1.62it/s]
 75%|███████▌  | 10023/13317 [1:43:36<33:48,  1.62it/s]
                                                       
{'loss': 0.1312, 'grad_norm': 0.8277016282081604, 'learning_rate': 1.7508914977834475e-06, 'epoch': 2.26}

 75%|███████▌  | 10023/13317 [1:43:36<33:48,  1.62it/s]
 75%|███████▌  | 10024/13317 [1:43:37<33:58,  1.62it/s]
                                                       
{'loss': 0.1789, 'grad_norm': 0.6479934453964233, 'learning_rate': 1.7498954130389635e-06, 'epoch': 2.26}

 75%|███████▌  | 10024/13317 [1:43:37<33:58,  1.62it/s]
 75%|███████▌  | 10025/13317 [1:43:37<33:51,  1.62it/s]
                                                       
{'loss': 0.149, 'grad_norm': 0.6560859084129333, 'learning_rate': 1.7488995516110774e-06, 'epoch': 2.26}

 75%|███████▌  | 10025/13317 [1:43:37<33:51,  1.62it/s]
 75%|███████▌  | 10026/13317 [1:43:38<33:50,  1.62it/s]
                                                       
{'loss': 0.1521, 'grad_norm': 0.5346865057945251, 'learning_rate': 1.7479039135682163e-06, 'epoch': 2.26}

 75%|███████▌  | 10026/13317 [1:43:38<33:50,  1.62it/s]
 75%|███████▌  | 10027/13317 [1:43:38<33:46,  1.62it/s]
                                                       
{'loss': 0.1444, 'grad_norm': 0.5328335762023926, 'learning_rate': 1.746908498978791e-06, 'epoch': 2.26}

 75%|███████▌  | 10027/13317 [1:43:39<33:46,  1.62it/s]
 75%|███████▌  | 10028/13317 [1:43:39<33:43,  1.63it/s]
                                                       
{'loss': 0.1535, 'grad_norm': 0.5860457420349121, 'learning_rate': 1.745913307911194e-06, 'epoch': 2.26}

 75%|███████▌  | 10028/13317 [1:43:39<33:43,  1.63it/s]
 75%|███████▌  | 10029/13317 [1:43:40<33:42,  1.63it/s]
                                                       
{'loss': 0.1309, 'grad_norm': 0.4939118027687073, 'learning_rate': 1.7449183404338071e-06, 'epoch': 2.26}

 75%|███████▌  | 10029/13317 [1:43:40<33:42,  1.63it/s]
 75%|███████▌  | 10030/13317 [1:43:40<33:41,  1.63it/s]
                                                       
{'loss': 0.1534, 'grad_norm': 0.5906164050102234, 'learning_rate': 1.7439235966149982e-06, 'epoch': 2.26}

 75%|███████▌  | 10030/13317 [1:43:40<33:41,  1.63it/s]
 75%|███████▌  | 10031/13317 [1:43:41<33:41,  1.63it/s]
                                                       
{'loss': 0.1459, 'grad_norm': 0.516021192073822, 'learning_rate': 1.7429290765231132e-06, 'epoch': 2.26}

 75%|███████▌  | 10031/13317 [1:43:41<33:41,  1.63it/s]
 75%|███████▌  | 10032/13317 [1:43:41<33:39,  1.63it/s]
                                                       
{'loss': 0.1647, 'grad_norm': 0.5938966274261475, 'learning_rate': 1.7419347802264858e-06, 'epoch': 2.26}

 75%|███████▌  | 10032/13317 [1:43:42<33:39,  1.63it/s]
 75%|███████▌  | 10033/13317 [1:43:42<33:37,  1.63it/s]
                                                       
{'loss': 0.1508, 'grad_norm': 0.6062262058258057, 'learning_rate': 1.7409407077934377e-06, 'epoch': 2.26}

 75%|███████▌  | 10033/13317 [1:43:42<33:37,  1.63it/s]
 75%|███████▌  | 10034/13317 [1:43:43<33:38,  1.63it/s]
                                                       
{'loss': 0.1166, 'grad_norm': 0.523393452167511, 'learning_rate': 1.7399468592922703e-06, 'epoch': 2.26}

 75%|███████▌  | 10034/13317 [1:43:43<33:38,  1.63it/s]
 75%|███████▌  | 10035/13317 [1:43:43<33:37,  1.63it/s]
                                                       
{'loss': 0.2004, 'grad_norm': 0.6448150277137756, 'learning_rate': 1.7389532347912702e-06, 'epoch': 2.26}

 75%|███████▌  | 10035/13317 [1:43:43<33:37,  1.63it/s]
 75%|███████▌  | 10036/13317 [1:43:44<33:47,  1.62it/s]
                                                       
{'loss': 0.1073, 'grad_norm': 0.45252203941345215, 'learning_rate': 1.7379598343587112e-06, 'epoch': 2.26}

 75%|███████▌  | 10036/13317 [1:43:44<33:47,  1.62it/s]
 75%|███████▌  | 10037/13317 [1:43:45<33:43,  1.62it/s]
                                                       
{'loss': 0.1332, 'grad_norm': 0.49337512254714966, 'learning_rate': 1.7369666580628524e-06, 'epoch': 2.26}

 75%|███████▌  | 10037/13317 [1:43:45<33:43,  1.62it/s]
 75%|███████▌  | 10038/13317 [1:43:45<33:40,  1.62it/s]
                                                       
{'loss': 0.1824, 'grad_norm': 0.6618478298187256, 'learning_rate': 1.7359737059719335e-06, 'epoch': 2.26}

 75%|███████▌  | 10038/13317 [1:43:45<33:40,  1.62it/s]
 75%|███████▌  | 10039/13317 [1:43:46<33:41,  1.62it/s]
                                                       
{'loss': 0.1545, 'grad_norm': 0.5657021403312683, 'learning_rate': 1.7349809781541798e-06, 'epoch': 2.26}

 75%|███████▌  | 10039/13317 [1:43:46<33:41,  1.62it/s]
 75%|███████▌  | 10040/13317 [1:43:46<33:38,  1.62it/s]
                                                       
{'loss': 0.162, 'grad_norm': 0.6070093512535095, 'learning_rate': 1.7339884746778046e-06, 'epoch': 2.26}

 75%|███████▌  | 10040/13317 [1:43:47<33:38,  1.62it/s]
 75%|███████▌  | 10041/13317 [1:43:47<33:37,  1.62it/s]
                                                       
{'loss': 0.1549, 'grad_norm': 0.5725731253623962, 'learning_rate': 1.732996195611003e-06, 'epoch': 2.26}

 75%|███████▌  | 10041/13317 [1:43:47<33:37,  1.62it/s]
 75%|███████▌  | 10042/13317 [1:43:48<33:34,  1.63it/s]
                                                       
{'loss': 0.1827, 'grad_norm': 0.6573620438575745, 'learning_rate': 1.7320041410219523e-06, 'epoch': 2.26}

 75%|███████▌  | 10042/13317 [1:43:48<33:34,  1.63it/s]
 75%|███████▌  | 10043/13317 [1:43:48<33:32,  1.63it/s]
                                                       
{'loss': 0.137, 'grad_norm': 0.6168983578681946, 'learning_rate': 1.7310123109788207e-06, 'epoch': 2.26}

 75%|███████▌  | 10043/13317 [1:43:48<33:32,  1.63it/s]
 75%|███████▌  | 10044/13317 [1:43:49<33:34,  1.62it/s]
                                                       
{'loss': 0.152, 'grad_norm': 0.5799864530563354, 'learning_rate': 1.7300207055497542e-06, 'epoch': 2.26}

 75%|███████▌  | 10044/13317 [1:43:49<33:34,  1.62it/s]
 75%|███████▌  | 10045/13317 [1:43:49<33:32,  1.63it/s]
                                                       
{'loss': 0.1451, 'grad_norm': 0.5097622275352478, 'learning_rate': 1.7290293248028906e-06, 'epoch': 2.26}

 75%|███████▌  | 10045/13317 [1:43:50<33:32,  1.63it/s]
 75%|███████▌  | 10046/13317 [1:43:50<33:33,  1.62it/s]
                                                       
{'loss': 0.1348, 'grad_norm': 0.5954887270927429, 'learning_rate': 1.7280381688063436e-06, 'epoch': 2.26}

 75%|███████▌  | 10046/13317 [1:43:50<33:33,  1.62it/s]
 75%|███████▌  | 10047/13317 [1:43:51<33:31,  1.63it/s]
                                                       
{'loss': 0.1333, 'grad_norm': 0.5147674679756165, 'learning_rate': 1.7270472376282205e-06, 'epoch': 2.26}

 75%|███████▌  | 10047/13317 [1:43:51<33:31,  1.63it/s]
 75%|███████▌  | 10048/13317 [1:43:51<33:29,  1.63it/s]
                                                       
{'loss': 0.1691, 'grad_norm': 0.6133910417556763, 'learning_rate': 1.7260565313366067e-06, 'epoch': 2.26}

 75%|███████▌  | 10048/13317 [1:43:51<33:29,  1.63it/s]
 75%|███████▌  | 10049/13317 [1:43:52<33:30,  1.63it/s]
                                                       
{'loss': 0.1153, 'grad_norm': 0.42462223768234253, 'learning_rate': 1.7250660499995726e-06, 'epoch': 2.26}

 75%|███████▌  | 10049/13317 [1:43:52<33:30,  1.63it/s]
 75%|███████▌  | 10050/13317 [1:43:53<33:28,  1.63it/s]
                                                       
{'loss': 0.1695, 'grad_norm': 0.6179179549217224, 'learning_rate': 1.7240757936851782e-06, 'epoch': 2.26}

 75%|███████▌  | 10050/13317 [1:43:53<33:28,  1.63it/s]
 75%|███████▌  | 10051/13317 [1:43:53<33:33,  1.62it/s]
                                                       
{'loss': 0.1361, 'grad_norm': 0.5058053731918335, 'learning_rate': 1.723085762461461e-06, 'epoch': 2.26}

 75%|███████▌  | 10051/13317 [1:43:53<33:33,  1.62it/s]
 75%|███████▌  | 10052/13317 [1:43:54<33:30,  1.62it/s]
                                                       
{'loss': 0.154, 'grad_norm': 0.5584905743598938, 'learning_rate': 1.7220959563964501e-06, 'epoch': 2.26}

 75%|███████▌  | 10052/13317 [1:43:54<33:30,  1.62it/s]
 75%|███████▌  | 10053/13317 [1:43:54<33:27,  1.63it/s]
                                                       
{'loss': 0.1306, 'grad_norm': 0.5564922094345093, 'learning_rate': 1.7211063755581524e-06, 'epoch': 2.26}

 75%|███████▌  | 10053/13317 [1:43:55<33:27,  1.63it/s]
 75%|███████▌  | 10054/13317 [1:43:55<33:27,  1.63it/s]
                                                       
{'loss': 0.1374, 'grad_norm': 0.6250126361846924, 'learning_rate': 1.7201170200145662e-06, 'epoch': 2.26}

 75%|███████▌  | 10054/13317 [1:43:55<33:27,  1.63it/s]
 76%|███████▌  | 10055/13317 [1:43:56<33:24,  1.63it/s]
                                                       
{'loss': 0.1692, 'grad_norm': 0.5865836143493652, 'learning_rate': 1.7191278898336666e-06, 'epoch': 2.27}

 76%|███████▌  | 10055/13317 [1:43:56<33:24,  1.63it/s]
 76%|███████▌  | 10056/13317 [1:43:56<33:23,  1.63it/s]
                                                       
{'loss': 0.1929, 'grad_norm': 0.7551862597465515, 'learning_rate': 1.7181389850834207e-06, 'epoch': 2.27}

 76%|███████▌  | 10056/13317 [1:43:56<33:23,  1.63it/s]
 76%|███████▌  | 10057/13317 [1:43:57<33:22,  1.63it/s]
                                                       
{'loss': 0.1518, 'grad_norm': 0.5595750212669373, 'learning_rate': 1.7171503058317763e-06, 'epoch': 2.27}

 76%|███████▌  | 10057/13317 [1:43:57<33:22,  1.63it/s]
 76%|███████▌  | 10058/13317 [1:43:57<33:21,  1.63it/s]
                                                       
{'loss': 0.1292, 'grad_norm': 0.49270474910736084, 'learning_rate': 1.7161618521466628e-06, 'epoch': 2.27}

 76%|███████▌  | 10058/13317 [1:43:58<33:21,  1.63it/s]
 76%|███████▌  | 10059/13317 [1:43:58<33:24,  1.63it/s]
                                                       
{'loss': 0.1444, 'grad_norm': 0.5728105902671814, 'learning_rate': 1.715173624096002e-06, 'epoch': 2.27}

 76%|███████▌  | 10059/13317 [1:43:58<33:24,  1.63it/s]
 76%|███████▌  | 10060/13317 [1:43:59<33:23,  1.63it/s]
                                                       
{'loss': 0.1525, 'grad_norm': 0.5072213411331177, 'learning_rate': 1.7141856217476938e-06, 'epoch': 2.27}

 76%|███████▌  | 10060/13317 [1:43:59<33:23,  1.63it/s]
 76%|███████▌  | 10061/13317 [1:43:59<33:22,  1.63it/s]
                                                       
{'loss': 0.1335, 'grad_norm': 0.49946799874305725, 'learning_rate': 1.7131978451696218e-06, 'epoch': 2.27}

 76%|███████▌  | 10061/13317 [1:43:59<33:22,  1.63it/s]
 76%|███████▌  | 10062/13317 [1:44:00<33:20,  1.63it/s]
                                                       
{'loss': 0.1517, 'grad_norm': 0.6791331768035889, 'learning_rate': 1.7122102944296593e-06, 'epoch': 2.27}

 76%|███████▌  | 10062/13317 [1:44:00<33:20,  1.63it/s]
 76%|███████▌  | 10063/13317 [1:44:01<33:18,  1.63it/s]
                                                       
{'loss': 0.135, 'grad_norm': 0.5399020910263062, 'learning_rate': 1.7112229695956628e-06, 'epoch': 2.27}

 76%|███████▌  | 10063/13317 [1:44:01<33:18,  1.63it/s]
 76%|███████▌  | 10064/13317 [1:44:01<33:19,  1.63it/s]
                                                       
{'loss': 0.1555, 'grad_norm': 0.5623376965522766, 'learning_rate': 1.7102358707354711e-06, 'epoch': 2.27}

 76%|███████▌  | 10064/13317 [1:44:01<33:19,  1.63it/s]
 76%|███████▌  | 10065/13317 [1:44:02<33:17,  1.63it/s]
                                                       
{'loss': 0.1402, 'grad_norm': 0.5838342905044556, 'learning_rate': 1.7092489979169052e-06, 'epoch': 2.27}

 76%|███████▌  | 10065/13317 [1:44:02<33:17,  1.63it/s]
 76%|███████▌  | 10066/13317 [1:44:02<33:19,  1.63it/s]
                                                       
{'loss': 0.1647, 'grad_norm': 0.5041776895523071, 'learning_rate': 1.7082623512077784e-06, 'epoch': 2.27}

 76%|███████▌  | 10066/13317 [1:44:02<33:19,  1.63it/s]
 76%|███████▌  | 10067/13317 [1:44:03<33:17,  1.63it/s]
                                                       
{'loss': 0.1699, 'grad_norm': 0.5782407522201538, 'learning_rate': 1.7072759306758813e-06, 'epoch': 2.27}

 76%|███████▌  | 10067/13317 [1:44:03<33:17,  1.63it/s]
 76%|███████▌  | 10068/13317 [1:44:04<33:15,  1.63it/s]
                                                       
{'loss': 0.14, 'grad_norm': 0.48687636852264404, 'learning_rate': 1.7062897363889896e-06, 'epoch': 2.27}

 76%|███████▌  | 10068/13317 [1:44:04<33:15,  1.63it/s]
 76%|███████▌  | 10069/13317 [1:44:04<33:18,  1.63it/s]
                                                       
{'loss': 0.2259, 'grad_norm': 0.9082575440406799, 'learning_rate': 1.7053037684148682e-06, 'epoch': 2.27}

 76%|███████▌  | 10069/13317 [1:44:04<33:18,  1.63it/s]
 76%|███████▌  | 10070/13317 [1:44:05<33:16,  1.63it/s]
                                                       
{'loss': 0.1704, 'grad_norm': 0.5608575344085693, 'learning_rate': 1.704318026821264e-06, 'epoch': 2.27}

 76%|███████▌  | 10070/13317 [1:44:05<33:16,  1.63it/s]
 76%|███████▌  | 10071/13317 [1:44:05<33:25,  1.62it/s]
                                                       
{'loss': 0.1295, 'grad_norm': 0.4487066864967346, 'learning_rate': 1.7033325116759064e-06, 'epoch': 2.27}

 76%|███████▌  | 10071/13317 [1:44:06<33:25,  1.62it/s]
 76%|███████▌  | 10072/13317 [1:44:06<33:20,  1.62it/s]
                                                       
{'loss': 0.1566, 'grad_norm': 0.6082825660705566, 'learning_rate': 1.7023472230465094e-06, 'epoch': 2.27}

 76%|███████▌  | 10072/13317 [1:44:06<33:20,  1.62it/s]
 76%|███████▌  | 10073/13317 [1:44:07<33:18,  1.62it/s]
                                                       
{'loss': 0.2265, 'grad_norm': 0.6675247550010681, 'learning_rate': 1.7013621610007759e-06, 'epoch': 2.27}

 76%|███████▌  | 10073/13317 [1:44:07<33:18,  1.62it/s]
 76%|███████▌  | 10074/13317 [1:44:07<33:17,  1.62it/s]
                                                       
{'loss': 0.1288, 'grad_norm': 0.4898954927921295, 'learning_rate': 1.7003773256063882e-06, 'epoch': 2.27}

 76%|███████▌  | 10074/13317 [1:44:07<33:17,  1.62it/s]
 76%|███████▌  | 10075/13317 [1:44:08<33:15,  1.62it/s]
                                                       
{'loss': 0.1516, 'grad_norm': 0.5317237973213196, 'learning_rate': 1.6993927169310142e-06, 'epoch': 2.27}

 76%|███████▌  | 10075/13317 [1:44:08<33:15,  1.62it/s]
 76%|███████▌  | 10076/13317 [1:44:09<33:15,  1.62it/s]
                                                       
{'loss': 0.1204, 'grad_norm': 0.4594806134700775, 'learning_rate': 1.6984083350423076e-06, 'epoch': 2.27}

 76%|███████▌  | 10076/13317 [1:44:09<33:15,  1.62it/s]
 76%|███████▌  | 10077/13317 [1:44:09<33:12,  1.63it/s]
                                                       
{'loss': 0.1681, 'grad_norm': 0.5579805374145508, 'learning_rate': 1.6974241800079078e-06, 'epoch': 2.27}

 76%|███████▌  | 10077/13317 [1:44:09<33:12,  1.63it/s]
 76%|███████▌  | 10078/13317 [1:44:10<33:10,  1.63it/s]
                                                       
{'loss': 0.1616, 'grad_norm': 0.5964553356170654, 'learning_rate': 1.6964402518954353e-06, 'epoch': 2.27}

 76%|███████▌  | 10078/13317 [1:44:10<33:10,  1.63it/s]
 76%|███████▌  | 10079/13317 [1:44:10<33:11,  1.63it/s]
                                                       
{'loss': 0.1357, 'grad_norm': 0.5533422231674194, 'learning_rate': 1.695456550772494e-06, 'epoch': 2.27}

 76%|███████▌  | 10079/13317 [1:44:10<33:11,  1.63it/s]
 76%|███████▌  | 10080/13317 [1:44:11<33:09,  1.63it/s]
                                                       
{'loss': 0.1335, 'grad_norm': 0.49804607033729553, 'learning_rate': 1.6944730767066785e-06, 'epoch': 2.27}

 76%|███████▌  | 10080/13317 [1:44:11<33:09,  1.63it/s]
 76%|███████▌  | 10081/13317 [1:44:12<33:07,  1.63it/s]
                                                       
{'loss': 0.1769, 'grad_norm': 0.6781308054924011, 'learning_rate': 1.6934898297655628e-06, 'epoch': 2.27}

 76%|███████▌  | 10081/13317 [1:44:12<33:07,  1.63it/s]
 76%|███████▌  | 10082/13317 [1:44:12<33:06,  1.63it/s]
                                                       
{'loss': 0.1247, 'grad_norm': 0.47168904542922974, 'learning_rate': 1.6925068100167035e-06, 'epoch': 2.27}

 76%|███████▌  | 10082/13317 [1:44:12<33:06,  1.63it/s]
 76%|███████▌  | 10083/13317 [1:44:13<33:05,  1.63it/s]
                                                       
{'loss': 0.1411, 'grad_norm': 0.6008332371711731, 'learning_rate': 1.6915240175276465e-06, 'epoch': 2.27}

 76%|███████▌  | 10083/13317 [1:44:13<33:05,  1.63it/s]
 76%|███████▌  | 10084/13317 [1:44:13<33:06,  1.63it/s]
                                                       
{'loss': 0.1515, 'grad_norm': 0.5699983835220337, 'learning_rate': 1.690541452365922e-06, 'epoch': 2.27}

 76%|███████▌  | 10084/13317 [1:44:14<33:06,  1.63it/s]
 76%|███████▌  | 10085/13317 [1:44:14<33:04,  1.63it/s]
                                                       
{'loss': 0.169, 'grad_norm': 0.8727960586547852, 'learning_rate': 1.6895591145990408e-06, 'epoch': 2.27}

 76%|███████▌  | 10085/13317 [1:44:14<33:04,  1.63it/s]
 76%|███████▌  | 10086/13317 [1:44:15<33:03,  1.63it/s]
                                                       
{'loss': 0.1537, 'grad_norm': 0.5890971422195435, 'learning_rate': 1.6885770042944988e-06, 'epoch': 2.27}

 76%|███████▌  | 10086/13317 [1:44:15<33:03,  1.63it/s]
 76%|███████▌  | 10087/13317 [1:44:15<33:01,  1.63it/s]
                                                       
{'loss': 0.13, 'grad_norm': 0.5306887030601501, 'learning_rate': 1.6875951215197779e-06, 'epoch': 2.27}

 76%|███████▌  | 10087/13317 [1:44:15<33:01,  1.63it/s]
 76%|███████▌  | 10088/13317 [1:44:16<33:01,  1.63it/s]
                                                       
{'loss': 0.1435, 'grad_norm': 0.5681071281433105, 'learning_rate': 1.6866134663423479e-06, 'epoch': 2.27}

 76%|███████▌  | 10088/13317 [1:44:16<33:01,  1.63it/s]
 76%|███████▌  | 10089/13317 [1:44:17<33:04,  1.63it/s]
                                                       
{'loss': 0.1268, 'grad_norm': 0.4964219927787781, 'learning_rate': 1.685632038829652e-06, 'epoch': 2.27}

 76%|███████▌  | 10089/13317 [1:44:17<33:04,  1.63it/s]
 76%|███████▌  | 10090/13317 [1:44:17<33:01,  1.63it/s]
                                                       
{'loss': 0.1261, 'grad_norm': 0.47570958733558655, 'learning_rate': 1.6846508390491283e-06, 'epoch': 2.27}

 76%|███████▌  | 10090/13317 [1:44:17<33:01,  1.63it/s]
 76%|███████▌  | 10091/13317 [1:44:18<33:01,  1.63it/s]
                                                       
{'loss': 0.1335, 'grad_norm': 0.5569438338279724, 'learning_rate': 1.6836698670681973e-06, 'epoch': 2.27}

 76%|███████▌  | 10091/13317 [1:44:18<33:01,  1.63it/s]
 76%|███████▌  | 10092/13317 [1:44:18<33:00,  1.63it/s]
                                                       
{'loss': 0.1399, 'grad_norm': 0.5579673647880554, 'learning_rate': 1.6826891229542603e-06, 'epoch': 2.27}

 76%|███████▌  | 10092/13317 [1:44:18<33:00,  1.63it/s]
 76%|███████▌  | 10093/13317 [1:44:19<32:58,  1.63it/s]
                                                       
{'loss': 0.1445, 'grad_norm': 0.5282683372497559, 'learning_rate': 1.681708606774703e-06, 'epoch': 2.27}

 76%|███████▌  | 10093/13317 [1:44:19<32:58,  1.63it/s]
 76%|███████▌  | 10094/13317 [1:44:20<33:01,  1.63it/s]
                                                       
{'loss': 0.1371, 'grad_norm': 0.5273414254188538, 'learning_rate': 1.6807283185969004e-06, 'epoch': 2.27}

 76%|███████▌  | 10094/13317 [1:44:20<33:01,  1.63it/s]
 76%|███████▌  | 10095/13317 [1:44:20<32:59,  1.63it/s]
                                                       
{'loss': 0.1104, 'grad_norm': 0.4719773828983307, 'learning_rate': 1.6797482584882075e-06, 'epoch': 2.27}

 76%|███████▌  | 10095/13317 [1:44:20<32:59,  1.63it/s]
 76%|███████▌  | 10096/13317 [1:44:21<32:58,  1.63it/s]
                                                       
{'loss': 0.1447, 'grad_norm': 0.5816099643707275, 'learning_rate': 1.678768426515963e-06, 'epoch': 2.27}

 76%|███████▌  | 10096/13317 [1:44:21<32:58,  1.63it/s]
 76%|███████▌  | 10097/13317 [1:44:21<32:58,  1.63it/s]
                                                       
{'loss': 0.143, 'grad_norm': 0.5480382442474365, 'learning_rate': 1.6777888227474936e-06, 'epoch': 2.27}

 76%|███████▌  | 10097/13317 [1:44:22<32:58,  1.63it/s]
 76%|███████▌  | 10098/13317 [1:44:22<32:56,  1.63it/s]
                                                       
{'loss': 0.141, 'grad_norm': 0.5500797033309937, 'learning_rate': 1.6768094472501096e-06, 'epoch': 2.27}

 76%|███████▌  | 10098/13317 [1:44:22<32:56,  1.63it/s]
 76%|███████▌  | 10099/13317 [1:44:23<32:59,  1.63it/s]
                                                       
{'loss': 0.1321, 'grad_norm': 0.4796143174171448, 'learning_rate': 1.675830300091103e-06, 'epoch': 2.28}

 76%|███████▌  | 10099/13317 [1:44:23<32:59,  1.63it/s]
 76%|███████▌  | 10100/13317 [1:44:23<32:59,  1.63it/s]
                                                       
{'loss': 0.1185, 'grad_norm': 0.42448028922080994, 'learning_rate': 1.6748513813377503e-06, 'epoch': 2.28}

 76%|███████▌  | 10100/13317 [1:44:23<32:59,  1.63it/s]
 76%|███████▌  | 10101/13317 [1:44:24<32:58,  1.63it/s]
                                                       
{'loss': 0.1819, 'grad_norm': 0.5713872313499451, 'learning_rate': 1.6738726910573167e-06, 'epoch': 2.28}

 76%|███████▌  | 10101/13317 [1:44:24<32:58,  1.63it/s]
 76%|███████▌  | 10102/13317 [1:44:24<32:57,  1.63it/s]
                                                       
{'loss': 0.1312, 'grad_norm': 0.4993138313293457, 'learning_rate': 1.6728942293170464e-06, 'epoch': 2.28}

 76%|███████▌  | 10102/13317 [1:44:25<32:57,  1.63it/s]
 76%|███████▌  | 10103/13317 [1:44:25<32:56,  1.63it/s]
                                                       
{'loss': 0.1217, 'grad_norm': 0.47532081604003906, 'learning_rate': 1.6719159961841691e-06, 'epoch': 2.28}

 76%|███████▌  | 10103/13317 [1:44:25<32:56,  1.63it/s]
 76%|███████▌  | 10104/13317 [1:44:26<32:57,  1.63it/s]
                                                       
{'loss': 0.1411, 'grad_norm': 0.561496913433075, 'learning_rate': 1.6709379917259028e-06, 'epoch': 2.28}

 76%|███████▌  | 10104/13317 [1:44:26<32:57,  1.63it/s]
 76%|███████▌  | 10105/13317 [1:44:26<32:55,  1.63it/s]
                                                       
{'loss': 0.1389, 'grad_norm': 0.5745574235916138, 'learning_rate': 1.6699602160094436e-06, 'epoch': 2.28}

 76%|███████▌  | 10105/13317 [1:44:26<32:55,  1.63it/s]
 76%|███████▌  | 10106/13317 [1:44:27<32:55,  1.63it/s]
                                                       
{'loss': 0.1421, 'grad_norm': 0.551829993724823, 'learning_rate': 1.6689826691019778e-06, 'epoch': 2.28}

 76%|███████▌  | 10106/13317 [1:44:27<32:55,  1.63it/s]
 76%|███████▌  | 10107/13317 [1:44:28<32:53,  1.63it/s]
                                                       
{'loss': 0.1467, 'grad_norm': 0.5885811448097229, 'learning_rate': 1.6680053510706707e-06, 'epoch': 2.28}

 76%|███████▌  | 10107/13317 [1:44:28<32:53,  1.63it/s]
 76%|███████▌  | 10108/13317 [1:44:28<32:52,  1.63it/s]
                                                       
{'loss': 0.1789, 'grad_norm': 0.6457898616790771, 'learning_rate': 1.6670282619826772e-06, 'epoch': 2.28}

 76%|███████▌  | 10108/13317 [1:44:28<32:52,  1.63it/s]
 76%|███████▌  | 10109/13317 [1:44:29<32:53,  1.63it/s]
                                                       
{'loss': 0.149, 'grad_norm': 0.5910893678665161, 'learning_rate': 1.666051401905131e-06, 'epoch': 2.28}

 76%|███████▌  | 10109/13317 [1:44:29<32:53,  1.63it/s]
 76%|███████▌  | 10110/13317 [1:44:29<32:51,  1.63it/s]
                                                       
{'loss': 0.2295, 'grad_norm': 0.6296184659004211, 'learning_rate': 1.6650747709051552e-06, 'epoch': 2.28}

 76%|███████▌  | 10110/13317 [1:44:30<32:51,  1.63it/s]
 76%|███████▌  | 10111/13317 [1:44:30<32:59,  1.62it/s]
                                                       
{'loss': 0.154, 'grad_norm': 0.5555558204650879, 'learning_rate': 1.664098369049853e-06, 'epoch': 2.28}

 76%|███████▌  | 10111/13317 [1:44:30<32:59,  1.62it/s]
 76%|███████▌  | 10112/13317 [1:44:31<32:55,  1.62it/s]
                                                       
{'loss': 0.1298, 'grad_norm': 0.49102962017059326, 'learning_rate': 1.6631221964063127e-06, 'epoch': 2.28}

 76%|███████▌  | 10112/13317 [1:44:31<32:55,  1.62it/s]
 76%|███████▌  | 10113/13317 [1:44:31<32:52,  1.62it/s]
                                                       
{'loss': 0.118, 'grad_norm': 0.540603756904602, 'learning_rate': 1.6621462530416104e-06, 'epoch': 2.28}

 76%|███████▌  | 10113/13317 [1:44:31<32:52,  1.62it/s]
 76%|███████▌  | 10114/13317 [1:44:32<32:52,  1.62it/s]
                                                       
{'loss': 0.1892, 'grad_norm': 0.675564706325531, 'learning_rate': 1.6611705390228011e-06, 'epoch': 2.28}

 76%|███████▌  | 10114/13317 [1:44:32<32:52,  1.62it/s]
 76%|███████▌  | 10115/13317 [1:44:33<34:30,  1.55it/s]
                                                       
{'loss': 0.1694, 'grad_norm': 0.6148865222930908, 'learning_rate': 1.6601950544169294e-06, 'epoch': 2.28}

 76%|███████▌  | 10115/13317 [1:44:33<34:30,  1.55it/s]
 76%|███████▌  | 10116/13317 [1:44:33<33:58,  1.57it/s]
                                                       
{'loss': 0.1582, 'grad_norm': 0.5727320909500122, 'learning_rate': 1.6592197992910175e-06, 'epoch': 2.28}

 76%|███████▌  | 10116/13317 [1:44:33<33:58,  1.57it/s]
 76%|███████▌  | 10117/13317 [1:44:34<33:35,  1.59it/s]
                                                       
{'loss': 0.1688, 'grad_norm': 0.6172579526901245, 'learning_rate': 1.6582447737120804e-06, 'epoch': 2.28}

 76%|███████▌  | 10117/13317 [1:44:34<33:35,  1.59it/s]
 76%|███████▌  | 10118/13317 [1:44:34<33:19,  1.60it/s]
                                                       
{'loss': 0.1436, 'grad_norm': 0.5169343948364258, 'learning_rate': 1.6572699777471108e-06, 'epoch': 2.28}

 76%|███████▌  | 10118/13317 [1:44:35<33:19,  1.60it/s]
 76%|███████▌  | 10119/13317 [1:44:35<33:08,  1.61it/s]
                                                       
{'loss': 0.132, 'grad_norm': 0.42563536763191223, 'learning_rate': 1.6562954114630847e-06, 'epoch': 2.28}

 76%|███████▌  | 10119/13317 [1:44:35<33:08,  1.61it/s]
 76%|███████▌  | 10120/13317 [1:44:36<32:59,  1.62it/s]
                                                       
{'loss': 0.1611, 'grad_norm': 0.7251240611076355, 'learning_rate': 1.6553210749269694e-06, 'epoch': 2.28}

 76%|███████▌  | 10120/13317 [1:44:36<32:59,  1.62it/s]
 76%|███████▌  | 10121/13317 [1:44:36<32:53,  1.62it/s]
                                                       
{'loss': 0.1857, 'grad_norm': 0.6577488780021667, 'learning_rate': 1.6543469682057105e-06, 'epoch': 2.28}

 76%|███████▌  | 10121/13317 [1:44:36<32:53,  1.62it/s]
 76%|███████▌  | 10122/13317 [1:44:37<32:49,  1.62it/s]
                                                       
{'loss': 0.2696, 'grad_norm': 0.6885810494422913, 'learning_rate': 1.6533730913662376e-06, 'epoch': 2.28}

 76%|███████▌  | 10122/13317 [1:44:37<32:49,  1.62it/s]
 76%|███████▌  | 10123/13317 [1:44:38<32:45,  1.62it/s]
                                                       
{'loss': 0.1459, 'grad_norm': 0.5162796378135681, 'learning_rate': 1.6523994444754677e-06, 'epoch': 2.28}

 76%|███████▌  | 10123/13317 [1:44:38<32:45,  1.62it/s]
 76%|███████▌  | 10124/13317 [1:44:38<32:43,  1.63it/s]
                                                       
{'loss': 0.1618, 'grad_norm': 0.6938615441322327, 'learning_rate': 1.6514260276003025e-06, 'epoch': 2.28}

 76%|███████▌  | 10124/13317 [1:44:38<32:43,  1.63it/s]
 76%|███████▌  | 10125/13317 [1:44:39<32:42,  1.63it/s]
                                                       
{'loss': 0.1435, 'grad_norm': 0.5398364663124084, 'learning_rate': 1.6504528408076249e-06, 'epoch': 2.28}

 76%|███████▌  | 10125/13317 [1:44:39<32:42,  1.63it/s]
 76%|███████▌  | 10126/13317 [1:44:39<32:48,  1.62it/s]
                                                       
{'loss': 0.1535, 'grad_norm': 0.5075201988220215, 'learning_rate': 1.6494798841643006e-06, 'epoch': 2.28}

 76%|███████▌  | 10126/13317 [1:44:40<32:48,  1.62it/s]
 76%|███████▌  | 10127/13317 [1:44:40<34:25,  1.54it/s]
                                                       
{'loss': 0.1686, 'grad_norm': 0.6747465133666992, 'learning_rate': 1.6485071577371863e-06, 'epoch': 2.28}

 76%|███████▌  | 10127/13317 [1:44:40<34:25,  1.54it/s]
 76%|███████▌  | 10128/13317 [1:44:41<33:53,  1.57it/s]
                                                       
{'loss': 0.1222, 'grad_norm': 0.4614529609680176, 'learning_rate': 1.647534661593116e-06, 'epoch': 2.28}

 76%|███████▌  | 10128/13317 [1:44:41<33:53,  1.57it/s]
 76%|███████▌  | 10129/13317 [1:44:41<33:40,  1.58it/s]
                                                       
{'loss': 0.1687, 'grad_norm': 0.6202313303947449, 'learning_rate': 1.6465623957989097e-06, 'epoch': 2.28}

 76%|███████▌  | 10129/13317 [1:44:41<33:40,  1.58it/s]
 76%|███████▌  | 10130/13317 [1:44:42<33:20,  1.59it/s]
                                                       
{'loss': 0.1428, 'grad_norm': 0.5709989070892334, 'learning_rate': 1.645590360421373e-06, 'epoch': 2.28}

 76%|███████▌  | 10130/13317 [1:44:42<33:20,  1.59it/s]
 76%|███████▌  | 10131/13317 [1:44:43<33:06,  1.60it/s]
                                                       
{'loss': 0.1639, 'grad_norm': 0.5840112566947937, 'learning_rate': 1.6446185555272975e-06, 'epoch': 2.28}

 76%|███████▌  | 10131/13317 [1:44:43<33:06,  1.60it/s]
 76%|███████▌  | 10132/13317 [1:44:43<32:56,  1.61it/s]
                                                       
{'loss': 0.1727, 'grad_norm': 0.6215143799781799, 'learning_rate': 1.643646981183455e-06, 'epoch': 2.28}

 76%|███████▌  | 10132/13317 [1:44:43<32:56,  1.61it/s]
 76%|███████▌  | 10133/13317 [1:44:44<32:49,  1.62it/s]
                                                       
{'loss': 0.166, 'grad_norm': 0.5506106019020081, 'learning_rate': 1.6426756374566005e-06, 'epoch': 2.28}

 76%|███████▌  | 10133/13317 [1:44:44<32:49,  1.62it/s]
 76%|███████▌  | 10134/13317 [1:44:44<32:45,  1.62it/s]
                                                       
{'loss': 0.1259, 'grad_norm': 0.46928277611732483, 'learning_rate': 1.6417045244134776e-06, 'epoch': 2.28}

 76%|███████▌  | 10134/13317 [1:44:45<32:45,  1.62it/s]
 76%|███████▌  | 10135/13317 [1:44:45<32:40,  1.62it/s]
                                                       
{'loss': 0.1536, 'grad_norm': 0.5277381539344788, 'learning_rate': 1.6407336421208154e-06, 'epoch': 2.28}

 76%|███████▌  | 10135/13317 [1:44:45<32:40,  1.62it/s]
 76%|███████▌  | 10136/13317 [1:44:46<32:36,  1.63it/s]
                                                       
{'loss': 0.1844, 'grad_norm': 0.8232296109199524, 'learning_rate': 1.639762990645317e-06, 'epoch': 2.28}

 76%|███████▌  | 10136/13317 [1:44:46<32:36,  1.63it/s]
 76%|███████▌  | 10137/13317 [1:44:46<32:37,  1.62it/s]
                                                       
{'loss': 0.1626, 'grad_norm': 0.5936422944068909, 'learning_rate': 1.63879257005368e-06, 'epoch': 2.28}

 76%|███████▌  | 10137/13317 [1:44:46<32:37,  1.62it/s]
 76%|███████▌  | 10138/13317 [1:44:47<32:33,  1.63it/s]
                                                       
{'loss': 0.1879, 'grad_norm': 0.6315288543701172, 'learning_rate': 1.6378223804125843e-06, 'epoch': 2.28}

 76%|███████▌  | 10138/13317 [1:44:47<32:33,  1.63it/s]
 76%|███████▌  | 10139/13317 [1:44:47<32:33,  1.63it/s]
                                                       
{'loss': 0.1111, 'grad_norm': 0.48236584663391113, 'learning_rate': 1.6368524217886906e-06, 'epoch': 2.28}

 76%|███████▌  | 10139/13317 [1:44:48<32:33,  1.63it/s]
 76%|███████▌  | 10140/13317 [1:44:48<32:31,  1.63it/s]
                                                       
{'loss': 0.1316, 'grad_norm': 0.5280783176422119, 'learning_rate': 1.6358826942486427e-06, 'epoch': 2.28}

 76%|███████▌  | 10140/13317 [1:44:48<32:31,  1.63it/s]
 76%|███████▌  | 10141/13317 [1:44:49<32:30,  1.63it/s]
                                                       
{'loss': 0.1474, 'grad_norm': 0.5631553530693054, 'learning_rate': 1.6349131978590743e-06, 'epoch': 2.28}

 76%|███████▌  | 10141/13317 [1:44:49<32:30,  1.63it/s]
 76%|███████▌  | 10142/13317 [1:44:49<32:32,  1.63it/s]
                                                       
{'loss': 0.1095, 'grad_norm': 0.41386377811431885, 'learning_rate': 1.6339439326866024e-06, 'epoch': 2.28}

 76%|███████▌  | 10142/13317 [1:44:49<32:32,  1.63it/s]
 76%|███████▌  | 10143/13317 [1:44:50<32:31,  1.63it/s]
                                                       
{'loss': 0.1531, 'grad_norm': 0.5780370831489563, 'learning_rate': 1.632974898797819e-06, 'epoch': 2.28}

 76%|███████▌  | 10143/13317 [1:44:50<32:31,  1.63it/s]
 76%|███████▌  | 10144/13317 [1:44:51<32:30,  1.63it/s]
                                                       
{'loss': 0.1649, 'grad_norm': 0.6296884417533875, 'learning_rate': 1.632006096259311e-06, 'epoch': 2.29}

 76%|███████▌  | 10144/13317 [1:44:51<32:30,  1.63it/s]
 76%|███████▌  | 10145/13317 [1:44:51<32:29,  1.63it/s]
                                                       
{'loss': 0.1328, 'grad_norm': 0.5041541457176208, 'learning_rate': 1.631037525137647e-06, 'epoch': 2.29}

 76%|███████▌  | 10145/13317 [1:44:51<32:29,  1.63it/s]
 76%|███████▌  | 10146/13317 [1:44:52<32:28,  1.63it/s]
                                                       
{'loss': 0.113, 'grad_norm': 0.39091846346855164, 'learning_rate': 1.6300691854993755e-06, 'epoch': 2.29}

 76%|███████▌  | 10146/13317 [1:44:52<32:28,  1.63it/s]
 76%|███████▌  | 10147/13317 [1:44:52<32:30,  1.62it/s]
                                                       
{'loss': 0.1336, 'grad_norm': 0.5025199055671692, 'learning_rate': 1.629101077411031e-06, 'epoch': 2.29}

 76%|███████▌  | 10147/13317 [1:44:53<32:30,  1.62it/s]
 76%|███████▌  | 10148/13317 [1:44:53<32:29,  1.63it/s]
                                                       
{'loss': 0.2036, 'grad_norm': 0.6304013729095459, 'learning_rate': 1.628133200939136e-06, 'epoch': 2.29}

 76%|███████▌  | 10148/13317 [1:44:53<32:29,  1.63it/s]
 76%|███████▌  | 10149/13317 [1:44:54<32:29,  1.62it/s]
                                                       
{'loss': 0.154, 'grad_norm': 0.6238538026809692, 'learning_rate': 1.627165556150192e-06, 'epoch': 2.29}

 76%|███████▌  | 10149/13317 [1:44:54<32:29,  1.62it/s]
 76%|███████▌  | 10150/13317 [1:44:54<32:26,  1.63it/s]
                                                       
{'loss': 0.1221, 'grad_norm': 0.4996837079524994, 'learning_rate': 1.626198143110685e-06, 'epoch': 2.29}

 76%|███████▌  | 10150/13317 [1:44:54<32:26,  1.63it/s]
 76%|███████▌  | 10151/13317 [1:44:55<32:24,  1.63it/s]
                                                       
{'loss': 0.1866, 'grad_norm': 0.6594658493995667, 'learning_rate': 1.6252309618870871e-06, 'epoch': 2.29}

 76%|███████▌  | 10151/13317 [1:44:55<32:24,  1.63it/s]
 76%|███████▌  | 10152/13317 [1:44:55<32:26,  1.63it/s]
                                                       
{'loss': 0.1584, 'grad_norm': 0.5788140892982483, 'learning_rate': 1.624264012545857e-06, 'epoch': 2.29}

 76%|███████▌  | 10152/13317 [1:44:56<32:26,  1.63it/s]
 76%|███████▌  | 10153/13317 [1:44:56<32:22,  1.63it/s]
                                                       
{'loss': 0.1329, 'grad_norm': 0.4856848120689392, 'learning_rate': 1.623297295153432e-06, 'epoch': 2.29}

 76%|███████▌  | 10153/13317 [1:44:56<32:22,  1.63it/s]
 76%|███████▌  | 10154/13317 [1:44:57<32:25,  1.63it/s]
                                                       
{'loss': 0.1848, 'grad_norm': 0.6999034881591797, 'learning_rate': 1.622330809776234e-06, 'epoch': 2.29}

 76%|███████▌  | 10154/13317 [1:44:57<32:25,  1.63it/s]
 76%|███████▋  | 10155/13317 [1:44:57<32:23,  1.63it/s]
                                                       
{'loss': 0.1424, 'grad_norm': 0.5419288277626038, 'learning_rate': 1.6213645564806751e-06, 'epoch': 2.29}

 76%|███████▋  | 10155/13317 [1:44:57<32:23,  1.63it/s]
 76%|███████▋  | 10156/13317 [1:44:58<32:21,  1.63it/s]
                                                       
{'loss': 0.1389, 'grad_norm': 0.5013055205345154, 'learning_rate': 1.6203985353331426e-06, 'epoch': 2.29}

 76%|███████▋  | 10156/13317 [1:44:58<32:21,  1.63it/s]
 76%|███████▋  | 10157/13317 [1:44:59<32:23,  1.63it/s]
                                                       
{'loss': 0.1618, 'grad_norm': 0.5847048163414001, 'learning_rate': 1.6194327464000159e-06, 'epoch': 2.29}

 76%|███████▋  | 10157/13317 [1:44:59<32:23,  1.63it/s]
 76%|███████▋  | 10158/13317 [1:44:59<32:21,  1.63it/s]
                                                       
{'loss': 0.1504, 'grad_norm': 0.5545573830604553, 'learning_rate': 1.6184671897476522e-06, 'epoch': 2.29}

 76%|███████▋  | 10158/13317 [1:44:59<32:21,  1.63it/s]
 76%|███████▋  | 10159/13317 [1:45:00<32:22,  1.63it/s]
                                                       
{'loss': 0.2083, 'grad_norm': 0.7217549085617065, 'learning_rate': 1.6175018654423986e-06, 'epoch': 2.29}

 76%|███████▋  | 10159/13317 [1:45:00<32:22,  1.63it/s]
 76%|███████▋  | 10160/13317 [1:45:00<32:19,  1.63it/s]
                                                       
{'loss': 0.1724, 'grad_norm': 0.674090564250946, 'learning_rate': 1.616536773550581e-06, 'epoch': 2.29}

 76%|███████▋  | 10160/13317 [1:45:00<32:19,  1.63it/s]
 76%|███████▋  | 10161/13317 [1:45:01<32:17,  1.63it/s]
                                                       
{'loss': 0.1636, 'grad_norm': 0.6334512233734131, 'learning_rate': 1.6155719141385107e-06, 'epoch': 2.29}

 76%|███████▋  | 10161/13317 [1:45:01<32:17,  1.63it/s]
 76%|███████▋  | 10162/13317 [1:45:02<32:25,  1.62it/s]
                                                       
{'loss': 0.1245, 'grad_norm': 0.4772702753543854, 'learning_rate': 1.6146072872724855e-06, 'epoch': 2.29}

 76%|███████▋  | 10162/13317 [1:45:02<32:25,  1.62it/s]
 76%|███████▋  | 10163/13317 [1:45:02<32:21,  1.62it/s]
                                                       
{'loss': 0.129, 'grad_norm': 0.4848686158657074, 'learning_rate': 1.6136428930187835e-06, 'epoch': 2.29}

 76%|███████▋  | 10163/13317 [1:45:02<32:21,  1.62it/s]
 76%|███████▋  | 10164/13317 [1:45:03<32:20,  1.62it/s]
                                                       
{'loss': 0.1142, 'grad_norm': 0.4541091322898865, 'learning_rate': 1.6126787314436721e-06, 'epoch': 2.29}

 76%|███████▋  | 10164/13317 [1:45:03<32:20,  1.62it/s]
 76%|███████▋  | 10165/13317 [1:45:03<32:18,  1.63it/s]
                                                       
{'loss': 0.1323, 'grad_norm': 0.5681717991828918, 'learning_rate': 1.6117148026133972e-06, 'epoch': 2.29}

 76%|███████▋  | 10165/13317 [1:45:04<32:18,  1.63it/s]
 76%|███████▋  | 10166/13317 [1:45:04<32:16,  1.63it/s]
                                                       
{'loss': 0.1311, 'grad_norm': 0.4625359773635864, 'learning_rate': 1.6107511065941894e-06, 'epoch': 2.29}

 76%|███████▋  | 10166/13317 [1:45:04<32:16,  1.63it/s]
 76%|███████▋  | 10167/13317 [1:45:05<32:18,  1.63it/s]
                                                       
{'loss': 0.1704, 'grad_norm': 0.5832375884056091, 'learning_rate': 1.609787643452268e-06, 'epoch': 2.29}

 76%|███████▋  | 10167/13317 [1:45:05<32:18,  1.63it/s]
 76%|███████▋  | 10168/13317 [1:45:05<32:16,  1.63it/s]
                                                       
{'loss': 0.1826, 'grad_norm': 0.6335359811782837, 'learning_rate': 1.60882441325383e-06, 'epoch': 2.29}

 76%|███████▋  | 10168/13317 [1:45:05<32:16,  1.63it/s]
 76%|███████▋  | 10169/13317 [1:45:06<32:15,  1.63it/s]
                                                       
{'loss': 0.1194, 'grad_norm': 0.4750259816646576, 'learning_rate': 1.607861416065063e-06, 'epoch': 2.29}

 76%|███████▋  | 10169/13317 [1:45:06<32:15,  1.63it/s]
 76%|███████▋  | 10170/13317 [1:45:07<32:13,  1.63it/s]
                                                       
{'loss': 0.1583, 'grad_norm': 0.5510023236274719, 'learning_rate': 1.6068986519521318e-06, 'epoch': 2.29}

 76%|███████▋  | 10170/13317 [1:45:07<32:13,  1.63it/s]
 76%|███████▋  | 10171/13317 [1:45:07<32:12,  1.63it/s]
                                                       
{'loss': 0.1467, 'grad_norm': 0.5456913113594055, 'learning_rate': 1.6059361209811914e-06, 'epoch': 2.29}

 76%|███████▋  | 10171/13317 [1:45:07<32:12,  1.63it/s]
 76%|███████▋  | 10172/13317 [1:45:08<32:14,  1.63it/s]
                                                       
{'loss': 0.1331, 'grad_norm': 0.5055966377258301, 'learning_rate': 1.604973823218376e-06, 'epoch': 2.29}

 76%|███████▋  | 10172/13317 [1:45:08<32:14,  1.63it/s]
 76%|███████▋  | 10173/13317 [1:45:08<32:12,  1.63it/s]
                                                       
{'loss': 0.1489, 'grad_norm': 0.5755211710929871, 'learning_rate': 1.604011758729805e-06, 'epoch': 2.29}

 76%|███████▋  | 10173/13317 [1:45:08<32:12,  1.63it/s]
 76%|███████▋  | 10174/13317 [1:45:09<32:13,  1.63it/s]
                                                       
{'loss': 0.1241, 'grad_norm': 0.48245981335639954, 'learning_rate': 1.6030499275815848e-06, 'epoch': 2.29}

 76%|███████▋  | 10174/13317 [1:45:09<32:13,  1.63it/s]
 76%|███████▋  | 10175/13317 [1:45:10<32:12,  1.63it/s]
                                                       
{'loss': 0.1169, 'grad_norm': 0.5072972178459167, 'learning_rate': 1.6020883298398006e-06, 'epoch': 2.29}

 76%|███████▋  | 10175/13317 [1:45:10<32:12,  1.63it/s]
 76%|███████▋  | 10176/13317 [1:45:10<32:09,  1.63it/s]
                                                       
{'loss': 0.1603, 'grad_norm': 0.5846529006958008, 'learning_rate': 1.601126965570527e-06, 'epoch': 2.29}

 76%|███████▋  | 10176/13317 [1:45:10<32:09,  1.63it/s]
 76%|███████▋  | 10177/13317 [1:45:11<32:11,  1.63it/s]
                                                       
{'loss': 0.1325, 'grad_norm': 0.5298675298690796, 'learning_rate': 1.6001658348398174e-06, 'epoch': 2.29}

 76%|███████▋  | 10177/13317 [1:45:11<32:11,  1.63it/s]
 76%|███████▋  | 10178/13317 [1:45:11<32:09,  1.63it/s]
                                                       
{'loss': 0.1429, 'grad_norm': 0.5276377201080322, 'learning_rate': 1.5992049377137147e-06, 'epoch': 2.29}

 76%|███████▋  | 10178/13317 [1:45:12<32:09,  1.63it/s]
 76%|███████▋  | 10179/13317 [1:45:12<32:08,  1.63it/s]
                                                       
{'loss': 0.1541, 'grad_norm': 0.5148496627807617, 'learning_rate': 1.598244274258241e-06, 'epoch': 2.29}

 76%|███████▋  | 10179/13317 [1:45:12<32:08,  1.63it/s]
 76%|███████▋  | 10180/13317 [1:45:13<32:06,  1.63it/s]
                                                       
{'loss': 0.1285, 'grad_norm': 0.46624502539634705, 'learning_rate': 1.5972838445394024e-06, 'epoch': 2.29}

 76%|███████▋  | 10180/13317 [1:45:13<32:06,  1.63it/s]
 76%|███████▋  | 10181/13317 [1:45:13<32:06,  1.63it/s]
                                                       
{'loss': 0.1754, 'grad_norm': 0.7772746086120605, 'learning_rate': 1.5963236486231936e-06, 'epoch': 2.29}

 76%|███████▋  | 10181/13317 [1:45:13<32:06,  1.63it/s]
 76%|███████▋  | 10182/13317 [1:45:14<32:04,  1.63it/s]
                                                       
{'loss': 0.1579, 'grad_norm': 0.6056097149848938, 'learning_rate': 1.595363686575589e-06, 'epoch': 2.29}

 76%|███████▋  | 10182/13317 [1:45:14<32:04,  1.63it/s]
 76%|███████▋  | 10183/13317 [1:45:14<32:02,  1.63it/s]
                                                       
{'loss': 0.1763, 'grad_norm': 0.7043097615242004, 'learning_rate': 1.594403958462546e-06, 'epoch': 2.29}

 76%|███████▋  | 10183/13317 [1:45:15<32:02,  1.63it/s]
 76%|███████▋  | 10184/13317 [1:45:15<32:03,  1.63it/s]
                                                       
{'loss': 0.1678, 'grad_norm': 0.749316930770874, 'learning_rate': 1.5934444643500107e-06, 'epoch': 2.29}

 76%|███████▋  | 10184/13317 [1:45:15<32:03,  1.63it/s]
 76%|███████▋  | 10185/13317 [1:45:16<32:01,  1.63it/s]
                                                       
{'loss': 0.0993, 'grad_norm': 0.39597412943840027, 'learning_rate': 1.5924852043039113e-06, 'epoch': 2.29}

 76%|███████▋  | 10185/13317 [1:45:16<32:01,  1.63it/s]
 76%|███████▋  | 10186/13317 [1:45:16<32:01,  1.63it/s]
                                                       
{'loss': 0.1418, 'grad_norm': 0.5362234115600586, 'learning_rate': 1.5915261783901576e-06, 'epoch': 2.29}

 76%|███████▋  | 10186/13317 [1:45:16<32:01,  1.63it/s]
 76%|███████▋  | 10187/13317 [1:45:17<32:01,  1.63it/s]
                                                       
{'loss': 0.1327, 'grad_norm': 0.42798784375190735, 'learning_rate': 1.590567386674643e-06, 'epoch': 2.29}

 76%|███████▋  | 10187/13317 [1:45:17<32:01,  1.63it/s]
 77%|███████▋  | 10188/13317 [1:45:18<32:00,  1.63it/s]
                                                       
{'loss': 0.1455, 'grad_norm': 0.5735514163970947, 'learning_rate': 1.5896088292232492e-06, 'epoch': 2.3}

 77%|███████▋  | 10188/13317 [1:45:18<32:00,  1.63it/s]
 77%|███████▋  | 10189/13317 [1:45:18<32:02,  1.63it/s]
                                                       
{'loss': 0.1392, 'grad_norm': 0.5652514100074768, 'learning_rate': 1.5886505061018415e-06, 'epoch': 2.3}

 77%|███████▋  | 10189/13317 [1:45:18<32:02,  1.63it/s]
 77%|███████▋  | 10190/13317 [1:45:19<32:01,  1.63it/s]
                                                       
{'loss': 0.1492, 'grad_norm': 0.5670727491378784, 'learning_rate': 1.587692417376261e-06, 'epoch': 2.3}

 77%|███████▋  | 10190/13317 [1:45:19<32:01,  1.63it/s]
 77%|███████▋  | 10191/13317 [1:45:19<32:00,  1.63it/s]
                                                       
{'loss': 0.146, 'grad_norm': 0.5092431306838989, 'learning_rate': 1.586734563112341e-06, 'epoch': 2.3}

 77%|███████▋  | 10191/13317 [1:45:20<32:00,  1.63it/s]
 77%|███████▋  | 10192/13317 [1:45:20<31:58,  1.63it/s]
                                                       
{'loss': 0.1509, 'grad_norm': 0.5436812043190002, 'learning_rate': 1.585776943375899e-06, 'epoch': 2.3}

 77%|███████▋  | 10192/13317 [1:45:20<31:58,  1.63it/s]
 77%|███████▋  | 10193/13317 [1:45:21<31:57,  1.63it/s]
                                                       
{'loss': 0.1594, 'grad_norm': 0.537223756313324, 'learning_rate': 1.5848195582327308e-06, 'epoch': 2.3}

 77%|███████▋  | 10193/13317 [1:45:21<31:57,  1.63it/s]
 77%|███████▋  | 10194/13317 [1:45:21<32:00,  1.63it/s]
                                                       
{'loss': 0.1331, 'grad_norm': 0.47801724076271057, 'learning_rate': 1.5838624077486186e-06, 'epoch': 2.3}

 77%|███████▋  | 10194/13317 [1:45:21<32:00,  1.63it/s]
 77%|███████▋  | 10195/13317 [1:45:22<31:58,  1.63it/s]
                                                       
{'loss': 0.1444, 'grad_norm': 0.5850033760070801, 'learning_rate': 1.5829054919893288e-06, 'epoch': 2.3}

 77%|███████▋  | 10195/13317 [1:45:22<31:58,  1.63it/s]
 77%|███████▋  | 10196/13317 [1:45:22<31:57,  1.63it/s]
                                                       
{'loss': 0.1389, 'grad_norm': 0.4969499707221985, 'learning_rate': 1.5819488110206161e-06, 'epoch': 2.3}

 77%|███████▋  | 10196/13317 [1:45:23<31:57,  1.63it/s]
 77%|███████▋  | 10197/13317 [1:45:23<31:59,  1.63it/s]
                                                       
{'loss': 0.1482, 'grad_norm': 0.6035985350608826, 'learning_rate': 1.5809923649082076e-06, 'epoch': 2.3}

 77%|███████▋  | 10197/13317 [1:45:23<31:59,  1.63it/s]
 77%|███████▋  | 10198/13317 [1:45:24<31:58,  1.63it/s]
                                                       
{'loss': 0.1352, 'grad_norm': 0.5031452178955078, 'learning_rate': 1.580036153717825e-06, 'epoch': 2.3}

 77%|███████▋  | 10198/13317 [1:45:24<31:58,  1.63it/s]
 77%|███████▋  | 10199/13317 [1:45:24<31:58,  1.63it/s]
                                                       
{'loss': 0.1734, 'grad_norm': 0.6346666216850281, 'learning_rate': 1.5790801775151714e-06, 'epoch': 2.3}

 77%|███████▋  | 10199/13317 [1:45:24<31:58,  1.63it/s]
 77%|███████▋  | 10200/13317 [1:45:25<31:56,  1.63it/s]
                                                       
{'loss': 0.1401, 'grad_norm': 0.5038657784461975, 'learning_rate': 1.5781244363659304e-06, 'epoch': 2.3}

 77%|███████▋  | 10200/13317 [1:45:25<31:56,  1.63it/s]
 77%|███████▋  | 10201/13317 [1:45:26<31:54,  1.63it/s]
                                                       
{'loss': 0.155, 'grad_norm': 0.5612280964851379, 'learning_rate': 1.5771689303357707e-06, 'epoch': 2.3}

 77%|███████▋  | 10201/13317 [1:45:26<31:54,  1.63it/s]
 77%|███████▋  | 10202/13317 [1:45:26<31:56,  1.63it/s]
                                                       
{'loss': 0.1364, 'grad_norm': 0.5678194165229797, 'learning_rate': 1.576213659490347e-06, 'epoch': 2.3}

 77%|███████▋  | 10202/13317 [1:45:26<31:56,  1.63it/s]
 77%|███████▋  | 10203/13317 [1:45:27<31:53,  1.63it/s]
                                                       
{'loss': 0.1296, 'grad_norm': 0.5161104202270508, 'learning_rate': 1.5752586238953e-06, 'epoch': 2.3}

 77%|███████▋  | 10203/13317 [1:45:27<31:53,  1.63it/s]
 77%|███████▋  | 10204/13317 [1:45:27<31:54,  1.63it/s]
                                                       
{'loss': 0.1405, 'grad_norm': 0.6451393961906433, 'learning_rate': 1.574303823616244e-06, 'epoch': 2.3}

 77%|███████▋  | 10204/13317 [1:45:28<31:54,  1.63it/s]
 77%|███████▋  | 10205/13317 [1:45:28<31:52,  1.63it/s]
                                                       
{'loss': 0.1042, 'grad_norm': 0.43031707406044006, 'learning_rate': 1.573349258718787e-06, 'epoch': 2.3}

 77%|███████▋  | 10205/13317 [1:45:28<31:52,  1.63it/s]
 77%|███████▋  | 10206/13317 [1:45:29<31:51,  1.63it/s]
                                                       
{'loss': 0.1609, 'grad_norm': 0.6092628240585327, 'learning_rate': 1.5723949292685193e-06, 'epoch': 2.3}

 77%|███████▋  | 10206/13317 [1:45:29<31:51,  1.63it/s]
 77%|███████▋  | 10207/13317 [1:45:29<31:53,  1.63it/s]
                                                       
{'loss': 0.1415, 'grad_norm': 0.5478536486625671, 'learning_rate': 1.5714408353310123e-06, 'epoch': 2.3}

 77%|███████▋  | 10207/13317 [1:45:29<31:53,  1.63it/s]
 77%|███████▋  | 10208/13317 [1:45:30<31:50,  1.63it/s]
                                                       
{'loss': 0.1526, 'grad_norm': 0.5942792296409607, 'learning_rate': 1.57048697697182e-06, 'epoch': 2.3}

 77%|███████▋  | 10208/13317 [1:45:30<31:50,  1.63it/s]
 77%|███████▋  | 10209/13317 [1:45:30<31:51,  1.63it/s]
                                                       
{'loss': 0.1759, 'grad_norm': 0.6347354054450989, 'learning_rate': 1.5695333542564867e-06, 'epoch': 2.3}

 77%|███████▋  | 10209/13317 [1:45:31<31:51,  1.63it/s]
 77%|███████▋  | 10210/13317 [1:45:31<31:49,  1.63it/s]
                                                       
{'loss': 0.1353, 'grad_norm': 0.49990248680114746, 'learning_rate': 1.568579967250533e-06, 'epoch': 2.3}

 77%|███████▋  | 10210/13317 [1:45:31<31:49,  1.63it/s]
 77%|███████▋  | 10211/13317 [1:45:32<31:48,  1.63it/s]
                                                       
{'loss': 0.1469, 'grad_norm': 0.5991837978363037, 'learning_rate': 1.5676268160194692e-06, 'epoch': 2.3}

 77%|███████▋  | 10211/13317 [1:45:32<31:48,  1.63it/s]
 77%|███████▋  | 10212/13317 [1:45:32<31:49,  1.63it/s]
                                                       
{'loss': 0.1891, 'grad_norm': 0.6919810771942139, 'learning_rate': 1.566673900628784e-06, 'epoch': 2.3}

 77%|███████▋  | 10212/13317 [1:45:32<31:49,  1.63it/s]
 77%|███████▋  | 10213/13317 [1:45:33<31:46,  1.63it/s]
                                                       
{'loss': 0.1507, 'grad_norm': 0.5432010889053345, 'learning_rate': 1.5657212211439565e-06, 'epoch': 2.3}

 77%|███████▋  | 10213/13317 [1:45:33<31:46,  1.63it/s]
 77%|███████▋  | 10214/13317 [1:45:34<31:46,  1.63it/s]
                                                       
{'loss': 0.1466, 'grad_norm': 0.5066789984703064, 'learning_rate': 1.5647687776304432e-06, 'epoch': 2.3}

 77%|███████▋  | 10214/13317 [1:45:34<31:46,  1.63it/s]
 77%|███████▋  | 10215/13317 [1:45:34<31:45,  1.63it/s]
                                                       
{'loss': 0.1538, 'grad_norm': 0.610893726348877, 'learning_rate': 1.5638165701536866e-06, 'epoch': 2.3}

 77%|███████▋  | 10215/13317 [1:45:34<31:45,  1.63it/s]
 77%|███████▋  | 10216/13317 [1:45:35<31:43,  1.63it/s]
                                                       
{'loss': 0.1455, 'grad_norm': 0.568861722946167, 'learning_rate': 1.5628645987791157e-06, 'epoch': 2.3}

 77%|███████▋  | 10216/13317 [1:45:35<31:43,  1.63it/s]
 77%|███████▋  | 10217/13317 [1:45:35<31:42,  1.63it/s]
                                                       
{'loss': 0.1387, 'grad_norm': 0.5397898554801941, 'learning_rate': 1.5619128635721386e-06, 'epoch': 2.3}

 77%|███████▋  | 10217/13317 [1:45:36<31:42,  1.63it/s]
 77%|███████▋  | 10218/13317 [1:45:36<31:42,  1.63it/s]
                                                       
{'loss': 0.1238, 'grad_norm': 0.46399760246276855, 'learning_rate': 1.5609613645981524e-06, 'epoch': 2.3}

 77%|███████▋  | 10218/13317 [1:45:36<31:42,  1.63it/s]
 77%|███████▋  | 10219/13317 [1:45:37<31:42,  1.63it/s]
                                                       
{'loss': 0.1418, 'grad_norm': 0.5133104920387268, 'learning_rate': 1.5600101019225322e-06, 'epoch': 2.3}

 77%|███████▋  | 10219/13317 [1:45:37<31:42,  1.63it/s]
 77%|███████▋  | 10220/13317 [1:45:37<31:41,  1.63it/s]
                                                       
{'loss': 0.1335, 'grad_norm': 0.5534493923187256, 'learning_rate': 1.5590590756106423e-06, 'epoch': 2.3}

 77%|███████▋  | 10220/13317 [1:45:37<31:41,  1.63it/s]
 77%|███████▋  | 10221/13317 [1:45:38<31:41,  1.63it/s]
                                                       
{'loss': 0.1615, 'grad_norm': 0.6060440540313721, 'learning_rate': 1.5581082857278273e-06, 'epoch': 2.3}

 77%|███████▋  | 10221/13317 [1:45:38<31:41,  1.63it/s]
 77%|███████▋  | 10222/13317 [1:45:38<31:42,  1.63it/s]
                                                       
{'loss': 0.1413, 'grad_norm': 0.5092533826828003, 'learning_rate': 1.5571577323394149e-06, 'epoch': 2.3}

 77%|███████▋  | 10222/13317 [1:45:39<31:42,  1.63it/s]
 77%|███████▋  | 10223/13317 [1:45:39<31:40,  1.63it/s]
                                                       
{'loss': 0.1524, 'grad_norm': 0.5408033132553101, 'learning_rate': 1.5562074155107216e-06, 'epoch': 2.3}

 77%|███████▋  | 10223/13317 [1:45:39<31:40,  1.63it/s]
 77%|███████▋  | 10224/13317 [1:45:40<31:41,  1.63it/s]
                                                       
{'loss': 0.135, 'grad_norm': 0.4681439995765686, 'learning_rate': 1.55525733530704e-06, 'epoch': 2.3}

 77%|███████▋  | 10224/13317 [1:45:40<31:41,  1.63it/s]
 77%|███████▋  | 10225/13317 [1:45:40<31:39,  1.63it/s]
                                                       
{'loss': 0.1424, 'grad_norm': 0.8777173757553101, 'learning_rate': 1.554307491793655e-06, 'epoch': 2.3}

 77%|███████▋  | 10225/13317 [1:45:40<31:39,  1.63it/s]
 77%|███████▋  | 10226/13317 [1:45:41<31:37,  1.63it/s]
                                                       
{'loss': 0.1463, 'grad_norm': 0.5332685708999634, 'learning_rate': 1.553357885035829e-06, 'epoch': 2.3}

 77%|███████▋  | 10226/13317 [1:45:41<31:37,  1.63it/s]
 77%|███████▋  | 10227/13317 [1:45:42<31:36,  1.63it/s]
                                                       
{'loss': 0.1435, 'grad_norm': 0.5533584952354431, 'learning_rate': 1.5524085150988077e-06, 'epoch': 2.3}

 77%|███████▋  | 10227/13317 [1:45:42<31:36,  1.63it/s]
 77%|███████▋  | 10228/13317 [1:45:42<31:35,  1.63it/s]
                                                       
{'loss': 0.1611, 'grad_norm': 0.583087146282196, 'learning_rate': 1.5514593820478274e-06, 'epoch': 2.3}

 77%|███████▋  | 10228/13317 [1:45:42<31:35,  1.63it/s]
 77%|███████▋  | 10229/13317 [1:45:43<31:43,  1.62it/s]
                                                       
{'loss': 0.1703, 'grad_norm': 0.6567412614822388, 'learning_rate': 1.5505104859480997e-06, 'epoch': 2.3}

 77%|███████▋  | 10229/13317 [1:45:43<31:43,  1.62it/s]
 77%|███████▋  | 10230/13317 [1:45:43<31:40,  1.62it/s]
                                                       
{'loss': 0.1333, 'grad_norm': 0.5071125626564026, 'learning_rate': 1.5495618268648265e-06, 'epoch': 2.3}

 77%|███████▋  | 10230/13317 [1:45:44<31:40,  1.62it/s]
 77%|███████▋  | 10231/13317 [1:45:44<31:39,  1.62it/s]
                                                       
{'loss': 0.18, 'grad_norm': 0.6548349261283875, 'learning_rate': 1.5486134048631884e-06, 'epoch': 2.3}

 77%|███████▋  | 10231/13317 [1:45:44<31:39,  1.62it/s]
 77%|███████▋  | 10232/13317 [1:45:45<31:38,  1.62it/s]
                                                       
{'loss': 0.1469, 'grad_norm': 0.5135277509689331, 'learning_rate': 1.5476652200083548e-06, 'epoch': 2.31}

 77%|███████▋  | 10232/13317 [1:45:45<31:38,  1.62it/s]
 77%|███████▋  | 10233/13317 [1:45:45<31:37,  1.63it/s]
                                                       
{'loss': 0.1747, 'grad_norm': 0.6342453360557556, 'learning_rate': 1.5467172723654745e-06, 'epoch': 2.31}

 77%|███████▋  | 10233/13317 [1:45:45<31:37,  1.63it/s]
 77%|███████▋  | 10234/13317 [1:45:46<31:39,  1.62it/s]
                                                       
{'loss': 0.1636, 'grad_norm': 0.6111482977867126, 'learning_rate': 1.5457695619996798e-06, 'epoch': 2.31}

 77%|███████▋  | 10234/13317 [1:45:46<31:39,  1.62it/s]
 77%|███████▋  | 10235/13317 [1:45:46<31:35,  1.63it/s]
                                                       
{'loss': 0.1494, 'grad_norm': 0.5550348162651062, 'learning_rate': 1.5448220889760902e-06, 'epoch': 2.31}

 77%|███████▋  | 10235/13317 [1:45:47<31:35,  1.63it/s]
 77%|███████▋  | 10236/13317 [1:45:47<31:33,  1.63it/s]
                                                       
{'loss': 0.1651, 'grad_norm': 0.5575845241546631, 'learning_rate': 1.5438748533598085e-06, 'epoch': 2.31}

 77%|███████▋  | 10236/13317 [1:45:47<31:33,  1.63it/s]
 77%|███████▋  | 10237/13317 [1:45:48<31:34,  1.63it/s]
                                                       
{'loss': 0.1334, 'grad_norm': 0.5210907459259033, 'learning_rate': 1.5429278552159189e-06, 'epoch': 2.31}

 77%|███████▋  | 10237/13317 [1:45:48<31:34,  1.63it/s]
 77%|███████▋  | 10238/13317 [1:45:48<31:32,  1.63it/s]
                                                       
{'loss': 0.1675, 'grad_norm': 0.5669235587120056, 'learning_rate': 1.5419810946094877e-06, 'epoch': 2.31}

 77%|███████▋  | 10238/13317 [1:45:48<31:32,  1.63it/s]
 77%|███████▋  | 10239/13317 [1:45:49<31:38,  1.62it/s]
                                                       
{'loss': 0.1214, 'grad_norm': 0.438597708940506, 'learning_rate': 1.5410345716055713e-06, 'epoch': 2.31}

 77%|███████▋  | 10239/13317 [1:45:49<31:38,  1.62it/s]
 77%|███████▋  | 10240/13317 [1:45:50<31:35,  1.62it/s]
                                                       
{'loss': 0.1272, 'grad_norm': 0.4762876629829407, 'learning_rate': 1.5400882862692035e-06, 'epoch': 2.31}

 77%|███████▋  | 10240/13317 [1:45:50<31:35,  1.62it/s]
 77%|███████▋  | 10241/13317 [1:45:50<31:32,  1.63it/s]
                                                       
{'loss': 0.202, 'grad_norm': 0.7001889944076538, 'learning_rate': 1.5391422386654032e-06, 'epoch': 2.31}

 77%|███████▋  | 10241/13317 [1:45:50<31:32,  1.63it/s]
 77%|███████▋  | 10242/13317 [1:45:51<31:34,  1.62it/s]
                                                       
{'loss': 0.1507, 'grad_norm': 0.5518068671226501, 'learning_rate': 1.5381964288591755e-06, 'epoch': 2.31}

 77%|███████▋  | 10242/13317 [1:45:51<31:34,  1.62it/s]
 77%|███████▋  | 10243/13317 [1:45:51<31:31,  1.63it/s]
                                                       
{'loss': 0.1926, 'grad_norm': 0.6905341744422913, 'learning_rate': 1.5372508569155104e-06, 'epoch': 2.31}

 77%|███████▋  | 10243/13317 [1:45:52<31:31,  1.63it/s]
 77%|███████▋  | 10244/13317 [1:45:52<31:38,  1.62it/s]
                                                       
{'loss': 0.1711, 'grad_norm': 0.6001460552215576, 'learning_rate': 1.5363055228993723e-06, 'epoch': 2.31}

 77%|███████▋  | 10244/13317 [1:45:52<31:38,  1.62it/s]
 77%|███████▋  | 10245/13317 [1:45:53<31:33,  1.62it/s]
                                                       
{'loss': 0.1367, 'grad_norm': 0.5035179853439331, 'learning_rate': 1.535360426875719e-06, 'epoch': 2.31}

 77%|███████▋  | 10245/13317 [1:45:53<31:33,  1.62it/s]
 77%|███████▋  | 10246/13317 [1:45:53<31:30,  1.62it/s]
                                                       
{'loss': 0.142, 'grad_norm': 0.5160623788833618, 'learning_rate': 1.5344155689094898e-06, 'epoch': 2.31}

 77%|███████▋  | 10246/13317 [1:45:53<31:30,  1.62it/s]
 77%|███████▋  | 10247/13317 [1:45:54<31:38,  1.62it/s]
                                                       
{'loss': 0.1484, 'grad_norm': 0.605839729309082, 'learning_rate': 1.5334709490656059e-06, 'epoch': 2.31}

 77%|███████▋  | 10247/13317 [1:45:54<31:38,  1.62it/s]
 77%|███████▋  | 10248/13317 [1:45:54<31:33,  1.62it/s]
                                                       
{'loss': 0.1622, 'grad_norm': 0.634394109249115, 'learning_rate': 1.5325265674089696e-06, 'epoch': 2.31}

 77%|███████▋  | 10248/13317 [1:45:55<31:33,  1.62it/s]
 77%|███████▋  | 10249/13317 [1:45:55<31:32,  1.62it/s]
                                                       
{'loss': 0.1512, 'grad_norm': 0.7008200287818909, 'learning_rate': 1.531582424004473e-06, 'epoch': 2.31}

 77%|███████▋  | 10249/13317 [1:45:55<31:32,  1.62it/s]
 77%|███████▋  | 10250/13317 [1:45:56<31:28,  1.62it/s]
                                                       
{'loss': 0.1484, 'grad_norm': 0.5933245420455933, 'learning_rate': 1.5306385189169909e-06, 'epoch': 2.31}

 77%|███████▋  | 10250/13317 [1:45:56<31:28,  1.62it/s]
 77%|███████▋  | 10251/13317 [1:45:56<31:25,  1.63it/s]
                                                       
{'loss': 0.1415, 'grad_norm': 0.486826628446579, 'learning_rate': 1.5296948522113736e-06, 'epoch': 2.31}

 77%|███████▋  | 10251/13317 [1:45:56<31:25,  1.63it/s]
 77%|███████▋  | 10252/13317 [1:45:57<31:26,  1.62it/s]
                                                       
{'loss': 0.1408, 'grad_norm': 0.5310570001602173, 'learning_rate': 1.5287514239524641e-06, 'epoch': 2.31}

 77%|███████▋  | 10252/13317 [1:45:57<31:26,  1.62it/s]
 77%|███████▋  | 10253/13317 [1:45:58<31:24,  1.63it/s]
                                                       
{'loss': 0.1607, 'grad_norm': 0.5478213429450989, 'learning_rate': 1.5278082342050877e-06, 'epoch': 2.31}

 77%|███████▋  | 10253/13317 [1:45:58<31:24,  1.63it/s]
 77%|███████▋  | 10254/13317 [1:45:58<31:24,  1.63it/s]
                                                       
{'loss': 0.1207, 'grad_norm': 0.4867706298828125, 'learning_rate': 1.5268652830340498e-06, 'epoch': 2.31}

 77%|███████▋  | 10254/13317 [1:45:58<31:24,  1.63it/s]
 77%|███████▋  | 10255/13317 [1:45:59<31:22,  1.63it/s]
                                                       
{'loss': 0.1568, 'grad_norm': 0.5867069959640503, 'learning_rate': 1.5259225705041398e-06, 'epoch': 2.31}

 77%|███████▋  | 10255/13317 [1:45:59<31:22,  1.63it/s]
 77%|███████▋  | 10256/13317 [1:45:59<31:20,  1.63it/s]
                                                       
{'loss': 0.2174, 'grad_norm': 0.6599308848381042, 'learning_rate': 1.5249800966801325e-06, 'epoch': 2.31}

 77%|███████▋  | 10256/13317 [1:46:00<31:20,  1.63it/s]
 77%|███████▋  | 10257/13317 [1:46:00<31:22,  1.63it/s]
                                                       
{'loss': 0.1788, 'grad_norm': 0.694874107837677, 'learning_rate': 1.5240378616267887e-06, 'epoch': 2.31}

 77%|███████▋  | 10257/13317 [1:46:00<31:22,  1.63it/s]
 77%|███████▋  | 10258/13317 [1:46:01<31:20,  1.63it/s]
                                                       
{'loss': 0.1683, 'grad_norm': 0.5646510124206543, 'learning_rate': 1.523095865408848e-06, 'epoch': 2.31}

 77%|███████▋  | 10258/13317 [1:46:01<31:20,  1.63it/s]
 77%|███████▋  | 10259/13317 [1:46:01<31:21,  1.63it/s]
                                                       
{'loss': 0.1602, 'grad_norm': 0.5591140985488892, 'learning_rate': 1.5221541080910336e-06, 'epoch': 2.31}

 77%|███████▋  | 10259/13317 [1:46:01<31:21,  1.63it/s]
 77%|███████▋  | 10260/13317 [1:46:02<31:19,  1.63it/s]
                                                       
{'loss': 0.0989, 'grad_norm': 0.39475712180137634, 'learning_rate': 1.5212125897380575e-06, 'epoch': 2.31}

 77%|███████▋  | 10260/13317 [1:46:02<31:19,  1.63it/s]
 77%|███████▋  | 10261/13317 [1:46:02<31:17,  1.63it/s]
                                                       
{'loss': 0.1717, 'grad_norm': 0.5392889976501465, 'learning_rate': 1.5202713104146105e-06, 'epoch': 2.31}

 77%|███████▋  | 10261/13317 [1:46:03<31:17,  1.63it/s]
 77%|███████▋  | 10262/13317 [1:46:03<31:27,  1.62it/s]
                                                       
{'loss': 0.1576, 'grad_norm': 0.6401129961013794, 'learning_rate': 1.5193302701853674e-06, 'epoch': 2.31}

 77%|███████▋  | 10262/13317 [1:46:03<31:27,  1.62it/s]
 77%|███████▋  | 10263/13317 [1:46:04<31:23,  1.62it/s]
                                                       
{'loss': 0.1357, 'grad_norm': 0.48966091871261597, 'learning_rate': 1.518389469114988e-06, 'epoch': 2.31}

 77%|███████▋  | 10263/13317 [1:46:04<31:23,  1.62it/s]
 77%|███████▋  | 10264/13317 [1:46:04<32:04,  1.59it/s]
                                                       
{'loss': 0.217, 'grad_norm': 0.8183326125144958, 'learning_rate': 1.517448907268118e-06, 'epoch': 2.31}

 77%|███████▋  | 10264/13317 [1:46:05<32:04,  1.59it/s]
 77%|███████▋  | 10265/13317 [1:46:05<31:55,  1.59it/s]
                                                       
{'loss': 0.1478, 'grad_norm': 0.534125566482544, 'learning_rate': 1.5165085847093814e-06, 'epoch': 2.31}

 77%|███████▋  | 10265/13317 [1:46:05<31:55,  1.59it/s]
 77%|███████▋  | 10266/13317 [1:46:06<31:41,  1.60it/s]
                                                       
{'loss': 0.1616, 'grad_norm': 0.6093271970748901, 'learning_rate': 1.5155685015033878e-06, 'epoch': 2.31}

 77%|███████▋  | 10266/13317 [1:46:06<31:41,  1.60it/s]
 77%|███████▋  | 10267/13317 [1:46:06<31:33,  1.61it/s]
                                                       
{'loss': 0.148, 'grad_norm': 0.5232399702072144, 'learning_rate': 1.514628657714733e-06, 'epoch': 2.31}

 77%|███████▋  | 10267/13317 [1:46:06<31:33,  1.61it/s]
 77%|███████▋  | 10268/13317 [1:46:07<31:26,  1.62it/s]
                                                       
{'loss': 0.1393, 'grad_norm': 0.5056501626968384, 'learning_rate': 1.5136890534079935e-06, 'epoch': 2.31}

 77%|███████▋  | 10268/13317 [1:46:07<31:26,  1.62it/s]
 77%|███████▋  | 10269/13317 [1:46:07<31:22,  1.62it/s]
                                                       
{'loss': 0.1883, 'grad_norm': 0.6362823247909546, 'learning_rate': 1.5127496886477278e-06, 'epoch': 2.31}

 77%|███████▋  | 10269/13317 [1:46:08<31:22,  1.62it/s]
 77%|███████▋  | 10270/13317 [1:46:08<31:18,  1.62it/s]
                                                       
{'loss': 0.1257, 'grad_norm': 0.5189551115036011, 'learning_rate': 1.5118105634984837e-06, 'epoch': 2.31}

 77%|███████▋  | 10270/13317 [1:46:08<31:18,  1.62it/s]
 77%|███████▋  | 10271/13317 [1:46:09<31:15,  1.62it/s]
                                                       
{'loss': 0.204, 'grad_norm': 0.6351556777954102, 'learning_rate': 1.5108716780247863e-06, 'epoch': 2.31}

 77%|███████▋  | 10271/13317 [1:46:09<31:15,  1.62it/s]
 77%|███████▋  | 10272/13317 [1:46:09<31:11,  1.63it/s]
                                                       
{'loss': 0.1504, 'grad_norm': 0.5261610746383667, 'learning_rate': 1.5099330322911493e-06, 'epoch': 2.31}

 77%|███████▋  | 10272/13317 [1:46:09<31:11,  1.63it/s]
 77%|███████▋  | 10273/13317 [1:46:10<31:10,  1.63it/s]
                                                       
{'loss': 0.1921, 'grad_norm': 0.7040725350379944, 'learning_rate': 1.5089946263620647e-06, 'epoch': 2.31}

 77%|███████▋  | 10273/13317 [1:46:10<31:10,  1.63it/s]
 77%|███████▋  | 10274/13317 [1:46:11<31:10,  1.63it/s]
                                                       
{'loss': 0.153, 'grad_norm': 0.5143675804138184, 'learning_rate': 1.5080564603020143e-06, 'epoch': 2.31}

 77%|███████▋  | 10274/13317 [1:46:11<31:10,  1.63it/s]
 77%|███████▋  | 10275/13317 [1:46:11<31:08,  1.63it/s]
                                                       
{'loss': 0.1847, 'grad_norm': 0.6292542815208435, 'learning_rate': 1.5071185341754585e-06, 'epoch': 2.31}

 77%|███████▋  | 10275/13317 [1:46:11<31:08,  1.63it/s]
 77%|███████▋  | 10276/13317 [1:46:12<31:07,  1.63it/s]
                                                       
{'loss': 0.13, 'grad_norm': 0.4891291856765747, 'learning_rate': 1.506180848046841e-06, 'epoch': 2.31}

 77%|███████▋  | 10276/13317 [1:46:12<31:07,  1.63it/s]
 77%|███████▋  | 10277/13317 [1:46:12<31:10,  1.63it/s]
                                                       
{'loss': 0.1586, 'grad_norm': 0.5593088865280151, 'learning_rate': 1.505243401980594e-06, 'epoch': 2.32}

 77%|███████▋  | 10277/13317 [1:46:12<31:10,  1.63it/s]
 77%|███████▋  | 10278/13317 [1:46:13<31:07,  1.63it/s]
                                                       
{'loss': 0.1583, 'grad_norm': 0.5680751800537109, 'learning_rate': 1.5043061960411266e-06, 'epoch': 2.32}

 77%|███████▋  | 10278/13317 [1:46:13<31:07,  1.63it/s]
 77%|███████▋  | 10279/13317 [1:46:14<31:08,  1.63it/s]
                                                       
{'loss': 0.1521, 'grad_norm': 0.6052099466323853, 'learning_rate': 1.5033692302928382e-06, 'epoch': 2.32}

 77%|███████▋  | 10279/13317 [1:46:14<31:08,  1.63it/s]
 77%|███████▋  | 10280/13317 [1:46:14<31:06,  1.63it/s]
                                                       
{'loss': 0.1091, 'grad_norm': 0.4896833002567291, 'learning_rate': 1.5024325048001054e-06, 'epoch': 2.32}

 77%|███████▋  | 10280/13317 [1:46:14<31:06,  1.63it/s]
 77%|███████▋  | 10281/13317 [1:46:15<31:04,  1.63it/s]
                                                       
{'loss': 0.154, 'grad_norm': 0.5842311978340149, 'learning_rate': 1.5014960196272931e-06, 'epoch': 2.32}

 77%|███████▋  | 10281/13317 [1:46:15<31:04,  1.63it/s]
 77%|███████▋  | 10282/13317 [1:46:15<31:05,  1.63it/s]
                                                       
{'loss': 0.1387, 'grad_norm': 0.5583600997924805, 'learning_rate': 1.500559774838748e-06, 'epoch': 2.32}

 77%|███████▋  | 10282/13317 [1:46:16<31:05,  1.63it/s]
 77%|███████▋  | 10283/13317 [1:46:16<31:03,  1.63it/s]
                                                       
{'loss': 0.1335, 'grad_norm': 0.5464246869087219, 'learning_rate': 1.4996237704987966e-06, 'epoch': 2.32}

 77%|███████▋  | 10283/13317 [1:46:16<31:03,  1.63it/s]
 77%|███████▋  | 10284/13317 [1:46:17<31:03,  1.63it/s]
                                                       
{'loss': 0.1399, 'grad_norm': 0.5773972868919373, 'learning_rate': 1.498688006671757e-06, 'epoch': 2.32}

 77%|███████▋  | 10284/13317 [1:46:17<31:03,  1.63it/s]
 77%|███████▋  | 10285/13317 [1:46:17<31:02,  1.63it/s]
                                                       
{'loss': 0.1372, 'grad_norm': 0.5019108653068542, 'learning_rate': 1.4977524834219215e-06, 'epoch': 2.32}

 77%|███████▋  | 10285/13317 [1:46:17<31:02,  1.63it/s]
 77%|███████▋  | 10286/13317 [1:46:18<31:01,  1.63it/s]
                                                       
{'loss': 0.1446, 'grad_norm': 0.6470386385917664, 'learning_rate': 1.496817200813575e-06, 'epoch': 2.32}

 77%|███████▋  | 10286/13317 [1:46:18<31:01,  1.63it/s]
 77%|███████▋  | 10287/13317 [1:46:18<30:59,  1.63it/s]
                                                       
{'loss': 0.1421, 'grad_norm': 0.6320403218269348, 'learning_rate': 1.495882158910979e-06, 'epoch': 2.32}

 77%|███████▋  | 10287/13317 [1:46:19<30:59,  1.63it/s]
 77%|███████▋  | 10288/13317 [1:46:19<30:58,  1.63it/s]
                                                       
{'loss': 0.1646, 'grad_norm': 0.6181575059890747, 'learning_rate': 1.494947357778379e-06, 'epoch': 2.32}

 77%|███████▋  | 10288/13317 [1:46:19<30:58,  1.63it/s]
 77%|███████▋  | 10289/13317 [1:46:20<31:00,  1.63it/s]
                                                       
{'loss': 0.1125, 'grad_norm': 0.47969162464141846, 'learning_rate': 1.4940127974800078e-06, 'epoch': 2.32}

 77%|███████▋  | 10289/13317 [1:46:20<31:00,  1.63it/s]
 77%|███████▋  | 10290/13317 [1:46:20<30:57,  1.63it/s]
                                                       
{'loss': 0.1535, 'grad_norm': 0.5863584280014038, 'learning_rate': 1.493078478080081e-06, 'epoch': 2.32}

 77%|███████▋  | 10290/13317 [1:46:20<30:57,  1.63it/s]
 77%|███████▋  | 10291/13317 [1:46:21<30:56,  1.63it/s]
                                                       
{'loss': 0.1774, 'grad_norm': 0.5975820422172546, 'learning_rate': 1.4921443996427948e-06, 'epoch': 2.32}

 77%|███████▋  | 10291/13317 [1:46:21<30:56,  1.63it/s]
 77%|███████▋  | 10292/13317 [1:46:22<30:55,  1.63it/s]
                                                       
{'loss': 0.1539, 'grad_norm': 0.5625325441360474, 'learning_rate': 1.491210562232328e-06, 'epoch': 2.32}

 77%|███████▋  | 10292/13317 [1:46:22<30:55,  1.63it/s]
 77%|███████▋  | 10293/13317 [1:46:22<30:55,  1.63it/s]
                                                       
{'loss': 0.1208, 'grad_norm': 0.5074368715286255, 'learning_rate': 1.490276965912849e-06, 'epoch': 2.32}

 77%|███████▋  | 10293/13317 [1:46:22<30:55,  1.63it/s]
 77%|███████▋  | 10294/13317 [1:46:23<30:57,  1.63it/s]
                                                       
{'loss': 0.1433, 'grad_norm': 0.5657719969749451, 'learning_rate': 1.4893436107485043e-06, 'epoch': 2.32}

 77%|███████▋  | 10294/13317 [1:46:23<30:57,  1.63it/s]
 77%|███████▋  | 10295/13317 [1:46:23<30:56,  1.63it/s]
                                                       
{'loss': 0.1314, 'grad_norm': 0.5671060085296631, 'learning_rate': 1.4884104968034234e-06, 'epoch': 2.32}

 77%|███████▋  | 10295/13317 [1:46:24<30:56,  1.63it/s]
 77%|███████▋  | 10296/13317 [1:46:24<30:54,  1.63it/s]
                                                       
{'loss': 0.1257, 'grad_norm': 0.5071542859077454, 'learning_rate': 1.4874776241417227e-06, 'epoch': 2.32}

 77%|███████▋  | 10296/13317 [1:46:24<30:54,  1.63it/s]
 77%|███████▋  | 10297/13317 [1:46:25<30:53,  1.63it/s]
                                                       
{'loss': 0.1342, 'grad_norm': 0.5465911626815796, 'learning_rate': 1.4865449928275017e-06, 'epoch': 2.32}

 77%|███████▋  | 10297/13317 [1:46:25<30:53,  1.63it/s]
 77%|███████▋  | 10298/13317 [1:46:25<30:53,  1.63it/s]
                                                       
{'loss': 0.1491, 'grad_norm': 0.5796815156936646, 'learning_rate': 1.4856126029248407e-06, 'epoch': 2.32}

 77%|███████▋  | 10298/13317 [1:46:25<30:53,  1.63it/s]
 77%|███████▋  | 10299/13317 [1:46:26<30:55,  1.63it/s]
                                                       
{'loss': 0.1832, 'grad_norm': 0.6045339703559875, 'learning_rate': 1.4846804544978034e-06, 'epoch': 2.32}

 77%|███████▋  | 10299/13317 [1:46:26<30:55,  1.63it/s]
 77%|███████▋  | 10300/13317 [1:46:26<30:55,  1.63it/s]
                                                       
{'loss': 0.1403, 'grad_norm': 0.6082249879837036, 'learning_rate': 1.4837485476104413e-06, 'epoch': 2.32}

 77%|███████▋  | 10300/13317 [1:46:27<30:55,  1.63it/s]
 77%|███████▋  | 10301/13317 [1:46:27<30:53,  1.63it/s]
                                                       
{'loss': 0.1572, 'grad_norm': 0.5254184603691101, 'learning_rate': 1.4828168823267847e-06, 'epoch': 2.32}

 77%|███████▋  | 10301/13317 [1:46:27<30:53,  1.63it/s]
 77%|███████▋  | 10302/13317 [1:46:28<30:53,  1.63it/s]
                                                       
{'loss': 0.1438, 'grad_norm': 0.5441808104515076, 'learning_rate': 1.4818854587108471e-06, 'epoch': 2.32}

 77%|███████▋  | 10302/13317 [1:46:28<30:53,  1.63it/s]
 77%|███████▋  | 10303/13317 [1:46:28<30:53,  1.63it/s]
                                                       
{'loss': 0.155, 'grad_norm': 0.5625491738319397, 'learning_rate': 1.4809542768266288e-06, 'epoch': 2.32}

 77%|███████▋  | 10303/13317 [1:46:28<30:53,  1.63it/s]
 77%|███████▋  | 10304/13317 [1:46:29<31:00,  1.62it/s]
                                                       
{'loss': 0.1236, 'grad_norm': 0.47428613901138306, 'learning_rate': 1.480023336738115e-06, 'epoch': 2.32}

 77%|███████▋  | 10304/13317 [1:46:29<31:00,  1.62it/s]
 77%|███████▋  | 10305/13317 [1:46:30<30:56,  1.62it/s]
                                                       
{'loss': 0.1593, 'grad_norm': 0.622397780418396, 'learning_rate': 1.4790926385092651e-06, 'epoch': 2.32}

 77%|███████▋  | 10305/13317 [1:46:30<30:56,  1.62it/s]
 77%|███████▋  | 10306/13317 [1:46:30<30:54,  1.62it/s]
                                                       
{'loss': 0.1351, 'grad_norm': 0.5499795079231262, 'learning_rate': 1.4781621822040309e-06, 'epoch': 2.32}

 77%|███████▋  | 10306/13317 [1:46:30<30:54,  1.62it/s]
 77%|███████▋  | 10307/13317 [1:46:31<30:54,  1.62it/s]
                                                       
{'loss': 0.1452, 'grad_norm': 0.5215097069740295, 'learning_rate': 1.4772319678863461e-06, 'epoch': 2.32}

 77%|███████▋  | 10307/13317 [1:46:31<30:54,  1.62it/s]
 77%|███████▋  | 10308/13317 [1:46:31<30:52,  1.62it/s]
                                                       
{'loss': 0.1223, 'grad_norm': 0.5340060591697693, 'learning_rate': 1.4763019956201252e-06, 'epoch': 2.32}

 77%|███████▋  | 10308/13317 [1:46:32<30:52,  1.62it/s]
 77%|███████▋  | 10309/13317 [1:46:32<30:51,  1.62it/s]
                                                       
{'loss': 0.1405, 'grad_norm': 0.5536866188049316, 'learning_rate': 1.475372265469265e-06, 'epoch': 2.32}

 77%|███████▋  | 10309/13317 [1:46:32<30:51,  1.62it/s]
 77%|███████▋  | 10310/13317 [1:46:33<30:48,  1.63it/s]
                                                       
{'loss': 0.2084, 'grad_norm': 0.7096543312072754, 'learning_rate': 1.4744427774976489e-06, 'epoch': 2.32}

 77%|███████▋  | 10310/13317 [1:46:33<30:48,  1.63it/s]
 77%|███████▋  | 10311/13317 [1:46:33<30:47,  1.63it/s]
                                                       
{'loss': 0.1565, 'grad_norm': 0.6302303671836853, 'learning_rate': 1.473513531769145e-06, 'epoch': 2.32}

 77%|███████▋  | 10311/13317 [1:46:33<30:47,  1.63it/s]
 77%|███████▋  | 10312/13317 [1:46:34<30:48,  1.63it/s]
                                                       
{'loss': 0.1622, 'grad_norm': 0.5841745138168335, 'learning_rate': 1.4725845283476008e-06, 'epoch': 2.32}

 77%|███████▋  | 10312/13317 [1:46:34<30:48,  1.63it/s]
 77%|███████▋  | 10313/13317 [1:46:34<30:45,  1.63it/s]
                                                       
{'loss': 0.1861, 'grad_norm': 0.5152497887611389, 'learning_rate': 1.4716557672968467e-06, 'epoch': 2.32}

 77%|███████▋  | 10313/13317 [1:46:35<30:45,  1.63it/s]
 77%|███████▋  | 10314/13317 [1:46:35<30:45,  1.63it/s]
                                                       
{'loss': 0.1691, 'grad_norm': 0.48360344767570496, 'learning_rate': 1.4707272486807016e-06, 'epoch': 2.32}

 77%|███████▋  | 10314/13317 [1:46:35<30:45,  1.63it/s]
 77%|███████▋  | 10315/13317 [1:46:36<30:43,  1.63it/s]
                                                       
{'loss': 0.1505, 'grad_norm': 0.5765366554260254, 'learning_rate': 1.469798972562963e-06, 'epoch': 2.32}

 77%|███████▋  | 10315/13317 [1:46:36<30:43,  1.63it/s]
 77%|███████▋  | 10316/13317 [1:46:36<30:43,  1.63it/s]
                                                       
{'loss': 0.1375, 'grad_norm': 0.5233107209205627, 'learning_rate': 1.468870939007412e-06, 'epoch': 2.32}

 77%|███████▋  | 10316/13317 [1:46:36<30:43,  1.63it/s]
 77%|███████▋  | 10317/13317 [1:46:37<30:43,  1.63it/s]
                                                       
{'loss': 0.1435, 'grad_norm': 0.6152898669242859, 'learning_rate': 1.4679431480778156e-06, 'epoch': 2.32}

 77%|███████▋  | 10317/13317 [1:46:37<30:43,  1.63it/s]
 77%|███████▋  | 10318/13317 [1:46:38<30:42,  1.63it/s]
                                                       
{'loss': 0.156, 'grad_norm': 0.5637812614440918, 'learning_rate': 1.4670155998379242e-06, 'epoch': 2.32}

 77%|███████▋  | 10318/13317 [1:46:38<30:42,  1.63it/s]
 77%|███████▋  | 10319/13317 [1:46:38<30:43,  1.63it/s]
                                                       
{'loss': 0.1399, 'grad_norm': 0.44450774788856506, 'learning_rate': 1.4660882943514693e-06, 'epoch': 2.32}

 77%|███████▋  | 10319/13317 [1:46:38<30:43,  1.63it/s]
 77%|███████▋  | 10320/13317 [1:46:39<30:41,  1.63it/s]
                                                       
{'loss': 0.1383, 'grad_norm': 0.5597094893455505, 'learning_rate': 1.4651612316821646e-06, 'epoch': 2.32}

 77%|███████▋  | 10320/13317 [1:46:39<30:41,  1.63it/s]
 78%|███████▊  | 10321/13317 [1:46:39<30:38,  1.63it/s]
                                                       
{'loss': 0.144, 'grad_norm': 0.5445024371147156, 'learning_rate': 1.4642344118937124e-06, 'epoch': 2.33}

 78%|███████▊  | 10321/13317 [1:46:40<30:38,  1.63it/s]
 78%|███████▊  | 10322/13317 [1:46:40<30:41,  1.63it/s]
                                                       
{'loss': 0.1175, 'grad_norm': 0.4626706838607788, 'learning_rate': 1.4633078350497932e-06, 'epoch': 2.33}

 78%|███████▊  | 10322/13317 [1:46:40<30:41,  1.63it/s]
 78%|███████▊  | 10323/13317 [1:46:41<30:40,  1.63it/s]
                                                       
{'loss': 0.1628, 'grad_norm': 0.5210431218147278, 'learning_rate': 1.4623815012140714e-06, 'epoch': 2.33}

 78%|███████▊  | 10323/13317 [1:46:41<30:40,  1.63it/s]
 78%|███████▊  | 10324/13317 [1:46:41<30:41,  1.63it/s]
                                                       
{'loss': 0.1546, 'grad_norm': 0.5336509346961975, 'learning_rate': 1.461455410450197e-06, 'epoch': 2.33}

 78%|███████▊  | 10324/13317 [1:46:41<30:41,  1.63it/s]
 78%|███████▊  | 10325/13317 [1:46:42<30:38,  1.63it/s]
                                                       
{'loss': 0.159, 'grad_norm': 0.5669369101524353, 'learning_rate': 1.4605295628218046e-06, 'epoch': 2.33}

 78%|███████▊  | 10325/13317 [1:46:42<30:38,  1.63it/s]
 78%|███████▊  | 10326/13317 [1:46:42<30:36,  1.63it/s]
                                                       
{'loss': 0.2062, 'grad_norm': 0.6868971586227417, 'learning_rate': 1.4596039583925082e-06, 'epoch': 2.33}

 78%|███████▊  | 10326/13317 [1:46:43<30:36,  1.63it/s]
 78%|███████▊  | 10327/13317 [1:46:43<30:39,  1.63it/s]
                                                       
{'loss': 0.1263, 'grad_norm': 0.4781598746776581, 'learning_rate': 1.4586785972259038e-06, 'epoch': 2.33}

 78%|███████▊  | 10327/13317 [1:46:43<30:39,  1.63it/s]
 78%|███████▊  | 10328/13317 [1:46:44<30:36,  1.63it/s]
                                                       
{'loss': 0.1977, 'grad_norm': 0.5641443133354187, 'learning_rate': 1.4577534793855774e-06, 'epoch': 2.33}

 78%|███████▊  | 10328/13317 [1:46:44<30:36,  1.63it/s]
 78%|███████▊  | 10329/13317 [1:46:44<30:34,  1.63it/s]
                                                       
{'loss': 0.1618, 'grad_norm': 0.6025774478912354, 'learning_rate': 1.4568286049350932e-06, 'epoch': 2.33}

 78%|███████▊  | 10329/13317 [1:46:44<30:34,  1.63it/s]
 78%|███████▊  | 10330/13317 [1:46:45<30:33,  1.63it/s]
                                                       
{'loss': 0.1544, 'grad_norm': 0.5616703629493713, 'learning_rate': 1.4559039739379982e-06, 'epoch': 2.33}

 78%|███████▊  | 10330/13317 [1:46:45<30:33,  1.63it/s]
 78%|███████▊  | 10331/13317 [1:46:46<30:31,  1.63it/s]
                                                       
{'loss': 0.1199, 'grad_norm': 0.5213028788566589, 'learning_rate': 1.454979586457827e-06, 'epoch': 2.33}

 78%|███████▊  | 10331/13317 [1:46:46<30:31,  1.63it/s]
 78%|███████▊  | 10332/13317 [1:46:46<30:31,  1.63it/s]
                                                       
{'loss': 0.1408, 'grad_norm': 0.5540585517883301, 'learning_rate': 1.4540554425580917e-06, 'epoch': 2.33}

 78%|███████▊  | 10332/13317 [1:46:46<30:31,  1.63it/s]
 78%|███████▊  | 10333/13317 [1:46:47<30:31,  1.63it/s]
                                                       
{'loss': 0.1682, 'grad_norm': 0.5846338868141174, 'learning_rate': 1.4531315423022946e-06, 'epoch': 2.33}

 78%|███████▊  | 10333/13317 [1:46:47<30:31,  1.63it/s]
 78%|███████▊  | 10334/13317 [1:46:47<30:30,  1.63it/s]
                                                       
{'loss': 0.1576, 'grad_norm': 0.605430543422699, 'learning_rate': 1.4522078857539135e-06, 'epoch': 2.33}

 78%|███████▊  | 10334/13317 [1:46:48<30:30,  1.63it/s]
 78%|███████▊  | 10335/13317 [1:46:48<30:28,  1.63it/s]
                                                       
{'loss': 0.1516, 'grad_norm': 0.5292304754257202, 'learning_rate': 1.451284472976417e-06, 'epoch': 2.33}

 78%|███████▊  | 10335/13317 [1:46:48<30:28,  1.63it/s]
 78%|███████▊  | 10336/13317 [1:46:49<30:28,  1.63it/s]
                                                       
{'loss': 0.1571, 'grad_norm': 0.6661537289619446, 'learning_rate': 1.4503613040332498e-06, 'epoch': 2.33}

 78%|███████▊  | 10336/13317 [1:46:49<30:28,  1.63it/s]
 78%|███████▊  | 10337/13317 [1:46:49<30:27,  1.63it/s]
                                                       
{'loss': 0.1567, 'grad_norm': 0.6880689859390259, 'learning_rate': 1.4494383789878468e-06, 'epoch': 2.33}

 78%|███████▊  | 10337/13317 [1:46:49<30:27,  1.63it/s]
 78%|███████▊  | 10338/13317 [1:46:50<30:27,  1.63it/s]
                                                       
{'loss': 0.1544, 'grad_norm': 0.5710260272026062, 'learning_rate': 1.4485156979036214e-06, 'epoch': 2.33}

 78%|███████▊  | 10338/13317 [1:46:50<30:27,  1.63it/s]
 78%|███████▊  | 10339/13317 [1:46:50<30:28,  1.63it/s]
                                                       
{'loss': 0.1555, 'grad_norm': 0.6195201873779297, 'learning_rate': 1.4475932608439697e-06, 'epoch': 2.33}

 78%|███████▊  | 10339/13317 [1:46:51<30:28,  1.63it/s]
 78%|███████▊  | 10340/13317 [1:46:51<30:26,  1.63it/s]
                                                       
{'loss': 0.1717, 'grad_norm': 0.6289955377578735, 'learning_rate': 1.4466710678722756e-06, 'epoch': 2.33}

 78%|███████▊  | 10340/13317 [1:46:51<30:26,  1.63it/s]
 78%|███████▊  | 10341/13317 [1:46:52<30:25,  1.63it/s]
                                                       
{'loss': 0.1435, 'grad_norm': 0.5326129198074341, 'learning_rate': 1.4457491190519013e-06, 'epoch': 2.33}

 78%|███████▊  | 10341/13317 [1:46:52<30:25,  1.63it/s]
 78%|███████▊  | 10342/13317 [1:46:52<30:24,  1.63it/s]
                                                       
{'loss': 0.1116, 'grad_norm': 0.4709962010383606, 'learning_rate': 1.4448274144461965e-06, 'epoch': 2.33}

 78%|███████▊  | 10342/13317 [1:46:52<30:24,  1.63it/s]
 78%|███████▊  | 10343/13317 [1:46:53<30:24,  1.63it/s]
                                                       
{'loss': 0.1411, 'grad_norm': 0.4788496494293213, 'learning_rate': 1.4439059541184902e-06, 'epoch': 2.33}

 78%|███████▊  | 10343/13317 [1:46:53<30:24,  1.63it/s]
 78%|███████▊  | 10344/13317 [1:46:54<30:58,  1.60it/s]
                                                       
{'loss': 0.1674, 'grad_norm': 0.6053767204284668, 'learning_rate': 1.4429847381320983e-06, 'epoch': 2.33}

 78%|███████▊  | 10344/13317 [1:46:54<30:58,  1.60it/s]
 78%|███████▊  | 10345/13317 [1:46:54<31:40,  1.56it/s]
                                                       
{'loss': 0.1231, 'grad_norm': 0.4826907515525818, 'learning_rate': 1.442063766550318e-06, 'epoch': 2.33}

 78%|███████▊  | 10345/13317 [1:46:54<31:40,  1.56it/s]
 78%|███████▊  | 10346/13317 [1:46:55<32:16,  1.53it/s]
                                                       
{'loss': 0.1197, 'grad_norm': 0.47324398159980774, 'learning_rate': 1.4411430394364267e-06, 'epoch': 2.33}

 78%|███████▊  | 10346/13317 [1:46:55<32:16,  1.53it/s]
 78%|███████▊  | 10347/13317 [1:46:56<32:36,  1.52it/s]
                                                       
{'loss': 0.1698, 'grad_norm': 0.6704502701759338, 'learning_rate': 1.440222556853692e-06, 'epoch': 2.33}

 78%|███████▊  | 10347/13317 [1:46:56<32:36,  1.52it/s]
 78%|███████▊  | 10348/13317 [1:46:56<31:56,  1.55it/s]
                                                       
{'loss': 0.1708, 'grad_norm': 0.5986996293067932, 'learning_rate': 1.4393023188653592e-06, 'epoch': 2.33}

 78%|███████▊  | 10348/13317 [1:46:56<31:56,  1.55it/s]
 78%|███████▊  | 10349/13317 [1:46:57<31:28,  1.57it/s]
                                                       
{'loss': 0.1786, 'grad_norm': 0.6932176947593689, 'learning_rate': 1.438382325534657e-06, 'epoch': 2.33}

 78%|███████▊  | 10349/13317 [1:46:57<31:28,  1.57it/s]
 78%|███████▊  | 10350/13317 [1:46:57<31:06,  1.59it/s]
                                                       
{'loss': 0.1629, 'grad_norm': 0.5685776472091675, 'learning_rate': 1.4374625769248e-06, 'epoch': 2.33}

 78%|███████▊  | 10350/13317 [1:46:58<31:06,  1.59it/s]
 78%|███████▊  | 10351/13317 [1:46:58<30:51,  1.60it/s]
                                                       
{'loss': 0.1709, 'grad_norm': 0.7042291164398193, 'learning_rate': 1.4365430730989866e-06, 'epoch': 2.33}

 78%|███████▊  | 10351/13317 [1:46:58<30:51,  1.60it/s]
 78%|███████▊  | 10352/13317 [1:46:59<30:43,  1.61it/s]
                                                       
{'loss': 0.1481, 'grad_norm': 0.5216469764709473, 'learning_rate': 1.4356238141203937e-06, 'epoch': 2.33}

 78%|███████▊  | 10352/13317 [1:46:59<30:43,  1.61it/s]
 78%|███████▊  | 10353/13317 [1:46:59<30:36,  1.61it/s]
                                                       
{'loss': 0.138, 'grad_norm': 0.49357911944389343, 'learning_rate': 1.434704800052184e-06, 'epoch': 2.33}

 78%|███████▊  | 10353/13317 [1:46:59<30:36,  1.61it/s]
 78%|███████▊  | 10354/13317 [1:47:00<30:30,  1.62it/s]
                                                       
{'loss': 0.1825, 'grad_norm': 0.6752015948295593, 'learning_rate': 1.4337860309575065e-06, 'epoch': 2.33}

 78%|███████▊  | 10354/13317 [1:47:00<30:30,  1.62it/s]
 78%|███████▊  | 10355/13317 [1:47:00<30:25,  1.62it/s]
                                                       
{'loss': 0.1369, 'grad_norm': 0.4891977310180664, 'learning_rate': 1.4328675068994873e-06, 'epoch': 2.33}

 78%|███████▊  | 10355/13317 [1:47:01<30:25,  1.62it/s]
 78%|███████▊  | 10356/13317 [1:47:01<30:21,  1.63it/s]
                                                       
{'loss': 0.1405, 'grad_norm': 0.5348027944564819, 'learning_rate': 1.4319492279412388e-06, 'epoch': 2.33}

 78%|███████▊  | 10356/13317 [1:47:01<30:21,  1.63it/s]
 78%|███████▊  | 10357/13317 [1:47:02<30:31,  1.62it/s]
                                                       
{'loss': 0.1392, 'grad_norm': 0.5196467638015747, 'learning_rate': 1.4310311941458576e-06, 'epoch': 2.33}

 78%|███████▊  | 10357/13317 [1:47:02<30:31,  1.62it/s]
 78%|███████▊  | 10358/13317 [1:47:02<30:25,  1.62it/s]
                                                       
{'loss': 0.1266, 'grad_norm': 0.4202362298965454, 'learning_rate': 1.4301134055764238e-06, 'epoch': 2.33}

 78%|███████▊  | 10358/13317 [1:47:02<30:25,  1.62it/s]
 78%|███████▊  | 10359/13317 [1:47:03<30:22,  1.62it/s]
                                                       
{'loss': 0.1352, 'grad_norm': 0.5353181958198547, 'learning_rate': 1.4291958622959972e-06, 'epoch': 2.33}

 78%|███████▊  | 10359/13317 [1:47:03<30:22,  1.62it/s]
 78%|███████▊  | 10360/13317 [1:47:04<30:19,  1.63it/s]
                                                       
{'loss': 0.1317, 'grad_norm': 0.6105872392654419, 'learning_rate': 1.4282785643676217e-06, 'epoch': 2.33}

 78%|███████▊  | 10360/13317 [1:47:04<30:19,  1.63it/s]
 78%|███████▊  | 10361/13317 [1:47:04<30:17,  1.63it/s]
                                                       
{'loss': 0.1431, 'grad_norm': 0.5427822470664978, 'learning_rate': 1.4273615118543288e-06, 'epoch': 2.33}

 78%|███████▊  | 10361/13317 [1:47:04<30:17,  1.63it/s]
 78%|███████▊  | 10362/13317 [1:47:05<30:18,  1.62it/s]
                                                       
{'loss': 0.1544, 'grad_norm': 0.587266206741333, 'learning_rate': 1.4264447048191272e-06, 'epoch': 2.33}

 78%|███████▊  | 10362/13317 [1:47:05<30:18,  1.62it/s]
 78%|███████▊  | 10363/13317 [1:47:05<30:18,  1.62it/s]
                                                       
{'loss': 0.1804, 'grad_norm': 0.6955783367156982, 'learning_rate': 1.4255281433250106e-06, 'epoch': 2.33}

 78%|███████▊  | 10363/13317 [1:47:06<30:18,  1.62it/s]
 78%|███████▊  | 10364/13317 [1:47:06<30:15,  1.63it/s]
                                                       
{'loss': 0.174, 'grad_norm': 0.7686485052108765, 'learning_rate': 1.4246118274349575e-06, 'epoch': 2.33}

 78%|███████▊  | 10364/13317 [1:47:06<30:15,  1.63it/s]
 78%|███████▊  | 10365/13317 [1:47:07<30:13,  1.63it/s]
                                                       
{'loss': 0.1488, 'grad_norm': 0.5705803632736206, 'learning_rate': 1.4236957572119304e-06, 'epoch': 2.33}

 78%|███████▊  | 10365/13317 [1:47:07<30:13,  1.63it/s]
 78%|███████▊  | 10366/13317 [1:47:07<30:12,  1.63it/s]
                                                       
{'loss': 0.1583, 'grad_norm': 0.5526302456855774, 'learning_rate': 1.4227799327188707e-06, 'epoch': 2.34}

 78%|███████▊  | 10366/13317 [1:47:07<30:12,  1.63it/s]
 78%|███████▊  | 10367/13317 [1:47:08<30:14,  1.63it/s]
                                                       
{'loss': 0.1428, 'grad_norm': 0.5412474870681763, 'learning_rate': 1.421864354018705e-06, 'epoch': 2.34}

 78%|███████▊  | 10367/13317 [1:47:08<30:14,  1.63it/s]
 78%|███████▊  | 10368/13317 [1:47:08<30:12,  1.63it/s]
                                                       
{'loss': 0.1319, 'grad_norm': 0.5104174613952637, 'learning_rate': 1.4209490211743438e-06, 'epoch': 2.34}

 78%|███████▊  | 10368/13317 [1:47:09<30:12,  1.63it/s]
 78%|███████▊  | 10369/13317 [1:47:09<30:11,  1.63it/s]
                                                       
{'loss': 0.1246, 'grad_norm': 0.503560483455658, 'learning_rate': 1.4200339342486836e-06, 'epoch': 2.34}

 78%|███████▊  | 10369/13317 [1:47:09<30:11,  1.63it/s]
 78%|███████▊  | 10370/13317 [1:47:10<30:10,  1.63it/s]
                                                       
{'loss': 0.1737, 'grad_norm': 0.608051598072052, 'learning_rate': 1.419119093304594e-06, 'epoch': 2.34}

 78%|███████▊  | 10370/13317 [1:47:10<30:10,  1.63it/s]
 78%|███████▊  | 10371/13317 [1:47:10<30:08,  1.63it/s]
                                                       
{'loss': 0.1717, 'grad_norm': 0.6148775815963745, 'learning_rate': 1.4182044984049381e-06, 'epoch': 2.34}

 78%|███████▊  | 10371/13317 [1:47:10<30:08,  1.63it/s]
 78%|███████▊  | 10372/13317 [1:47:11<30:10,  1.63it/s]
                                                       
{'loss': 0.152, 'grad_norm': 0.5645716190338135, 'learning_rate': 1.417290149612559e-06, 'epoch': 2.34}

 78%|███████▊  | 10372/13317 [1:47:11<30:10,  1.63it/s]
 78%|███████▊  | 10373/13317 [1:47:12<30:09,  1.63it/s]
                                                       
{'loss': 0.1489, 'grad_norm': 0.5665345788002014, 'learning_rate': 1.4163760469902815e-06, 'epoch': 2.34}

 78%|███████▊  | 10373/13317 [1:47:12<30:09,  1.63it/s]
 78%|███████▊  | 10374/13317 [1:47:12<30:06,  1.63it/s]
                                                       
{'loss': 0.141, 'grad_norm': 0.5541147589683533, 'learning_rate': 1.4154621906009113e-06, 'epoch': 2.34}

 78%|███████▊  | 10374/13317 [1:47:12<30:06,  1.63it/s]
 78%|███████▊  | 10375/13317 [1:47:13<30:05,  1.63it/s]
                                                       
{'loss': 0.1425, 'grad_norm': 0.5444077253341675, 'learning_rate': 1.4145485805072446e-06, 'epoch': 2.34}

 78%|███████▊  | 10375/13317 [1:47:13<30:05,  1.63it/s]
 78%|███████▊  | 10376/13317 [1:47:13<30:04,  1.63it/s]
                                                       
{'loss': 0.1502, 'grad_norm': 0.56072998046875, 'learning_rate': 1.413635216772053e-06, 'epoch': 2.34}

 78%|███████▊  | 10376/13317 [1:47:14<30:04,  1.63it/s]
 78%|███████▊  | 10377/13317 [1:47:14<30:05,  1.63it/s]
                                                       
{'loss': 0.122, 'grad_norm': 0.4793855547904968, 'learning_rate': 1.4127220994580943e-06, 'epoch': 2.34}

 78%|███████▊  | 10377/13317 [1:47:14<30:05,  1.63it/s]
 78%|███████▊  | 10378/13317 [1:47:15<30:03,  1.63it/s]
                                                       
{'loss': 0.1547, 'grad_norm': 0.5885844230651855, 'learning_rate': 1.4118092286281092e-06, 'epoch': 2.34}

 78%|███████▊  | 10378/13317 [1:47:15<30:03,  1.63it/s]
 78%|███████▊  | 10379/13317 [1:47:15<30:04,  1.63it/s]
                                                       
{'loss': 0.1367, 'grad_norm': 0.5128939747810364, 'learning_rate': 1.4108966043448247e-06, 'epoch': 2.34}

 78%|███████▊  | 10379/13317 [1:47:15<30:04,  1.63it/s]
 78%|███████▊  | 10380/13317 [1:47:16<30:02,  1.63it/s]
                                                       
{'loss': 0.1443, 'grad_norm': 0.5386395454406738, 'learning_rate': 1.4099842266709456e-06, 'epoch': 2.34}

 78%|███████▊  | 10380/13317 [1:47:16<30:02,  1.63it/s]
 78%|███████▊  | 10381/13317 [1:47:16<30:00,  1.63it/s]
                                                       
{'loss': 0.1551, 'grad_norm': 0.5429352521896362, 'learning_rate': 1.4090720956691601e-06, 'epoch': 2.34}

 78%|███████▊  | 10381/13317 [1:47:17<30:00,  1.63it/s]
 78%|███████▊  | 10382/13317 [1:47:17<30:02,  1.63it/s]
                                                       
{'loss': 0.1551, 'grad_norm': 0.6065042018890381, 'learning_rate': 1.4081602114021442e-06, 'epoch': 2.34}

 78%|███████▊  | 10382/13317 [1:47:17<30:02,  1.63it/s]
 78%|███████▊  | 10383/13317 [1:47:18<30:01,  1.63it/s]
                                                       
{'loss': 0.1355, 'grad_norm': 0.5277395844459534, 'learning_rate': 1.4072485739325532e-06, 'epoch': 2.34}

 78%|███████▊  | 10383/13317 [1:47:18<30:01,  1.63it/s]
 78%|███████▊  | 10384/13317 [1:47:18<30:00,  1.63it/s]
                                                       
{'loss': 0.1684, 'grad_norm': 0.5918450951576233, 'learning_rate': 1.4063371833230237e-06, 'epoch': 2.34}

 78%|███████▊  | 10384/13317 [1:47:18<30:00,  1.63it/s]
 78%|███████▊  | 10385/13317 [1:47:19<29:59,  1.63it/s]
                                                       
{'loss': 0.1582, 'grad_norm': 0.5922707915306091, 'learning_rate': 1.4054260396361806e-06, 'epoch': 2.34}

 78%|███████▊  | 10385/13317 [1:47:19<29:59,  1.63it/s]
 78%|███████▊  | 10386/13317 [1:47:20<29:58,  1.63it/s]
                                                       
{'loss': 0.1206, 'grad_norm': 0.46487724781036377, 'learning_rate': 1.4045151429346299e-06, 'epoch': 2.34}

 78%|███████▊  | 10386/13317 [1:47:20<29:58,  1.63it/s]
 78%|███████▊  | 10387/13317 [1:47:20<30:00,  1.63it/s]
                                                       
{'loss': 0.2191, 'grad_norm': 0.557555615901947, 'learning_rate': 1.4036044932809584e-06, 'epoch': 2.34}

 78%|███████▊  | 10387/13317 [1:47:20<30:00,  1.63it/s]
 78%|███████▊  | 10388/13317 [1:47:21<29:59,  1.63it/s]
                                                       
{'loss': 0.1782, 'grad_norm': 0.6901652812957764, 'learning_rate': 1.4026940907377361e-06, 'epoch': 2.34}

 78%|███████▊  | 10388/13317 [1:47:21<29:59,  1.63it/s]
 78%|███████▊  | 10389/13317 [1:47:21<29:57,  1.63it/s]
                                                       
{'loss': 0.1373, 'grad_norm': 0.500480592250824, 'learning_rate': 1.40178393536752e-06, 'epoch': 2.34}

 78%|███████▊  | 10389/13317 [1:47:22<29:57,  1.63it/s]
 78%|███████▊  | 10390/13317 [1:47:22<29:56,  1.63it/s]
                                                       
{'loss': 0.1295, 'grad_norm': 0.4923560619354248, 'learning_rate': 1.4008740272328446e-06, 'epoch': 2.34}

 78%|███████▊  | 10390/13317 [1:47:22<29:56,  1.63it/s]
 78%|███████▊  | 10391/13317 [1:47:23<29:56,  1.63it/s]
                                                       
{'loss': 0.1193, 'grad_norm': 0.48163312673568726, 'learning_rate': 1.3999643663962336e-06, 'epoch': 2.34}

 78%|███████▊  | 10391/13317 [1:47:23<29:56,  1.63it/s]
 78%|███████▊  | 10392/13317 [1:47:23<29:57,  1.63it/s]
                                                       
{'loss': 0.1286, 'grad_norm': 0.5867041349411011, 'learning_rate': 1.399054952920188e-06, 'epoch': 2.34}

 78%|███████▊  | 10392/13317 [1:47:23<29:57,  1.63it/s]
 78%|███████▊  | 10393/13317 [1:47:24<29:55,  1.63it/s]
                                                       
{'loss': 0.1701, 'grad_norm': 0.653157651424408, 'learning_rate': 1.3981457868671926e-06, 'epoch': 2.34}

 78%|███████▊  | 10393/13317 [1:47:24<29:55,  1.63it/s]
 78%|███████▊  | 10394/13317 [1:47:24<29:54,  1.63it/s]
                                                       
{'loss': 0.1447, 'grad_norm': 0.5445458292961121, 'learning_rate': 1.3972368682997212e-06, 'epoch': 2.34}

 78%|███████▊  | 10394/13317 [1:47:25<29:54,  1.63it/s]
 78%|███████▊  | 10395/13317 [1:47:25<29:53,  1.63it/s]
                                                       
{'loss': 0.1486, 'grad_norm': 0.6410472393035889, 'learning_rate': 1.3963281972802211e-06, 'epoch': 2.34}

 78%|███████▊  | 10395/13317 [1:47:25<29:53,  1.63it/s]
 78%|███████▊  | 10396/13317 [1:47:26<29:51,  1.63it/s]
                                                       
{'loss': 0.1309, 'grad_norm': 0.5451064705848694, 'learning_rate': 1.3954197738711322e-06, 'epoch': 2.34}

 78%|███████▊  | 10396/13317 [1:47:26<29:51,  1.63it/s]
 78%|███████▊  | 10397/13317 [1:47:26<29:52,  1.63it/s]
                                                       
{'loss': 0.1358, 'grad_norm': 0.5002039074897766, 'learning_rate': 1.3945115981348695e-06, 'epoch': 2.34}

 78%|███████▊  | 10397/13317 [1:47:26<29:52,  1.63it/s]
 78%|███████▊  | 10398/13317 [1:47:27<29:51,  1.63it/s]
                                                       
{'loss': 0.1485, 'grad_norm': 0.5791774392127991, 'learning_rate': 1.3936036701338368e-06, 'epoch': 2.34}

 78%|███████▊  | 10398/13317 [1:47:27<29:51,  1.63it/s]
 78%|███████▊  | 10399/13317 [1:47:28<29:50,  1.63it/s]
                                                       
{'loss': 0.1402, 'grad_norm': 0.5251458883285522, 'learning_rate': 1.3926959899304172e-06, 'epoch': 2.34}

 78%|███████▊  | 10399/13317 [1:47:28<29:50,  1.63it/s]
 78%|███████▊  | 10400/13317 [1:47:28<29:50,  1.63it/s]
                                                       
{'loss': 0.1572, 'grad_norm': 0.594960629940033, 'learning_rate': 1.3917885575869755e-06, 'epoch': 2.34}

 78%|███████▊  | 10400/13317 [1:47:28<29:50,  1.63it/s]
 78%|███████▊  | 10401/13317 [1:47:29<29:50,  1.63it/s]
                                                       
{'loss': 0.1321, 'grad_norm': 0.5160961747169495, 'learning_rate': 1.3908813731658666e-06, 'epoch': 2.34}

 78%|███████▊  | 10401/13317 [1:47:29<29:50,  1.63it/s]
 78%|███████▊  | 10402/13317 [1:47:29<29:53,  1.63it/s]
                                                       
{'loss': 0.141, 'grad_norm': 0.5475253462791443, 'learning_rate': 1.38997443672942e-06, 'epoch': 2.34}

 78%|███████▊  | 10402/13317 [1:47:29<29:53,  1.63it/s]
 78%|███████▊  | 10403/13317 [1:47:30<29:52,  1.63it/s]
                                                       
{'loss': 0.1495, 'grad_norm': 0.535986065864563, 'learning_rate': 1.389067748339954e-06, 'epoch': 2.34}

 78%|███████▊  | 10403/13317 [1:47:30<29:52,  1.63it/s]
 78%|███████▊  | 10404/13317 [1:47:31<29:51,  1.63it/s]
                                                       
{'loss': 0.1896, 'grad_norm': 0.7272257804870605, 'learning_rate': 1.3881613080597655e-06, 'epoch': 2.34}

 78%|███████▊  | 10404/13317 [1:47:31<29:51,  1.63it/s]
 78%|███████▊  | 10405/13317 [1:47:31<29:50,  1.63it/s]
                                                       
{'loss': 0.1255, 'grad_norm': 0.5340612530708313, 'learning_rate': 1.3872551159511394e-06, 'epoch': 2.34}

 78%|███████▊  | 10405/13317 [1:47:31<29:50,  1.63it/s]
 78%|███████▊  | 10406/13317 [1:47:32<29:48,  1.63it/s]
                                                       
{'loss': 0.14, 'grad_norm': 0.5902377963066101, 'learning_rate': 1.3863491720763395e-06, 'epoch': 2.34}

 78%|███████▊  | 10406/13317 [1:47:32<29:48,  1.63it/s]
 78%|███████▊  | 10407/13317 [1:47:32<29:50,  1.63it/s]
                                                       
{'loss': 0.1549, 'grad_norm': 0.5860907435417175, 'learning_rate': 1.3854434764976116e-06, 'epoch': 2.34}

 78%|███████▊  | 10407/13317 [1:47:33<29:50,  1.63it/s]
 78%|███████▊  | 10408/13317 [1:47:33<29:47,  1.63it/s]
                                                       
{'loss': 0.1594, 'grad_norm': 0.5766873359680176, 'learning_rate': 1.3845380292771903e-06, 'epoch': 2.34}

 78%|███████▊  | 10408/13317 [1:47:33<29:47,  1.63it/s]
 78%|███████▊  | 10409/13317 [1:47:34<29:45,  1.63it/s]
                                                       
{'loss': 0.1395, 'grad_norm': 0.5499057769775391, 'learning_rate': 1.3836328304772872e-06, 'epoch': 2.34}

 78%|███████▊  | 10409/13317 [1:47:34<29:45,  1.63it/s]
 78%|███████▊  | 10410/13317 [1:47:34<29:45,  1.63it/s]
                                                       
{'loss': 0.1627, 'grad_norm': 0.6011548638343811, 'learning_rate': 1.382727880160098e-06, 'epoch': 2.35}

 78%|███████▊  | 10410/13317 [1:47:34<29:45,  1.63it/s]
 78%|███████▊  | 10411/13317 [1:47:35<29:44,  1.63it/s]
                                                       
{'loss': 0.1894, 'grad_norm': 0.6914164423942566, 'learning_rate': 1.3818231783878033e-06, 'epoch': 2.35}

 78%|███████▊  | 10411/13317 [1:47:35<29:44,  1.63it/s]
 78%|███████▊  | 10412/13317 [1:47:36<29:43,  1.63it/s]
                                                       
{'loss': 0.1404, 'grad_norm': 0.5376487374305725, 'learning_rate': 1.3809187252225681e-06, 'epoch': 2.35}

 78%|███████▊  | 10412/13317 [1:47:36<29:43,  1.63it/s]
 78%|███████▊  | 10413/13317 [1:47:36<29:44,  1.63it/s]
                                                       
{'loss': 0.1899, 'grad_norm': 0.5877652168273926, 'learning_rate': 1.3800145207265358e-06, 'epoch': 2.35}

 78%|███████▊  | 10413/13317 [1:47:36<29:44,  1.63it/s]
 78%|███████▊  | 10414/13317 [1:47:37<29:42,  1.63it/s]
                                                       
{'loss': 0.1248, 'grad_norm': 0.4941878020763397, 'learning_rate': 1.3791105649618335e-06, 'epoch': 2.35}

 78%|███████▊  | 10414/13317 [1:47:37<29:42,  1.63it/s]
 78%|███████▊  | 10415/13317 [1:47:37<29:41,  1.63it/s]
                                                       
{'loss': 0.1848, 'grad_norm': 0.6460784077644348, 'learning_rate': 1.3782068579905739e-06, 'epoch': 2.35}

 78%|███████▊  | 10415/13317 [1:47:37<29:41,  1.63it/s]
 78%|███████▊  | 10416/13317 [1:47:38<29:40,  1.63it/s]
                                                       
{'loss': 0.1445, 'grad_norm': 0.5531423687934875, 'learning_rate': 1.3773033998748542e-06, 'epoch': 2.35}

 78%|███████▊  | 10416/13317 [1:47:38<29:40,  1.63it/s]
 78%|███████▊  | 10417/13317 [1:47:39<29:41,  1.63it/s]
                                                       
{'loss': 0.1449, 'grad_norm': 0.5691308975219727, 'learning_rate': 1.376400190676746e-06, 'epoch': 2.35}

 78%|███████▊  | 10417/13317 [1:47:39<29:41,  1.63it/s]
 78%|███████▊  | 10418/13317 [1:47:39<29:39,  1.63it/s]
                                                       
{'loss': 0.157, 'grad_norm': 0.5633912086486816, 'learning_rate': 1.3754972304583114e-06, 'epoch': 2.35}

 78%|███████▊  | 10418/13317 [1:47:39<29:39,  1.63it/s]
 78%|███████▊  | 10419/13317 [1:47:40<29:38,  1.63it/s]
                                                       
{'loss': 0.1141, 'grad_norm': 0.4461636245250702, 'learning_rate': 1.3745945192815962e-06, 'epoch': 2.35}

 78%|███████▊  | 10419/13317 [1:47:40<29:38,  1.63it/s]
 78%|███████▊  | 10420/13317 [1:47:40<29:37,  1.63it/s]
                                                       
{'loss': 0.1435, 'grad_norm': 0.5775041580200195, 'learning_rate': 1.3736920572086231e-06, 'epoch': 2.35}

 78%|███████▊  | 10420/13317 [1:47:41<29:37,  1.63it/s]
 78%|███████▊  | 10421/13317 [1:47:41<29:35,  1.63it/s]
                                                       
{'loss': 0.1499, 'grad_norm': 0.6330040693283081, 'learning_rate': 1.372789844301401e-06, 'epoch': 2.35}

 78%|███████▊  | 10421/13317 [1:47:41<29:35,  1.63it/s]
 78%|███████▊  | 10422/13317 [1:47:42<29:37,  1.63it/s]
                                                       
{'loss': 0.1491, 'grad_norm': 0.5686747431755066, 'learning_rate': 1.3718878806219215e-06, 'epoch': 2.35}

 78%|███████▊  | 10422/13317 [1:47:42<29:37,  1.63it/s]
 78%|███████▊  | 10423/13317 [1:47:42<29:36,  1.63it/s]
                                                       
{'loss': 0.1306, 'grad_norm': 0.4685397446155548, 'learning_rate': 1.3709861662321627e-06, 'epoch': 2.35}

 78%|███████▊  | 10423/13317 [1:47:42<29:36,  1.63it/s]
 78%|███████▊  | 10424/13317 [1:47:43<29:35,  1.63it/s]
                                                       
{'loss': 0.1562, 'grad_norm': 0.5634613633155823, 'learning_rate': 1.370084701194076e-06, 'epoch': 2.35}

 78%|███████▊  | 10424/13317 [1:47:43<29:35,  1.63it/s]
 78%|███████▊  | 10425/13317 [1:47:43<29:34,  1.63it/s]
                                                       
{'loss': 0.1471, 'grad_norm': 0.61932373046875, 'learning_rate': 1.3691834855696045e-06, 'epoch': 2.35}

 78%|███████▊  | 10425/13317 [1:47:44<29:34,  1.63it/s]
 78%|███████▊  | 10426/13317 [1:47:44<29:34,  1.63it/s]
                                                       
{'loss': 0.1365, 'grad_norm': 0.5351934432983398, 'learning_rate': 1.3682825194206729e-06, 'epoch': 2.35}

 78%|███████▊  | 10426/13317 [1:47:44<29:34,  1.63it/s]
 78%|███████▊  | 10427/13317 [1:47:45<29:34,  1.63it/s]
                                                       
{'loss': 0.1625, 'grad_norm': 0.6136506199836731, 'learning_rate': 1.367381802809185e-06, 'epoch': 2.35}

 78%|███████▊  | 10427/13317 [1:47:45<29:34,  1.63it/s]
 78%|███████▊  | 10428/13317 [1:47:45<29:34,  1.63it/s]
                                                       
{'loss': 0.1406, 'grad_norm': 0.48006805777549744, 'learning_rate': 1.3664813357970291e-06, 'epoch': 2.35}

 78%|███████▊  | 10428/13317 [1:47:45<29:34,  1.63it/s]
 78%|███████▊  | 10429/13317 [1:47:46<29:33,  1.63it/s]
                                                       
{'loss': 0.1808, 'grad_norm': 0.7956777811050415, 'learning_rate': 1.3655811184460776e-06, 'epoch': 2.35}

 78%|███████▊  | 10429/13317 [1:47:46<29:33,  1.63it/s]
 78%|███████▊  | 10430/13317 [1:47:47<29:31,  1.63it/s]
                                                       
{'loss': 0.1812, 'grad_norm': 0.7458754777908325, 'learning_rate': 1.364681150818188e-06, 'epoch': 2.35}

 78%|███████▊  | 10430/13317 [1:47:47<29:31,  1.63it/s]
 78%|███████▊  | 10431/13317 [1:47:47<29:30,  1.63it/s]
                                                       
{'loss': 0.123, 'grad_norm': 0.4301273226737976, 'learning_rate': 1.3637814329751913e-06, 'epoch': 2.35}

 78%|███████▊  | 10431/13317 [1:47:47<29:30,  1.63it/s]
 78%|███████▊  | 10432/13317 [1:47:48<29:31,  1.63it/s]
                                                       
{'loss': 0.1242, 'grad_norm': 0.4700429439544678, 'learning_rate': 1.3628819649789117e-06, 'epoch': 2.35}

 78%|███████▊  | 10432/13317 [1:47:48<29:31,  1.63it/s]
 78%|███████▊  | 10433/13317 [1:47:48<29:29,  1.63it/s]
                                                       
{'loss': 0.1908, 'grad_norm': 0.6426117420196533, 'learning_rate': 1.361982746891153e-06, 'epoch': 2.35}

 78%|███████▊  | 10433/13317 [1:47:49<29:29,  1.63it/s]
 78%|███████▊  | 10434/13317 [1:47:49<29:29,  1.63it/s]
                                                       
{'loss': 0.1461, 'grad_norm': 0.5440555810928345, 'learning_rate': 1.3610837787736992e-06, 'epoch': 2.35}

 78%|███████▊  | 10434/13317 [1:47:49<29:29,  1.63it/s]
 78%|███████▊  | 10435/13317 [1:47:50<29:28,  1.63it/s]
                                                       
{'loss': 0.1466, 'grad_norm': 0.5394548177719116, 'learning_rate': 1.3601850606883177e-06, 'epoch': 2.35}

 78%|███████▊  | 10435/13317 [1:47:50<29:28,  1.63it/s]
 78%|███████▊  | 10436/13317 [1:47:50<29:27,  1.63it/s]
                                                       
{'loss': 0.116, 'grad_norm': 0.4948824942111969, 'learning_rate': 1.359286592696763e-06, 'epoch': 2.35}

 78%|███████▊  | 10436/13317 [1:47:50<29:27,  1.63it/s]
 78%|███████▊  | 10437/13317 [1:47:51<29:29,  1.63it/s]
                                                       
{'loss': 0.1678, 'grad_norm': 0.7473019957542419, 'learning_rate': 1.358388374860767e-06, 'epoch': 2.35}

 78%|███████▊  | 10437/13317 [1:47:51<29:29,  1.63it/s]
 78%|███████▊  | 10438/13317 [1:47:51<29:27,  1.63it/s]
                                                       
{'loss': 0.1221, 'grad_norm': 0.45806384086608887, 'learning_rate': 1.3574904072420485e-06, 'epoch': 2.35}

 78%|███████▊  | 10438/13317 [1:47:52<29:27,  1.63it/s]
 78%|███████▊  | 10439/13317 [1:47:52<29:26,  1.63it/s]
                                                       
{'loss': 0.1396, 'grad_norm': 0.5333892703056335, 'learning_rate': 1.3565926899023051e-06, 'epoch': 2.35}

 78%|███████▊  | 10439/13317 [1:47:52<29:26,  1.63it/s]
 78%|███████▊  | 10440/13317 [1:47:53<29:25,  1.63it/s]
                                                       
{'loss': 0.1004, 'grad_norm': 0.4523099362850189, 'learning_rate': 1.3556952229032228e-06, 'epoch': 2.35}

 78%|███████▊  | 10440/13317 [1:47:53<29:25,  1.63it/s]
 78%|███████▊  | 10441/13317 [1:47:53<29:24,  1.63it/s]
                                                       
{'loss': 0.1657, 'grad_norm': 0.575038492679596, 'learning_rate': 1.3547980063064647e-06, 'epoch': 2.35}

 78%|███████▊  | 10441/13317 [1:47:53<29:24,  1.63it/s]
 78%|███████▊  | 10442/13317 [1:47:54<29:26,  1.63it/s]
                                                       
{'loss': 0.1235, 'grad_norm': 0.44543737173080444, 'learning_rate': 1.3539010401736784e-06, 'epoch': 2.35}

 78%|███████▊  | 10442/13317 [1:47:54<29:26,  1.63it/s]
 78%|███████▊  | 10443/13317 [1:47:55<29:25,  1.63it/s]
                                                       
{'loss': 0.1539, 'grad_norm': 0.6704545617103577, 'learning_rate': 1.3530043245664976e-06, 'epoch': 2.35}

 78%|███████▊  | 10443/13317 [1:47:55<29:25,  1.63it/s]
 78%|███████▊  | 10444/13317 [1:47:55<29:23,  1.63it/s]
                                                       
{'loss': 0.186, 'grad_norm': 0.6105055809020996, 'learning_rate': 1.3521078595465332e-06, 'epoch': 2.35}

 78%|███████▊  | 10444/13317 [1:47:55<29:23,  1.63it/s]
 78%|███████▊  | 10445/13317 [1:47:56<29:23,  1.63it/s]
                                                       
{'loss': 0.141, 'grad_norm': 0.583052396774292, 'learning_rate': 1.3512116451753848e-06, 'epoch': 2.35}

 78%|███████▊  | 10445/13317 [1:47:56<29:23,  1.63it/s]
 78%|███████▊  | 10446/13317 [1:47:56<29:22,  1.63it/s]
                                                       
{'loss': 0.1561, 'grad_norm': 0.6126515865325928, 'learning_rate': 1.350315681514629e-06, 'epoch': 2.35}

 78%|███████▊  | 10446/13317 [1:47:57<29:22,  1.63it/s]
 78%|███████▊  | 10447/13317 [1:47:57<29:25,  1.63it/s]
                                                       
{'loss': 0.1271, 'grad_norm': 0.5147011876106262, 'learning_rate': 1.3494199686258309e-06, 'epoch': 2.35}

 78%|███████▊  | 10447/13317 [1:47:57<29:25,  1.63it/s]
 78%|███████▊  | 10448/13317 [1:47:58<29:23,  1.63it/s]
                                                       
{'loss': 0.1376, 'grad_norm': 0.49366021156311035, 'learning_rate': 1.3485245065705337e-06, 'epoch': 2.35}

 78%|███████▊  | 10448/13317 [1:47:58<29:23,  1.63it/s]
 78%|███████▊  | 10449/13317 [1:47:58<29:21,  1.63it/s]
                                                       
{'loss': 0.1393, 'grad_norm': 0.5774400234222412, 'learning_rate': 1.3476292954102637e-06, 'epoch': 2.35}

 78%|███████▊  | 10449/13317 [1:47:58<29:21,  1.63it/s]
 78%|███████▊  | 10450/13317 [1:47:59<29:20,  1.63it/s]
                                                       
{'loss': 0.1642, 'grad_norm': 0.625899076461792, 'learning_rate': 1.3467343352065349e-06, 'epoch': 2.35}

 78%|███████▊  | 10450/13317 [1:47:59<29:20,  1.63it/s]
 78%|███████▊  | 10451/13317 [1:47:59<29:19,  1.63it/s]
                                                       
{'loss': 0.1573, 'grad_norm': 0.535287618637085, 'learning_rate': 1.3458396260208373e-06, 'epoch': 2.35}

 78%|███████▊  | 10451/13317 [1:48:00<29:19,  1.63it/s]
 78%|███████▊  | 10452/13317 [1:48:00<29:18,  1.63it/s]
                                                       
{'loss': 0.1507, 'grad_norm': 0.5786340832710266, 'learning_rate': 1.3449451679146491e-06, 'epoch': 2.35}

 78%|███████▊  | 10452/13317 [1:48:00<29:18,  1.63it/s]
 78%|███████▊  | 10453/13317 [1:48:01<29:18,  1.63it/s]
                                                       
{'loss': 0.1744, 'grad_norm': 0.6884837746620178, 'learning_rate': 1.344050960949429e-06, 'epoch': 2.35}

 78%|███████▊  | 10453/13317 [1:48:01<29:18,  1.63it/s]
 79%|███████▊  | 10454/13317 [1:48:01<29:16,  1.63it/s]
                                                       
{'loss': 0.2108, 'grad_norm': 0.8407454490661621, 'learning_rate': 1.3431570051866154e-06, 'epoch': 2.36}

 79%|███████▊  | 10454/13317 [1:48:01<29:16,  1.63it/s]
 79%|███████▊  | 10455/13317 [1:48:02<29:15,  1.63it/s]
                                                       
{'loss': 0.1839, 'grad_norm': 0.6755461692810059, 'learning_rate': 1.3422633006876367e-06, 'epoch': 2.36}

 79%|███████▊  | 10455/13317 [1:48:02<29:15,  1.63it/s]
 79%|███████▊  | 10456/13317 [1:48:03<29:15,  1.63it/s]
                                                       
{'loss': 0.1322, 'grad_norm': 0.519328236579895, 'learning_rate': 1.341369847513897e-06, 'epoch': 2.36}

 79%|███████▊  | 10456/13317 [1:48:03<29:15,  1.63it/s]
 79%|███████▊  | 10457/13317 [1:48:03<29:14,  1.63it/s]
                                                       
{'loss': 0.1297, 'grad_norm': 0.5032452344894409, 'learning_rate': 1.3404766457267875e-06, 'epoch': 2.36}

 79%|███████▊  | 10457/13317 [1:48:03<29:14,  1.63it/s]
 79%|███████▊  | 10458/13317 [1:48:04<29:14,  1.63it/s]
                                                       
{'loss': 0.1341, 'grad_norm': 0.5453071594238281, 'learning_rate': 1.3395836953876795e-06, 'epoch': 2.36}

 79%|███████▊  | 10458/13317 [1:48:04<29:14,  1.63it/s]
 79%|███████▊  | 10459/13317 [1:48:04<29:13,  1.63it/s]
                                                       
{'loss': 0.1665, 'grad_norm': 0.5484411716461182, 'learning_rate': 1.3386909965579297e-06, 'epoch': 2.36}

 79%|███████▊  | 10459/13317 [1:48:04<29:13,  1.63it/s]
 79%|███████▊  | 10460/13317 [1:48:05<29:13,  1.63it/s]
                                                       
{'loss': 0.1873, 'grad_norm': 0.6430842280387878, 'learning_rate': 1.3377985492988748e-06, 'epoch': 2.36}

 79%|███████▊  | 10460/13317 [1:48:05<29:13,  1.63it/s]
 79%|███████▊  | 10461/13317 [1:48:06<29:12,  1.63it/s]
                                                       
{'loss': 0.1189, 'grad_norm': 0.4833471179008484, 'learning_rate': 1.3369063536718347e-06, 'epoch': 2.36}

 79%|███████▊  | 10461/13317 [1:48:06<29:12,  1.63it/s]
 79%|███████▊  | 10462/13317 [1:48:06<29:13,  1.63it/s]
                                                       
{'loss': 0.1453, 'grad_norm': 0.5629954934120178, 'learning_rate': 1.3360144097381145e-06, 'epoch': 2.36}

 79%|███████▊  | 10462/13317 [1:48:06<29:13,  1.63it/s]
 79%|███████▊  | 10463/13317 [1:48:07<29:12,  1.63it/s]
                                                       
{'loss': 0.1516, 'grad_norm': 0.6046710014343262, 'learning_rate': 1.3351227175589982e-06, 'epoch': 2.36}

 79%|███████▊  | 10463/13317 [1:48:07<29:12,  1.63it/s]
 79%|███████▊  | 10464/13317 [1:48:07<29:11,  1.63it/s]
                                                       
{'loss': 0.222, 'grad_norm': 0.46179357171058655, 'learning_rate': 1.3342312771957566e-06, 'epoch': 2.36}

 79%|███████▊  | 10464/13317 [1:48:08<29:11,  1.63it/s]
 79%|███████▊  | 10465/13317 [1:48:08<29:09,  1.63it/s]
                                                       
{'loss': 0.1444, 'grad_norm': 0.5737559795379639, 'learning_rate': 1.3333400887096386e-06, 'epoch': 2.36}

 79%|███████▊  | 10465/13317 [1:48:08<29:09,  1.63it/s]
 79%|███████▊  | 10466/13317 [1:48:09<29:09,  1.63it/s]
                                                       
{'loss': 0.159, 'grad_norm': 0.6167683005332947, 'learning_rate': 1.3324491521618815e-06, 'epoch': 2.36}

 79%|███████▊  | 10466/13317 [1:48:09<29:09,  1.63it/s]
 79%|███████▊  | 10467/13317 [1:48:09<29:08,  1.63it/s]
                                                       
{'loss': 0.1362, 'grad_norm': 0.5225697159767151, 'learning_rate': 1.3315584676137e-06, 'epoch': 2.36}

 79%|███████▊  | 10467/13317 [1:48:09<29:08,  1.63it/s]
 79%|███████▊  | 10468/13317 [1:48:10<29:08,  1.63it/s]
                                                       
{'loss': 0.2166, 'grad_norm': 0.6887527108192444, 'learning_rate': 1.330668035126293e-06, 'epoch': 2.36}

 79%|███████▊  | 10468/13317 [1:48:10<29:08,  1.63it/s]
 79%|███████▊  | 10469/13317 [1:48:10<29:06,  1.63it/s]
                                                       
{'loss': 0.1309, 'grad_norm': 0.5389360785484314, 'learning_rate': 1.3297778547608426e-06, 'epoch': 2.36}

 79%|███████▊  | 10469/13317 [1:48:11<29:06,  1.63it/s]
 79%|███████▊  | 10470/13317 [1:48:11<29:06,  1.63it/s]
                                                       
{'loss': 0.1301, 'grad_norm': 0.5215939283370972, 'learning_rate': 1.3288879265785181e-06, 'epoch': 2.36}

 79%|███████▊  | 10470/13317 [1:48:11<29:06,  1.63it/s]
 79%|███████▊  | 10471/13317 [1:48:12<29:05,  1.63it/s]
                                                       
{'loss': 0.1577, 'grad_norm': 0.5949761271476746, 'learning_rate': 1.3279982506404605e-06, 'epoch': 2.36}

 79%|███████▊  | 10471/13317 [1:48:12<29:05,  1.63it/s]
 79%|███████▊  | 10472/13317 [1:48:12<29:04,  1.63it/s]
                                                       
{'loss': 0.1656, 'grad_norm': 0.6494801044464111, 'learning_rate': 1.3271088270078026e-06, 'epoch': 2.36}

 79%|███████▊  | 10472/13317 [1:48:12<29:04,  1.63it/s]
 79%|███████▊  | 10473/13317 [1:48:13<29:03,  1.63it/s]
                                                       
{'loss': 0.163, 'grad_norm': 0.5899677276611328, 'learning_rate': 1.326219655741659e-06, 'epoch': 2.36}

 79%|███████▊  | 10473/13317 [1:48:13<29:03,  1.63it/s]
 79%|███████▊  | 10474/13317 [1:48:14<29:04,  1.63it/s]
                                                       
{'loss': 0.1343, 'grad_norm': 0.5348420739173889, 'learning_rate': 1.3253307369031232e-06, 'epoch': 2.36}

 79%|███████▊  | 10474/13317 [1:48:14<29:04,  1.63it/s]
 79%|███████▊  | 10475/13317 [1:48:14<29:03,  1.63it/s]
                                                       
{'loss': 0.1594, 'grad_norm': 0.6265484690666199, 'learning_rate': 1.324442070553273e-06, 'epoch': 2.36}

 79%|███████▊  | 10475/13317 [1:48:14<29:03,  1.63it/s]
 79%|███████▊  | 10476/13317 [1:48:15<29:02,  1.63it/s]
                                                       
{'loss': 0.1238, 'grad_norm': 0.5289396643638611, 'learning_rate': 1.3235536567531698e-06, 'epoch': 2.36}

 79%|███████▊  | 10476/13317 [1:48:15<29:02,  1.63it/s]
 79%|███████▊  | 10477/13317 [1:48:15<29:02,  1.63it/s]
                                                       
{'loss': 0.1691, 'grad_norm': 0.59294193983078, 'learning_rate': 1.3226654955638602e-06, 'epoch': 2.36}

 79%|███████▊  | 10477/13317 [1:48:16<29:02,  1.63it/s]
 79%|███████▊  | 10478/13317 [1:48:16<29:01,  1.63it/s]
                                                       
{'loss': 0.1589, 'grad_norm': 0.6515262722969055, 'learning_rate': 1.321777587046364e-06, 'epoch': 2.36}

 79%|███████▊  | 10478/13317 [1:48:16<29:01,  1.63it/s]
 79%|███████▊  | 10479/13317 [1:48:17<29:00,  1.63it/s]
                                                       
{'loss': 0.1532, 'grad_norm': 0.675297200679779, 'learning_rate': 1.3208899312616935e-06, 'epoch': 2.36}

 79%|███████▊  | 10479/13317 [1:48:17<29:00,  1.63it/s]
 79%|███████▊  | 10480/13317 [1:48:17<28:59,  1.63it/s]
                                                       
{'loss': 0.1723, 'grad_norm': 0.6153677105903625, 'learning_rate': 1.3200025282708412e-06, 'epoch': 2.36}

 79%|███████▊  | 10480/13317 [1:48:17<28:59,  1.63it/s]
 79%|███████▊  | 10481/13317 [1:48:18<28:59,  1.63it/s]
                                                       
{'loss': 0.1385, 'grad_norm': 0.5390361547470093, 'learning_rate': 1.3191153781347798e-06, 'epoch': 2.36}

 79%|███████▊  | 10481/13317 [1:48:18<28:59,  1.63it/s]
 79%|███████▊  | 10482/13317 [1:48:18<28:59,  1.63it/s]
                                                       
{'loss': 0.1416, 'grad_norm': 0.5758960843086243, 'learning_rate': 1.318228480914464e-06, 'epoch': 2.36}

 79%|███████▊  | 10482/13317 [1:48:19<28:59,  1.63it/s]
 79%|███████▊  | 10483/13317 [1:48:19<28:58,  1.63it/s]
                                                       
{'loss': 0.1777, 'grad_norm': 0.6381306052207947, 'learning_rate': 1.317341836670835e-06, 'epoch': 2.36}

 79%|███████▊  | 10483/13317 [1:48:19<28:58,  1.63it/s]
 79%|███████▊  | 10484/13317 [1:48:20<28:57,  1.63it/s]
                                                       
{'loss': 0.1369, 'grad_norm': 0.49241092801094055, 'learning_rate': 1.316455445464817e-06, 'epoch': 2.36}

 79%|███████▊  | 10484/13317 [1:48:20<28:57,  1.63it/s]
 79%|███████▊  | 10485/13317 [1:48:20<28:56,  1.63it/s]
                                                       
{'loss': 0.1208, 'grad_norm': 0.544287919998169, 'learning_rate': 1.3155693073573095e-06, 'epoch': 2.36}

 79%|███████▊  | 10485/13317 [1:48:20<28:56,  1.63it/s]
 79%|███████▊  | 10486/13317 [1:48:21<28:56,  1.63it/s]
                                                       
{'loss': 0.117, 'grad_norm': 0.46863022446632385, 'learning_rate': 1.3146834224092015e-06, 'epoch': 2.36}

 79%|███████▊  | 10486/13317 [1:48:21<28:56,  1.63it/s]
 79%|███████▊  | 10487/13317 [1:48:22<28:55,  1.63it/s]
                                                       
{'loss': 0.1419, 'grad_norm': 0.5626978278160095, 'learning_rate': 1.3137977906813648e-06, 'epoch': 2.36}

 79%|███████▊  | 10487/13317 [1:48:22<28:55,  1.63it/s]
 79%|███████▉  | 10488/13317 [1:48:22<28:55,  1.63it/s]
                                                       
{'loss': 0.1306, 'grad_norm': 0.4755609631538391, 'learning_rate': 1.3129124122346505e-06, 'epoch': 2.36}

 79%|███████▉  | 10488/13317 [1:48:22<28:55,  1.63it/s]
 79%|███████▉  | 10489/13317 [1:48:23<28:55,  1.63it/s]
                                                       
{'loss': 0.1254, 'grad_norm': 0.4631372094154358, 'learning_rate': 1.3120272871298907e-06, 'epoch': 2.36}

 79%|███████▉  | 10489/13317 [1:48:23<28:55,  1.63it/s]
 79%|███████▉  | 10490/13317 [1:48:23<28:55,  1.63it/s]
                                                       
{'loss': 0.1335, 'grad_norm': 0.49804991483688354, 'learning_rate': 1.3111424154279056e-06, 'epoch': 2.36}

 79%|███████▉  | 10490/13317 [1:48:24<28:55,  1.63it/s]
 79%|███████▉  | 10491/13317 [1:48:24<28:56,  1.63it/s]
                                                       
{'loss': 0.15, 'grad_norm': 0.5546602606773376, 'learning_rate': 1.3102577971894958e-06, 'epoch': 2.36}

 79%|███████▉  | 10491/13317 [1:48:24<28:56,  1.63it/s]
 79%|███████▉  | 10492/13317 [1:48:25<28:55,  1.63it/s]
                                                       
{'loss': 0.1254, 'grad_norm': 0.4572524130344391, 'learning_rate': 1.3093734324754437e-06, 'epoch': 2.36}

 79%|███████▉  | 10492/13317 [1:48:25<28:55,  1.63it/s]
 79%|███████▉  | 10493/13317 [1:48:25<28:53,  1.63it/s]
                                                       
{'loss': 0.1548, 'grad_norm': 0.5838390588760376, 'learning_rate': 1.3084893213465116e-06, 'epoch': 2.36}

 79%|███████▉  | 10493/13317 [1:48:25<28:53,  1.63it/s]
 79%|███████▉  | 10494/13317 [1:48:26<28:53,  1.63it/s]
                                                       
{'loss': 0.1582, 'grad_norm': 0.6428651213645935, 'learning_rate': 1.3076054638634516e-06, 'epoch': 2.36}

 79%|███████▉  | 10494/13317 [1:48:26<28:53,  1.63it/s]
 79%|███████▉  | 10495/13317 [1:48:26<28:51,  1.63it/s]
                                                       
{'loss': 0.1645, 'grad_norm': 0.6179766654968262, 'learning_rate': 1.3067218600869914e-06, 'epoch': 2.36}

 79%|███████▉  | 10495/13317 [1:48:27<28:51,  1.63it/s]
 79%|███████▉  | 10496/13317 [1:48:27<28:50,  1.63it/s]
                                                       
{'loss': 0.1498, 'grad_norm': 0.5021267533302307, 'learning_rate': 1.3058385100778432e-06, 'epoch': 2.36}

 79%|███████▉  | 10496/13317 [1:48:27<28:50,  1.63it/s]
 79%|███████▉  | 10497/13317 [1:48:28<28:49,  1.63it/s]
                                                       
{'loss': 0.1655, 'grad_norm': 0.6002082228660583, 'learning_rate': 1.3049554138967052e-06, 'epoch': 2.36}

 79%|███████▉  | 10497/13317 [1:48:28<28:49,  1.63it/s]
 79%|███████▉  | 10498/13317 [1:48:28<28:49,  1.63it/s]
                                                       
{'loss': 0.1348, 'grad_norm': 0.5121265053749084, 'learning_rate': 1.3040725716042524e-06, 'epoch': 2.36}

 79%|███████▉  | 10498/13317 [1:48:28<28:49,  1.63it/s]
 79%|███████▉  | 10499/13317 [1:48:29<28:48,  1.63it/s]
                                                       
{'loss': 0.1204, 'grad_norm': 0.4958946108818054, 'learning_rate': 1.3031899832611488e-06, 'epoch': 2.37}

 79%|███████▉  | 10499/13317 [1:48:29<28:48,  1.63it/s]
 79%|███████▉  | 10500/13317 [1:48:30<28:48,  1.63it/s]
                                                       
{'loss': 0.1578, 'grad_norm': 0.49950987100601196, 'learning_rate': 1.3023076489280345e-06, 'epoch': 2.37}

 79%|███████▉  | 10500/13317 [1:48:30<28:48,  1.63it/s]

  0%|          | 0/13 [00:00<?, ?it/s]

 15%|█▌        | 2/13 [00:00<00:00, 13.39it/s]

 31%|███       | 4/13 [00:00<00:01,  8.41it/s]

 38%|███▊      | 5/13 [00:00<00:01,  7.81it/s]

 46%|████▌     | 6/13 [00:00<00:00,  7.43it/s]

 54%|█████▍    | 7/13 [00:00<00:00,  7.19it/s]

 62%|██████▏   | 8/13 [00:01<00:00,  7.03it/s]

 69%|██████▉   | 9/13 [00:01<00:00,  6.91it/s]

 77%|███████▋  | 10/13 [00:01<00:00,  6.84it/s]

 85%|████████▍ | 11/13 [00:01<00:00,  6.79it/s]

 92%|█████████▏| 12/13 [00:01<00:00,  6.74it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]
                                                       

                                               
{'eval_loss': 0.5470743775367737, 'eval_runtime': 1.9513, 'eval_samples_per_second': 102.496, 'eval_steps_per_second': 6.662, 'epoch': 2.37}

 79%|███████▉  | 10500/13317 [1:48:32<28:48,  1.63it/s]

100%|██████████| 13/13 [00:01<00:00,  6.76it/s]

                                               
 79%|███████▉  | 10501/13317 [1:48:32<56:20,  1.20s/it]
                                                       
{'loss': 0.1438, 'grad_norm': 0.5133366584777832, 'learning_rate': 1.3014255686655374e-06, 'epoch': 2.37}

 79%|███████▉  | 10501/13317 [1:48:32<56:20,  1.20s/it]
 79%|███████▉  | 10502/13317 [1:48:33<48:03,  1.02s/it]
                                                       
{'loss': 0.1305, 'grad_norm': 0.48089104890823364, 'learning_rate': 1.3005437425342653e-06, 'epoch': 2.37}

 79%|███████▉  | 10502/13317 [1:48:33<48:03,  1.02s/it]
 79%|███████▉  | 10503/13317 [1:48:33<42:16,  1.11it/s]
                                                       
{'loss': 0.1744, 'grad_norm': 0.626433253288269, 'learning_rate': 1.299662170594807e-06, 'epoch': 2.37}

 79%|███████▉  | 10503/13317 [1:48:33<42:16,  1.11it/s]
 79%|███████▉  | 10504/13317 [1:48:34<38:12,  1.23it/s]
                                                       
{'loss': 0.129, 'grad_norm': 0.5146748423576355, 'learning_rate': 1.298780852907739e-06, 'epoch': 2.37}

 79%|███████▉  | 10504/13317 [1:48:34<38:12,  1.23it/s]
 79%|███████▉  | 10505/13317 [1:48:35<35:21,  1.33it/s]
                                                       
{'loss': 0.1807, 'grad_norm': 0.6430657505989075, 'learning_rate': 1.2978997895336137e-06, 'epoch': 2.37}

 79%|███████▉  | 10505/13317 [1:48:35<35:21,  1.33it/s]
 79%|███████▉  | 10506/13317 [1:48:35<33:21,  1.40it/s]
                                                       
{'loss': 0.1444, 'grad_norm': 0.5028954148292542, 'learning_rate': 1.297018980532973e-06, 'epoch': 2.37}

 79%|███████▉  | 10506/13317 [1:48:35<33:21,  1.40it/s]
 79%|███████▉  | 10507/13317 [1:48:36<31:58,  1.46it/s]
                                                       
{'loss': 0.1741, 'grad_norm': 0.6334424614906311, 'learning_rate': 1.2961384259663345e-06, 'epoch': 2.37}

 79%|███████▉  | 10507/13317 [1:48:36<31:58,  1.46it/s]
 79%|███████▉  | 10508/13317 [1:48:36<30:59,  1.51it/s]
                                                       
{'loss': 0.174, 'grad_norm': 0.6902424097061157, 'learning_rate': 1.2952581258942048e-06, 'epoch': 2.37}

 79%|███████▉  | 10508/13317 [1:48:37<30:59,  1.51it/s]
 79%|███████▉  | 10509/13317 [1:48:37<30:18,  1.54it/s]
                                                       
{'loss': 0.1511, 'grad_norm': 0.5797167420387268, 'learning_rate': 1.2943780803770683e-06, 'epoch': 2.37}

 79%|███████▉  | 10509/13317 [1:48:37<30:18,  1.54it/s]
 79%|███████▉  | 10510/13317 [1:48:38<29:49,  1.57it/s]
                                                       
{'loss': 0.1376, 'grad_norm': 0.5443874597549438, 'learning_rate': 1.2934982894753922e-06, 'epoch': 2.37}

 79%|███████▉  | 10510/13317 [1:48:38<29:49,  1.57it/s]
 79%|███████▉  | 10511/13317 [1:48:38<29:27,  1.59it/s]
                                                       
{'loss': 0.1622, 'grad_norm': 0.5865097641944885, 'learning_rate': 1.2926187532496298e-06, 'epoch': 2.37}

 79%|███████▉  | 10511/13317 [1:48:38<29:27,  1.59it/s]
 79%|███████▉  | 10512/13317 [1:48:39<29:12,  1.60it/s]
                                                       
{'loss': 0.1716, 'grad_norm': 0.6126824617385864, 'learning_rate': 1.2917394717602123e-06, 'epoch': 2.37}

 79%|███████▉  | 10512/13317 [1:48:39<29:12,  1.60it/s]
 79%|███████▉  | 10513/13317 [1:48:39<29:03,  1.61it/s]
                                                       
{'loss': 0.1527, 'grad_norm': 0.5726668834686279, 'learning_rate': 1.2908604450675583e-06, 'epoch': 2.37}

 79%|███████▉  | 10513/13317 [1:48:40<29:03,  1.61it/s]
 79%|███████▉  | 10514/13317 [1:48:40<28:55,  1.62it/s]
                                                       
{'loss': 0.1341, 'grad_norm': 0.4948689341545105, 'learning_rate': 1.2899816732320642e-06, 'epoch': 2.37}

 79%|███████▉  | 10514/13317 [1:48:40<28:55,  1.62it/s]
 79%|███████▉  | 10515/13317 [1:48:41<28:50,  1.62it/s]
                                                       
{'loss': 0.169, 'grad_norm': 0.6704115271568298, 'learning_rate': 1.2891031563141104e-06, 'epoch': 2.37}

 79%|███████▉  | 10515/13317 [1:48:41<28:50,  1.62it/s]
 79%|███████▉  | 10516/13317 [1:48:41<28:46,  1.62it/s]
                                                       
{'loss': 0.1419, 'grad_norm': 0.7611289620399475, 'learning_rate': 1.288224894374061e-06, 'epoch': 2.37}

 79%|███████▉  | 10516/13317 [1:48:41<28:46,  1.62it/s]
 79%|███████▉  | 10517/13317 [1:48:42<28:42,  1.63it/s]
                                                       
{'loss': 0.1437, 'grad_norm': 0.5004695057868958, 'learning_rate': 1.287346887472264e-06, 'epoch': 2.37}

 79%|███████▉  | 10517/13317 [1:48:42<28:42,  1.63it/s]
 79%|███████▉  | 10518/13317 [1:48:43<28:41,  1.63it/s]
                                                       
{'loss': 0.1333, 'grad_norm': 0.5466312170028687, 'learning_rate': 1.2864691356690462e-06, 'epoch': 2.37}

 79%|███████▉  | 10518/13317 [1:48:43<28:41,  1.63it/s]
 79%|███████▉  | 10519/13317 [1:48:43<28:40,  1.63it/s]
                                                       
{'loss': 0.1532, 'grad_norm': 0.5772519111633301, 'learning_rate': 1.2855916390247165e-06, 'epoch': 2.37}

 79%|███████▉  | 10519/13317 [1:48:43<28:40,  1.63it/s]
 79%|███████▉  | 10520/13317 [1:48:44<28:38,  1.63it/s]
                                                       
{'loss': 0.1525, 'grad_norm': 0.6291557550430298, 'learning_rate': 1.2847143975995718e-06, 'epoch': 2.37}

 79%|███████▉  | 10520/13317 [1:48:44<28:38,  1.63it/s]
 79%|███████▉  | 10521/13317 [1:48:44<28:37,  1.63it/s]
                                                       
{'loss': 0.1305, 'grad_norm': 0.49966755509376526, 'learning_rate': 1.2838374114538853e-06, 'epoch': 2.37}

 79%|███████▉  | 10521/13317 [1:48:44<28:37,  1.63it/s]
 79%|███████▉  | 10522/13317 [1:48:45<28:36,  1.63it/s]
                                                       
{'loss': 0.1309, 'grad_norm': 0.5395436882972717, 'learning_rate': 1.2829606806479155e-06, 'epoch': 2.37}

 79%|███████▉  | 10522/13317 [1:48:45<28:36,  1.63it/s]
 79%|███████▉  | 10523/13317 [1:48:46<28:34,  1.63it/s]
                                                       
{'loss': 0.1347, 'grad_norm': 0.66013503074646, 'learning_rate': 1.282084205241903e-06, 'epoch': 2.37}

 79%|███████▉  | 10523/13317 [1:48:46<28:34,  1.63it/s]
 79%|███████▉  | 10524/13317 [1:48:46<28:34,  1.63it/s]
                                                       
{'loss': 0.1616, 'grad_norm': 0.6058743596076965, 'learning_rate': 1.281207985296073e-06, 'epoch': 2.37}

 79%|███████▉  | 10524/13317 [1:48:46<28:34,  1.63it/s]
 79%|███████▉  | 10525/13317 [1:48:47<28:34,  1.63it/s]
                                                       
{'loss': 0.1376, 'grad_norm': 0.5575799942016602, 'learning_rate': 1.28033202087063e-06, 'epoch': 2.37}

 79%|███████▉  | 10525/13317 [1:48:47<28:34,  1.63it/s]
 79%|███████▉  | 10526/13317 [1:48:47<28:33,  1.63it/s]
                                                       
{'loss': 0.1727, 'grad_norm': 0.5291580557823181, 'learning_rate': 1.2794563120257597e-06, 'epoch': 2.37}

 79%|███████▉  | 10526/13317 [1:48:48<28:33,  1.63it/s]
 79%|███████▉  | 10527/13317 [1:48:48<28:32,  1.63it/s]
                                                       
{'loss': 0.1809, 'grad_norm': 0.6829010248184204, 'learning_rate': 1.2785808588216365e-06, 'epoch': 2.37}

 79%|███████▉  | 10527/13317 [1:48:48<28:32,  1.63it/s]
 79%|███████▉  | 10528/13317 [1:48:49<28:32,  1.63it/s]
                                                       
{'loss': 0.1429, 'grad_norm': 0.5027305483818054, 'learning_rate': 1.2777056613184109e-06, 'epoch': 2.37}

 79%|███████▉  | 10528/13317 [1:48:49<28:32,  1.63it/s]
 79%|███████▉  | 10529/13317 [1:48:49<28:32,  1.63it/s]
                                                       
{'loss': 0.1575, 'grad_norm': 0.5248318910598755, 'learning_rate': 1.276830719576217e-06, 'epoch': 2.37}

 79%|███████▉  | 10529/13317 [1:48:49<28:32,  1.63it/s]
 79%|███████▉  | 10530/13317 [1:48:50<28:30,  1.63it/s]
                                                       
{'loss': 0.1597, 'grad_norm': 0.629015326499939, 'learning_rate': 1.2759560336551747e-06, 'epoch': 2.37}

 79%|███████▉  | 10530/13317 [1:48:50<28:30,  1.63it/s]
 79%|███████▉  | 10531/13317 [1:48:51<28:31,  1.63it/s]
                                                       
{'loss': 0.1469, 'grad_norm': 0.5237376093864441, 'learning_rate': 1.2750816036153863e-06, 'epoch': 2.37}

 79%|███████▉  | 10531/13317 [1:48:51<28:31,  1.63it/s]
 79%|███████▉  | 10532/13317 [1:48:51<28:30,  1.63it/s]
                                                       
{'loss': 0.1559, 'grad_norm': 0.5575289130210876, 'learning_rate': 1.2742074295169293e-06, 'epoch': 2.37}

 79%|███████▉  | 10532/13317 [1:48:51<28:30,  1.63it/s]
 79%|███████▉  | 10533/13317 [1:48:52<28:28,  1.63it/s]
                                                       
{'loss': 0.163, 'grad_norm': 0.6416277885437012, 'learning_rate': 1.273333511419871e-06, 'epoch': 2.37}

 79%|███████▉  | 10533/13317 [1:48:52<28:28,  1.63it/s]
 79%|███████▉  | 10534/13317 [1:48:52<28:28,  1.63it/s]
                                                       
{'loss': 0.1454, 'grad_norm': 0.5101035833358765, 'learning_rate': 1.27245984938426e-06, 'epoch': 2.37}

 79%|███████▉  | 10534/13317 [1:48:52<28:28,  1.63it/s]
 79%|███████▉  | 10535/13317 [1:48:53<28:26,  1.63it/s]
                                                       
{'loss': 0.1273, 'grad_norm': 0.5074505805969238, 'learning_rate': 1.2715864434701258e-06, 'epoch': 2.37}

 79%|███████▉  | 10535/13317 [1:48:53<28:26,  1.63it/s]
 79%|███████▉  | 10536/13317 [1:48:54<28:25,  1.63it/s]
                                                       
{'loss': 0.1783, 'grad_norm': 0.7777487635612488, 'learning_rate': 1.2707132937374783e-06, 'epoch': 2.37}

 79%|███████▉  | 10536/13317 [1:48:54<28:25,  1.63it/s]
 79%|███████▉  | 10537/13317 [1:48:54<28:27,  1.63it/s]
                                                       
{'loss': 0.1399, 'grad_norm': 0.5694535374641418, 'learning_rate': 1.269840400246314e-06, 'epoch': 2.37}

 79%|███████▉  | 10537/13317 [1:48:54<28:27,  1.63it/s]
 79%|███████▉  | 10538/13317 [1:48:55<28:25,  1.63it/s]
                                                       
{'loss': 0.2058, 'grad_norm': 0.6407250165939331, 'learning_rate': 1.2689677630566111e-06, 'epoch': 2.37}

 79%|███████▉  | 10538/13317 [1:48:55<28:25,  1.63it/s]
 79%|███████▉  | 10539/13317 [1:48:55<28:24,  1.63it/s]
                                                       
{'loss': 0.1768, 'grad_norm': 0.6412564516067505, 'learning_rate': 1.2680953822283276e-06, 'epoch': 2.37}

 79%|███████▉  | 10539/13317 [1:48:56<28:24,  1.63it/s]
 79%|███████▉  | 10540/13317 [1:48:56<28:23,  1.63it/s]
                                                       
{'loss': 0.1466, 'grad_norm': 0.5370169281959534, 'learning_rate': 1.267223257821404e-06, 'epoch': 2.37}

 79%|███████▉  | 10540/13317 [1:48:56<28:23,  1.63it/s]
 79%|███████▉  | 10541/13317 [1:48:57<28:22,  1.63it/s]
                                                       
{'loss': 0.1624, 'grad_norm': 0.5867446064949036, 'learning_rate': 1.2663513898957674e-06, 'epoch': 2.37}

 79%|███████▉  | 10541/13317 [1:48:57<28:22,  1.63it/s]
 79%|███████▉  | 10542/13317 [1:48:57<28:22,  1.63it/s]
                                                       
{'loss': 0.2498, 'grad_norm': 0.6228817105293274, 'learning_rate': 1.2654797785113226e-06, 'epoch': 2.37}

 79%|███████▉  | 10542/13317 [1:48:57<28:22,  1.63it/s]
 79%|███████▉  | 10543/13317 [1:48:58<28:22,  1.63it/s]
                                                       
{'loss': 0.1561, 'grad_norm': 0.6144649982452393, 'learning_rate': 1.2646084237279576e-06, 'epoch': 2.38}

 79%|███████▉  | 10543/13317 [1:48:58<28:22,  1.63it/s]
 79%|███████▉  | 10544/13317 [1:48:58<28:26,  1.62it/s]
                                                       
{'loss': 0.1578, 'grad_norm': 0.5000789165496826, 'learning_rate': 1.2637373256055445e-06, 'epoch': 2.38}

 79%|███████▉  | 10544/13317 [1:48:59<28:26,  1.62it/s]
 79%|███████▉  | 10545/13317 [1:48:59<28:22,  1.63it/s]
                                                       
{'loss': 0.1432, 'grad_norm': 0.5275683999061584, 'learning_rate': 1.262866484203939e-06, 'epoch': 2.38}

 79%|███████▉  | 10545/13317 [1:48:59<28:22,  1.63it/s]
 79%|███████▉  | 10546/13317 [1:49:00<28:21,  1.63it/s]
                                                       
{'loss': 0.1465, 'grad_norm': 0.5734187960624695, 'learning_rate': 1.2619958995829757e-06, 'epoch': 2.38}

 79%|███████▉  | 10546/13317 [1:49:00<28:21,  1.63it/s]
 79%|███████▉  | 10547/13317 [1:49:00<28:20,  1.63it/s]
                                                       
{'loss': 0.129, 'grad_norm': 0.5069543123245239, 'learning_rate': 1.2611255718024705e-06, 'epoch': 2.38}

 79%|███████▉  | 10547/13317 [1:49:00<28:20,  1.63it/s]
 79%|███████▉  | 10548/13317 [1:49:01<28:19,  1.63it/s]
                                                       
{'loss': 0.1315, 'grad_norm': 0.5462568402290344, 'learning_rate': 1.2602555009222285e-06, 'epoch': 2.38}

 79%|███████▉  | 10548/13317 [1:49:01<28:19,  1.63it/s]
 79%|███████▉  | 10549/13317 [1:49:02<28:18,  1.63it/s]
                                                       
{'loss': 0.1622, 'grad_norm': 0.5528610348701477, 'learning_rate': 1.2593856870020304e-06, 'epoch': 2.38}

 79%|███████▉  | 10549/13317 [1:49:02<28:18,  1.63it/s]
 79%|███████▉  | 10550/13317 [1:49:02<28:18,  1.63it/s]
                                                       
{'loss': 0.1318, 'grad_norm': 0.525002658367157, 'learning_rate': 1.2585161301016407e-06, 'epoch': 2.38}

 79%|███████▉  | 10550/13317 [1:49:02<28:18,  1.63it/s]
 79%|███████▉  | 10551/13317 [1:49:03<28:17,  1.63it/s]
                                                       
{'loss': 0.1595, 'grad_norm': 0.5765630006790161, 'learning_rate': 1.2576468302808082e-06, 'epoch': 2.38}

 79%|███████▉  | 10551/13317 [1:49:03<28:17,  1.63it/s]
 79%|███████▉  | 10552/13317 [1:49:03<28:18,  1.63it/s]
                                                       
{'loss': 0.136, 'grad_norm': 0.5679701566696167, 'learning_rate': 1.2567777875992643e-06, 'epoch': 2.38}

 79%|███████▉  | 10552/13317 [1:49:04<28:18,  1.63it/s]
 79%|███████▉  | 10553/13317 [1:49:04<28:17,  1.63it/s]
                                                       
{'loss': 0.1529, 'grad_norm': 0.537553608417511, 'learning_rate': 1.2559090021167202e-06, 'epoch': 2.38}

 79%|███████▉  | 10553/13317 [1:49:04<28:17,  1.63it/s]
 79%|███████▉  | 10554/13317 [1:49:05<28:16,  1.63it/s]
                                                       
{'loss': 0.1503, 'grad_norm': 0.603168249130249, 'learning_rate': 1.2550404738928695e-06, 'epoch': 2.38}

 79%|███████▉  | 10554/13317 [1:49:05<28:16,  1.63it/s]
 79%|███████▉  | 10555/13317 [1:49:05<28:15,  1.63it/s]
                                                       
{'loss': 0.1567, 'grad_norm': 0.5241350531578064, 'learning_rate': 1.254172202987392e-06, 'epoch': 2.38}

 79%|███████▉  | 10555/13317 [1:49:05<28:15,  1.63it/s]
 79%|███████▉  | 10556/13317 [1:49:06<28:14,  1.63it/s]
                                                       
{'loss': 0.1594, 'grad_norm': 0.5134358406066895, 'learning_rate': 1.2533041894599456e-06, 'epoch': 2.38}

 79%|███████▉  | 10556/13317 [1:49:06<28:14,  1.63it/s]
 79%|███████▉  | 10557/13317 [1:49:06<28:16,  1.63it/s]
                                                       
{'loss': 0.1848, 'grad_norm': 0.5815954208374023, 'learning_rate': 1.2524364333701705e-06, 'epoch': 2.38}

 79%|███████▉  | 10557/13317 [1:49:07<28:16,  1.63it/s]
 79%|███████▉  | 10558/13317 [1:49:07<28:15,  1.63it/s]
                                                       
{'loss': 0.1845, 'grad_norm': 0.6518412828445435, 'learning_rate': 1.2515689347776937e-06, 'epoch': 2.38}

 79%|███████▉  | 10558/13317 [1:49:07<28:15,  1.63it/s]
 79%|███████▉  | 10559/13317 [1:49:08<28:14,  1.63it/s]
                                                       
{'loss': 0.1455, 'grad_norm': 0.6328931450843811, 'learning_rate': 1.2507016937421184e-06, 'epoch': 2.38}

 79%|███████▉  | 10559/13317 [1:49:08<28:14,  1.63it/s]
 79%|███████▉  | 10560/13317 [1:49:08<28:12,  1.63it/s]
                                                       
{'loss': 0.1164, 'grad_norm': 1.0531030893325806, 'learning_rate': 1.2498347103230368e-06, 'epoch': 2.38}

 79%|███████▉  | 10560/13317 [1:49:08<28:12,  1.63it/s]
 79%|███████▉  | 10561/13317 [1:49:09<28:10,  1.63it/s]
                                                       
{'loss': 0.1558, 'grad_norm': 0.5593199729919434, 'learning_rate': 1.2489679845800168e-06, 'epoch': 2.38}

 79%|███████▉  | 10561/13317 [1:49:09<28:10,  1.63it/s]
 79%|███████▉  | 10562/13317 [1:49:10<28:12,  1.63it/s]
                                                       
{'loss': 0.1518, 'grad_norm': 0.656342625617981, 'learning_rate': 1.248101516572614e-06, 'epoch': 2.38}

 79%|███████▉  | 10562/13317 [1:49:10<28:12,  1.63it/s]
 79%|███████▉  | 10563/13317 [1:49:10<28:11,  1.63it/s]
                                                       
{'loss': 0.1371, 'grad_norm': 0.5204088091850281, 'learning_rate': 1.2472353063603626e-06, 'epoch': 2.38}

 79%|███████▉  | 10563/13317 [1:49:10<28:11,  1.63it/s]
 79%|███████▉  | 10564/13317 [1:49:11<28:10,  1.63it/s]
                                                       
{'loss': 0.1498, 'grad_norm': 0.5431369543075562, 'learning_rate': 1.2463693540027787e-06, 'epoch': 2.38}

 79%|███████▉  | 10564/13317 [1:49:11<28:10,  1.63it/s]
 79%|███████▉  | 10565/13317 [1:49:11<28:10,  1.63it/s]
                                                       
{'loss': 0.1382, 'grad_norm': 0.5374271869659424, 'learning_rate': 1.2455036595593666e-06, 'epoch': 2.38}

 79%|███████▉  | 10565/13317 [1:49:12<28:10,  1.63it/s]
 79%|███████▉  | 10566/13317 [1:49:12<28:09,  1.63it/s]
                                                       
{'loss': 0.123, 'grad_norm': 0.586236834526062, 'learning_rate': 1.2446382230896042e-06, 'epoch': 2.38}

 79%|███████▉  | 10566/13317 [1:49:12<28:09,  1.63it/s]
 79%|███████▉  | 10567/13317 [1:49:13<28:09,  1.63it/s]
                                                       
{'loss': 0.1458, 'grad_norm': 0.5290654301643372, 'learning_rate': 1.24377304465296e-06, 'epoch': 2.38}

 79%|███████▉  | 10567/13317 [1:49:13<28:09,  1.63it/s]
 79%|███████▉  | 10568/13317 [1:49:13<28:08,  1.63it/s]
                                                       
{'loss': 0.1957, 'grad_norm': 0.6260567307472229, 'learning_rate': 1.242908124308878e-06, 'epoch': 2.38}

 79%|███████▉  | 10568/13317 [1:49:13<28:08,  1.63it/s]
 79%|███████▉  | 10569/13317 [1:49:14<28:07,  1.63it/s]
                                                       
{'loss': 0.1889, 'grad_norm': 0.6105358004570007, 'learning_rate': 1.2420434621167898e-06, 'epoch': 2.38}

 79%|███████▉  | 10569/13317 [1:49:14<28:07,  1.63it/s]
 79%|███████▉  | 10570/13317 [1:49:14<28:05,  1.63it/s]
                                                       
{'loss': 0.1341, 'grad_norm': 0.5142709016799927, 'learning_rate': 1.241179058136104e-06, 'epoch': 2.38}

 79%|███████▉  | 10570/13317 [1:49:15<28:05,  1.63it/s]
 79%|███████▉  | 10571/13317 [1:49:15<28:06,  1.63it/s]
                                                       
{'loss': 0.123, 'grad_norm': 0.4761323034763336, 'learning_rate': 1.2403149124262176e-06, 'epoch': 2.38}

 79%|███████▉  | 10571/13317 [1:49:15<28:06,  1.63it/s]
 79%|███████▉  | 10572/13317 [1:49:16<28:07,  1.63it/s]
                                                       
{'loss': 0.1626, 'grad_norm': 0.6899129152297974, 'learning_rate': 1.2394510250465047e-06, 'epoch': 2.38}

 79%|███████▉  | 10572/13317 [1:49:16<28:07,  1.63it/s]
 79%|███████▉  | 10573/13317 [1:49:16<28:05,  1.63it/s]
                                                       
{'loss': 0.1358, 'grad_norm': 0.5070132613182068, 'learning_rate': 1.2385873960563227e-06, 'epoch': 2.38}

 79%|███████▉  | 10573/13317 [1:49:16<28:05,  1.63it/s]
 79%|███████▉  | 10574/13317 [1:49:17<28:06,  1.63it/s]
                                                       
{'loss': 0.1271, 'grad_norm': 0.54564368724823, 'learning_rate': 1.237724025515014e-06, 'epoch': 2.38}

 79%|███████▉  | 10574/13317 [1:49:17<28:06,  1.63it/s]
 79%|███████▉  | 10575/13317 [1:49:18<28:05,  1.63it/s]
                                                       
{'loss': 0.1206, 'grad_norm': 0.4926641881465912, 'learning_rate': 1.236860913481901e-06, 'epoch': 2.38}

 79%|███████▉  | 10575/13317 [1:49:18<28:05,  1.63it/s]
 79%|███████▉  | 10576/13317 [1:49:18<28:03,  1.63it/s]
                                                       
{'loss': 0.1582, 'grad_norm': 0.6916219592094421, 'learning_rate': 1.2359980600162858e-06, 'epoch': 2.38}

 79%|███████▉  | 10576/13317 [1:49:18<28:03,  1.63it/s]
 79%|███████▉  | 10577/13317 [1:49:19<28:04,  1.63it/s]
                                                       
{'loss': 0.1666, 'grad_norm': 0.6198058724403381, 'learning_rate': 1.2351354651774583e-06, 'epoch': 2.38}

 79%|███████▉  | 10577/13317 [1:49:19<28:04,  1.63it/s]
 79%|███████▉  | 10578/13317 [1:49:19<28:03,  1.63it/s]
                                                       
{'loss': 0.149, 'grad_norm': 0.5648500919342041, 'learning_rate': 1.2342731290246885e-06, 'epoch': 2.38}

 79%|███████▉  | 10578/13317 [1:49:19<28:03,  1.63it/s]
 79%|███████▉  | 10579/13317 [1:49:20<28:01,  1.63it/s]
                                                       
{'loss': 0.1614, 'grad_norm': 0.43698975443840027, 'learning_rate': 1.233411051617227e-06, 'epoch': 2.38}

 79%|███████▉  | 10579/13317 [1:49:20<28:01,  1.63it/s]
 79%|███████▉  | 10580/13317 [1:49:21<28:00,  1.63it/s]
                                                       
{'loss': 0.1259, 'grad_norm': 0.48426613211631775, 'learning_rate': 1.2325492330143062e-06, 'epoch': 2.38}

 79%|███████▉  | 10580/13317 [1:49:21<28:00,  1.63it/s]
 79%|███████▉  | 10581/13317 [1:49:21<27:59,  1.63it/s]
                                                       
{'loss': 0.1762, 'grad_norm': 0.5847159028053284, 'learning_rate': 1.2316876732751447e-06, 'epoch': 2.38}

 79%|███████▉  | 10581/13317 [1:49:21<27:59,  1.63it/s]
 79%|███████▉  | 10582/13317 [1:49:22<27:59,  1.63it/s]
                                                       
{'loss': 0.1482, 'grad_norm': 0.5548790693283081, 'learning_rate': 1.2308263724589397e-06, 'epoch': 2.38}

 79%|███████▉  | 10582/13317 [1:49:22<27:59,  1.63it/s]
 79%|███████▉  | 10583/13317 [1:49:22<27:59,  1.63it/s]
                                                       
{'loss': 0.1228, 'grad_norm': 0.49849122762680054, 'learning_rate': 1.2299653306248705e-06, 'epoch': 2.38}

 79%|███████▉  | 10583/13317 [1:49:23<27:59,  1.63it/s]
 79%|███████▉  | 10584/13317 [1:49:23<27:59,  1.63it/s]
                                                       
{'loss': 0.1687, 'grad_norm': 0.6751514077186584, 'learning_rate': 1.2291045478321e-06, 'epoch': 2.38}

 79%|███████▉  | 10584/13317 [1:49:23<27:59,  1.63it/s]
 79%|███████▉  | 10585/13317 [1:49:24<27:57,  1.63it/s]
                                                       
{'loss': 0.1344, 'grad_norm': 0.4643966555595398, 'learning_rate': 1.2282440241397763e-06, 'epoch': 2.38}

 79%|███████▉  | 10585/13317 [1:49:24<27:57,  1.63it/s]
 79%|███████▉  | 10586/13317 [1:49:24<27:57,  1.63it/s]
                                                       
{'loss': 0.1721, 'grad_norm': 0.6736175417900085, 'learning_rate': 1.2273837596070238e-06, 'epoch': 2.38}

 79%|███████▉  | 10586/13317 [1:49:24<27:57,  1.63it/s]
 79%|███████▉  | 10587/13317 [1:49:25<27:58,  1.63it/s]
                                                       
{'loss': 0.1138, 'grad_norm': 0.519016683101654, 'learning_rate': 1.2265237542929504e-06, 'epoch': 2.38}

 79%|███████▉  | 10587/13317 [1:49:25<27:58,  1.63it/s]
 80%|███████▉  | 10588/13317 [1:49:26<27:56,  1.63it/s]
                                                       
{'loss': 0.1481, 'grad_norm': 0.5373290777206421, 'learning_rate': 1.2256640082566513e-06, 'epoch': 2.39}

 80%|███████▉  | 10588/13317 [1:49:26<27:56,  1.63it/s]
 80%|███████▉  | 10589/13317 [1:49:26<27:55,  1.63it/s]
                                                       
{'loss': 0.1581, 'grad_norm': 0.5912209749221802, 'learning_rate': 1.2248045215571974e-06, 'epoch': 2.39}

 80%|███████▉  | 10589/13317 [1:49:26<27:55,  1.63it/s]
 80%|███████▉  | 10590/13317 [1:49:27<27:54,  1.63it/s]
                                                       
{'loss': 0.1744, 'grad_norm': 0.7017930150032043, 'learning_rate': 1.2239452942536445e-06, 'epoch': 2.39}

 80%|███████▉  | 10590/13317 [1:49:27<27:54,  1.63it/s]
 80%|███████▉  | 10591/13317 [1:49:27<27:53,  1.63it/s]
                                                       
{'loss': 0.2215, 'grad_norm': 0.647148847579956, 'learning_rate': 1.2230863264050308e-06, 'epoch': 2.39}

 80%|███████▉  | 10591/13317 [1:49:27<27:53,  1.63it/s]
 80%|███████▉  | 10592/13317 [1:49:28<27:55,  1.63it/s]
                                                       
{'loss': 0.1907, 'grad_norm': 0.6545592546463013, 'learning_rate': 1.222227618070379e-06, 'epoch': 2.39}

 80%|███████▉  | 10592/13317 [1:49:28<27:55,  1.63it/s]
 80%|███████▉  | 10593/13317 [1:49:29<27:54,  1.63it/s]
                                                       
{'loss': 0.182, 'grad_norm': 0.5842074751853943, 'learning_rate': 1.2213691693086892e-06, 'epoch': 2.39}

 80%|███████▉  | 10593/13317 [1:49:29<27:54,  1.63it/s]
 80%|███████▉  | 10594/13317 [1:49:29<27:52,  1.63it/s]
                                                       
{'loss': 0.1497, 'grad_norm': 0.6045913100242615, 'learning_rate': 1.220510980178945e-06, 'epoch': 2.39}

 80%|███████▉  | 10594/13317 [1:49:29<27:52,  1.63it/s]
 80%|███████▉  | 10595/13317 [1:49:30<27:51,  1.63it/s]
                                                       
{'loss': 0.1459, 'grad_norm': 0.5668658018112183, 'learning_rate': 1.2196530507401144e-06, 'epoch': 2.39}

 80%|███████▉  | 10595/13317 [1:49:30<27:51,  1.63it/s]
 80%|███████▉  | 10596/13317 [1:49:30<27:50,  1.63it/s]
                                                       
{'loss': 0.1153, 'grad_norm': 0.4758627414703369, 'learning_rate': 1.2187953810511489e-06, 'epoch': 2.39}

 80%|███████▉  | 10596/13317 [1:49:31<27:50,  1.63it/s]
 80%|███████▉  | 10597/13317 [1:49:31<27:52,  1.63it/s]
                                                       
{'loss': 0.1482, 'grad_norm': 0.5732858777046204, 'learning_rate': 1.2179379711709738e-06, 'epoch': 2.39}

 80%|███████▉  | 10597/13317 [1:49:31<27:52,  1.63it/s]
 80%|███████▉  | 10598/13317 [1:49:32<27:51,  1.63it/s]
                                                       
{'loss': 0.1803, 'grad_norm': 0.6079492568969727, 'learning_rate': 1.2170808211585056e-06, 'epoch': 2.39}

 80%|███████▉  | 10598/13317 [1:49:32<27:51,  1.63it/s]
 80%|███████▉  | 10599/13317 [1:49:32<27:49,  1.63it/s]
                                                       
{'loss': 0.1565, 'grad_norm': 0.571689248085022, 'learning_rate': 1.2162239310726403e-06, 'epoch': 2.39}

 80%|███████▉  | 10599/13317 [1:49:32<27:49,  1.63it/s]
 80%|███████▉  | 10600/13317 [1:49:33<27:49,  1.63it/s]
                                                       
{'loss': 0.1487, 'grad_norm': 0.5876469612121582, 'learning_rate': 1.2153673009722539e-06, 'epoch': 2.39}

 80%|███████▉  | 10600/13317 [1:49:33<27:49,  1.63it/s]
 80%|███████▉  | 10601/13317 [1:49:33<27:47,  1.63it/s]
                                                       
{'loss': 0.1476, 'grad_norm': 0.5590060949325562, 'learning_rate': 1.214510930916205e-06, 'epoch': 2.39}

 80%|███████▉  | 10601/13317 [1:49:34<27:47,  1.63it/s]
 80%|███████▉  | 10602/13317 [1:49:34<27:49,  1.63it/s]
                                                       
{'loss': 0.1443, 'grad_norm': 0.5228622555732727, 'learning_rate': 1.2136548209633385e-06, 'epoch': 2.39}

 80%|███████▉  | 10602/13317 [1:49:34<27:49,  1.63it/s]
 80%|███████▉  | 10603/13317 [1:49:35<27:47,  1.63it/s]
                                                       
{'loss': 0.2015, 'grad_norm': 0.5858786702156067, 'learning_rate': 1.2127989711724758e-06, 'epoch': 2.39}

 80%|███████▉  | 10603/13317 [1:49:35<27:47,  1.63it/s]
 80%|███████▉  | 10604/13317 [1:49:35<27:47,  1.63it/s]
                                                       
{'loss': 0.1818, 'grad_norm': 0.5145465731620789, 'learning_rate': 1.2119433816024222e-06, 'epoch': 2.39}

 80%|███████▉  | 10604/13317 [1:49:35<27:47,  1.63it/s]
 80%|███████▉  | 10605/13317 [1:49:36<27:46,  1.63it/s]
                                                       
{'loss': 0.1198, 'grad_norm': 0.5012165904045105, 'learning_rate': 1.2110880523119667e-06, 'epoch': 2.39}

 80%|███████▉  | 10605/13317 [1:49:36<27:46,  1.63it/s]
 80%|███████▉  | 10606/13317 [1:49:37<27:44,  1.63it/s]
                                                       
{'loss': 0.1219, 'grad_norm': 0.4904228150844574, 'learning_rate': 1.2102329833598808e-06, 'epoch': 2.39}

 80%|███████▉  | 10606/13317 [1:49:37<27:44,  1.63it/s]
 80%|███████▉  | 10607/13317 [1:49:37<27:47,  1.63it/s]
                                                       
{'loss': 0.1994, 'grad_norm': 0.6478523015975952, 'learning_rate': 1.209378174804916e-06, 'epoch': 2.39}

 80%|███████▉  | 10607/13317 [1:49:37<27:47,  1.63it/s]
 80%|███████▉  | 10608/13317 [1:49:38<27:45,  1.63it/s]
                                                       
{'loss': 0.1401, 'grad_norm': 0.5185469388961792, 'learning_rate': 1.2085236267058047e-06, 'epoch': 2.39}

 80%|███████▉  | 10608/13317 [1:49:38<27:45,  1.63it/s]
 80%|███████▉  | 10609/13317 [1:49:38<27:43,  1.63it/s]
                                                       
{'loss': 0.1344, 'grad_norm': 0.5708844661712646, 'learning_rate': 1.2076693391212663e-06, 'epoch': 2.39}

 80%|███████▉  | 10609/13317 [1:49:39<27:43,  1.63it/s]
 80%|███████▉  | 10610/13317 [1:49:39<27:42,  1.63it/s]
                                                       
{'loss': 0.1733, 'grad_norm': 0.5616685748100281, 'learning_rate': 1.2068153121099985e-06, 'epoch': 2.39}

 80%|███████▉  | 10610/13317 [1:49:39<27:42,  1.63it/s]
 80%|███████▉  | 10611/13317 [1:49:40<27:41,  1.63it/s]
                                                       
{'loss': 0.1311, 'grad_norm': 0.5404543876647949, 'learning_rate': 1.20596154573068e-06, 'epoch': 2.39}

 80%|███████▉  | 10611/13317 [1:49:40<27:41,  1.63it/s]
 80%|███████▉  | 10612/13317 [1:49:40<27:42,  1.63it/s]
                                                       
{'loss': 0.1465, 'grad_norm': 0.5155220031738281, 'learning_rate': 1.2051080400419756e-06, 'epoch': 2.39}

 80%|███████▉  | 10612/13317 [1:49:40<27:42,  1.63it/s]
 80%|███████▉  | 10613/13317 [1:49:41<27:40,  1.63it/s]
                                                       
{'loss': 0.1437, 'grad_norm': 0.5870513319969177, 'learning_rate': 1.2042547951025307e-06, 'epoch': 2.39}

 80%|███████▉  | 10613/13317 [1:49:41<27:40,  1.63it/s]
 80%|███████▉  | 10614/13317 [1:49:41<27:39,  1.63it/s]
                                                       
{'loss': 0.2031, 'grad_norm': 0.736530065536499, 'learning_rate': 1.2034018109709716e-06, 'epoch': 2.39}

 80%|███████▉  | 10614/13317 [1:49:42<27:39,  1.63it/s]
 80%|███████▉  | 10615/13317 [1:49:42<27:38,  1.63it/s]
                                                       
{'loss': 0.1537, 'grad_norm': 0.5730637907981873, 'learning_rate': 1.202549087705906e-06, 'epoch': 2.39}

 80%|███████▉  | 10615/13317 [1:49:42<27:38,  1.63it/s]
 80%|███████▉  | 10616/13317 [1:49:43<27:37,  1.63it/s]
                                                       
{'loss': 0.1245, 'grad_norm': 0.4973234534263611, 'learning_rate': 1.2016966253659273e-06, 'epoch': 2.39}

 80%|███████▉  | 10616/13317 [1:49:43<27:37,  1.63it/s]
 80%|███████▉  | 10617/13317 [1:49:43<27:39,  1.63it/s]
                                                       
{'loss': 0.1587, 'grad_norm': 0.5752674341201782, 'learning_rate': 1.2008444240096062e-06, 'epoch': 2.39}

 80%|███████▉  | 10617/13317 [1:49:43<27:39,  1.63it/s]
 80%|███████▉  | 10618/13317 [1:49:44<27:39,  1.63it/s]
                                                       
{'loss': 0.1798, 'grad_norm': 0.6391801238059998, 'learning_rate': 1.1999924836955001e-06, 'epoch': 2.39}

 80%|███████▉  | 10618/13317 [1:49:44<27:39,  1.63it/s]
 80%|███████▉  | 10619/13317 [1:49:45<27:37,  1.63it/s]
                                                       
{'loss': 0.1801, 'grad_norm': 0.6805635690689087, 'learning_rate': 1.1991408044821463e-06, 'epoch': 2.39}

 80%|███████▉  | 10619/13317 [1:49:45<27:37,  1.63it/s]
 80%|███████▉  | 10620/13317 [1:49:45<27:35,  1.63it/s]
                                                       
{'loss': 0.1422, 'grad_norm': 0.4954685866832733, 'learning_rate': 1.1982893864280615e-06, 'epoch': 2.39}

 80%|███████▉  | 10620/13317 [1:49:45<27:35,  1.63it/s]
 80%|███████▉  | 10621/13317 [1:49:46<27:35,  1.63it/s]
                                                       
{'loss': 0.1398, 'grad_norm': 0.504743754863739, 'learning_rate': 1.1974382295917508e-06, 'epoch': 2.39}

 80%|███████▉  | 10621/13317 [1:49:46<27:35,  1.63it/s]
 80%|███████▉  | 10622/13317 [1:49:46<27:36,  1.63it/s]
                                                       
{'loss': 0.1228, 'grad_norm': 0.4954816699028015, 'learning_rate': 1.1965873340316942e-06, 'epoch': 2.39}

 80%|███████▉  | 10622/13317 [1:49:47<27:36,  1.63it/s]
 80%|███████▉  | 10623/13317 [1:49:47<27:34,  1.63it/s]
                                                       
{'loss': 0.1399, 'grad_norm': 0.5541040897369385, 'learning_rate': 1.1957366998063602e-06, 'epoch': 2.39}

 80%|███████▉  | 10623/13317 [1:49:47<27:34,  1.63it/s]
 80%|███████▉  | 10624/13317 [1:49:48<27:34,  1.63it/s]
                                                       
{'loss': 0.1535, 'grad_norm': 0.5730574131011963, 'learning_rate': 1.194886326974194e-06, 'epoch': 2.39}

 80%|███████▉  | 10624/13317 [1:49:48<27:34,  1.63it/s]
 80%|███████▉  | 10625/13317 [1:49:48<27:32,  1.63it/s]
                                                       
{'loss': 0.1752, 'grad_norm': 0.6491177082061768, 'learning_rate': 1.1940362155936275e-06, 'epoch': 2.39}

 80%|███████▉  | 10625/13317 [1:49:48<27:32,  1.63it/s]
 80%|███████▉  | 10626/13317 [1:49:49<27:31,  1.63it/s]
                                                       
{'loss': 0.1419, 'grad_norm': 0.5647819638252258, 'learning_rate': 1.1931863657230714e-06, 'epoch': 2.39}

 80%|███████▉  | 10626/13317 [1:49:49<27:31,  1.63it/s]
 80%|███████▉  | 10627/13317 [1:49:49<27:33,  1.63it/s]
                                                       
{'loss': 0.1588, 'grad_norm': 0.5201312303543091, 'learning_rate': 1.1923367774209177e-06, 'epoch': 2.39}

 80%|███████▉  | 10627/13317 [1:49:50<27:33,  1.63it/s]
 80%|███████▉  | 10628/13317 [1:49:50<27:30,  1.63it/s]
                                                       
{'loss': 0.1566, 'grad_norm': 0.5399367213249207, 'learning_rate': 1.1914874507455448e-06, 'epoch': 2.39}

 80%|███████▉  | 10628/13317 [1:49:50<27:30,  1.63it/s]
 80%|███████▉  | 10629/13317 [1:49:51<27:29,  1.63it/s]
                                                       
{'loss': 0.1377, 'grad_norm': 0.5516841411590576, 'learning_rate': 1.1906383857553078e-06, 'epoch': 2.39}

 80%|███████▉  | 10629/13317 [1:49:51<27:29,  1.63it/s]
 80%|███████▉  | 10630/13317 [1:49:51<27:28,  1.63it/s]
                                                       
{'loss': 0.1401, 'grad_norm': 0.5581779479980469, 'learning_rate': 1.1897895825085498e-06, 'epoch': 2.39}

 80%|███████▉  | 10630/13317 [1:49:51<27:28,  1.63it/s]
 80%|███████▉  | 10631/13317 [1:49:52<27:28,  1.63it/s]
                                                       
{'loss': 0.2357, 'grad_norm': 0.8551298975944519, 'learning_rate': 1.1889410410635887e-06, 'epoch': 2.39}

 80%|███████▉  | 10631/13317 [1:49:52<27:28,  1.63it/s]
 80%|███████▉  | 10632/13317 [1:49:53<27:30,  1.63it/s]
                                                       
{'loss': 0.1596, 'grad_norm': 0.6049717664718628, 'learning_rate': 1.1880927614787318e-06, 'epoch': 2.4}

 80%|███████▉  | 10632/13317 [1:49:53<27:30,  1.63it/s]
 80%|███████▉  | 10633/13317 [1:49:53<27:28,  1.63it/s]
                                                       
{'loss': 0.1239, 'grad_norm': 0.5416741371154785, 'learning_rate': 1.1872447438122631e-06, 'epoch': 2.4}

 80%|███████▉  | 10633/13317 [1:49:53<27:28,  1.63it/s]
 80%|███████▉  | 10634/13317 [1:49:54<27:27,  1.63it/s]
                                                       
{'loss': 0.1246, 'grad_norm': 0.47594165802001953, 'learning_rate': 1.186396988122449e-06, 'epoch': 2.4}

 80%|███████▉  | 10634/13317 [1:49:54<27:27,  1.63it/s]
 80%|███████▉  | 10635/13317 [1:49:54<27:25,  1.63it/s]
                                                       
{'loss': 0.1382, 'grad_norm': 0.5458714365959167, 'learning_rate': 1.1855494944675428e-06, 'epoch': 2.4}

 80%|███████▉  | 10635/13317 [1:49:55<27:25,  1.63it/s]
 80%|███████▉  | 10636/13317 [1:49:55<27:25,  1.63it/s]
                                                       
{'loss': 0.1714, 'grad_norm': 0.6528377532958984, 'learning_rate': 1.1847022629057747e-06, 'epoch': 2.4}

 80%|███████▉  | 10636/13317 [1:49:55<27:25,  1.63it/s]
 80%|███████▉  | 10637/13317 [1:49:56<27:26,  1.63it/s]
                                                       
{'loss': 0.1261, 'grad_norm': 0.4686213433742523, 'learning_rate': 1.1838552934953562e-06, 'epoch': 2.4}

 80%|███████▉  | 10637/13317 [1:49:56<27:26,  1.63it/s]
 80%|███████▉  | 10638/13317 [1:49:56<27:25,  1.63it/s]
                                                       
{'loss': 0.1325, 'grad_norm': 0.5155843496322632, 'learning_rate': 1.1830085862944851e-06, 'epoch': 2.4}

 80%|███████▉  | 10638/13317 [1:49:56<27:25,  1.63it/s]
 80%|███████▉  | 10639/13317 [1:49:57<27:23,  1.63it/s]
                                                       
{'loss': 0.2079, 'grad_norm': 0.7289064526557922, 'learning_rate': 1.1821621413613405e-06, 'epoch': 2.4}

 80%|███████▉  | 10639/13317 [1:49:57<27:23,  1.63it/s]
 80%|███████▉  | 10640/13317 [1:49:57<27:23,  1.63it/s]
                                                       
{'loss': 0.1529, 'grad_norm': 0.5669398307800293, 'learning_rate': 1.1813159587540807e-06, 'epoch': 2.4}

 80%|███████▉  | 10640/13317 [1:49:58<27:23,  1.63it/s]
 80%|███████▉  | 10641/13317 [1:49:58<27:22,  1.63it/s]
                                                       
{'loss': 0.1632, 'grad_norm': 0.6411999464035034, 'learning_rate': 1.180470038530846e-06, 'epoch': 2.4}

 80%|███████▉  | 10641/13317 [1:49:58<27:22,  1.63it/s]
 80%|███████▉  | 10642/13317 [1:49:59<27:23,  1.63it/s]
                                                       
{'loss': 0.1628, 'grad_norm': 0.5684486627578735, 'learning_rate': 1.1796243807497614e-06, 'epoch': 2.4}

 80%|███████▉  | 10642/13317 [1:49:59<27:23,  1.63it/s]
 80%|███████▉  | 10643/13317 [1:49:59<27:22,  1.63it/s]
                                                       
{'loss': 0.1673, 'grad_norm': 0.6239591836929321, 'learning_rate': 1.1787789854689352e-06, 'epoch': 2.4}

 80%|███████▉  | 10643/13317 [1:49:59<27:22,  1.63it/s]
 80%|███████▉  | 10644/13317 [1:50:00<27:21,  1.63it/s]
                                                       
{'loss': 0.1609, 'grad_norm': 0.6560501456260681, 'learning_rate': 1.17793385274645e-06, 'epoch': 2.4}

 80%|███████▉  | 10644/13317 [1:50:00<27:21,  1.63it/s]
 80%|███████▉  | 10645/13317 [1:50:01<28:46,  1.55it/s]
                                                       
{'loss': 0.1339, 'grad_norm': 0.49985504150390625, 'learning_rate': 1.177088982640378e-06, 'epoch': 2.4}

 80%|███████▉  | 10645/13317 [1:50:01<28:46,  1.55it/s]
 80%|███████▉  | 10646/13317 [1:50:01<28:19,  1.57it/s]
                                                       
{'loss': 0.1158, 'grad_norm': 0.4797385632991791, 'learning_rate': 1.1762443752087716e-06, 'epoch': 2.4}

 80%|███████▉  | 10646/13317 [1:50:01<28:19,  1.57it/s]
 80%|███████▉  | 10647/13317 [1:50:02<28:01,  1.59it/s]
                                                       
{'loss': 0.235, 'grad_norm': 0.6245350241661072, 'learning_rate': 1.175400030509663e-06, 'epoch': 2.4}

 80%|███████▉  | 10647/13317 [1:50:02<28:01,  1.59it/s]
 80%|███████▉  | 10648/13317 [1:50:02<27:48,  1.60it/s]
                                                       
{'loss': 0.1343, 'grad_norm': 0.5327560901641846, 'learning_rate': 1.174555948601067e-06, 'epoch': 2.4}

 80%|███████▉  | 10648/13317 [1:50:03<27:48,  1.60it/s]
 80%|███████▉  | 10649/13317 [1:50:03<27:38,  1.61it/s]
                                                       
{'loss': 0.1455, 'grad_norm': 0.5773393511772156, 'learning_rate': 1.1737121295409815e-06, 'epoch': 2.4}

 80%|███████▉  | 10649/13317 [1:50:03<27:38,  1.61it/s]
 80%|███████▉  | 10650/13317 [1:50:04<27:31,  1.61it/s]
                                                       
{'loss': 0.1567, 'grad_norm': 0.5491563081741333, 'learning_rate': 1.172868573387389e-06, 'epoch': 2.4}

 80%|███████▉  | 10650/13317 [1:50:04<27:31,  1.61it/s]
 80%|███████▉  | 10651/13317 [1:50:04<27:26,  1.62it/s]
                                                       
{'loss': 0.1345, 'grad_norm': 0.5086719393730164, 'learning_rate': 1.1720252801982446e-06, 'epoch': 2.4}

 80%|███████▉  | 10651/13317 [1:50:04<27:26,  1.62it/s]
 80%|███████▉  | 10652/13317 [1:50:05<27:24,  1.62it/s]
                                                       
{'loss': 0.1785, 'grad_norm': 0.7343682050704956, 'learning_rate': 1.1711822500314951e-06, 'epoch': 2.4}

 80%|███████▉  | 10652/13317 [1:50:05<27:24,  1.62it/s]
 80%|███████▉  | 10653/13317 [1:50:06<27:20,  1.62it/s]
                                                       
{'loss': 0.1588, 'grad_norm': 0.5977708101272583, 'learning_rate': 1.1703394829450665e-06, 'epoch': 2.4}

 80%|███████▉  | 10653/13317 [1:50:06<27:20,  1.62it/s]
 80%|████████  | 10654/13317 [1:50:06<27:18,  1.63it/s]
                                                       
{'loss': 0.1421, 'grad_norm': 0.6008825898170471, 'learning_rate': 1.1694969789968646e-06, 'epoch': 2.4}

 80%|████████  | 10654/13317 [1:50:06<27:18,  1.63it/s]
 80%|████████  | 10655/13317 [1:50:07<27:15,  1.63it/s]
                                                       
{'loss': 0.1422, 'grad_norm': 0.549206554889679, 'learning_rate': 1.1686547382447767e-06, 'epoch': 2.4}

 80%|████████  | 10655/13317 [1:50:07<27:15,  1.63it/s]
 80%|████████  | 10656/13317 [1:50:07<27:13,  1.63it/s]
                                                       
{'loss': 0.1337, 'grad_norm': 0.5243009924888611, 'learning_rate': 1.167812760746675e-06, 'epoch': 2.4}

 80%|████████  | 10656/13317 [1:50:08<27:13,  1.63it/s]
 80%|████████  | 10657/13317 [1:50:08<27:14,  1.63it/s]
                                                       
{'loss': 0.1786, 'grad_norm': 0.6137104034423828, 'learning_rate': 1.1669710465604156e-06, 'epoch': 2.4}

 80%|████████  | 10657/13317 [1:50:08<27:14,  1.63it/s]
 80%|████████  | 10658/13317 [1:50:09<27:13,  1.63it/s]
                                                       
{'loss': 0.1432, 'grad_norm': 0.5880927443504333, 'learning_rate': 1.1661295957438273e-06, 'epoch': 2.4}

 80%|████████  | 10658/13317 [1:50:09<27:13,  1.63it/s]
 80%|████████  | 10659/13317 [1:50:09<27:12,  1.63it/s]
                                                       
{'loss': 0.1597, 'grad_norm': 0.5823058485984802, 'learning_rate': 1.1652884083547295e-06, 'epoch': 2.4}

 80%|████████  | 10659/13317 [1:50:09<27:12,  1.63it/s]
 80%|████████  | 10660/13317 [1:50:10<27:12,  1.63it/s]
                                                       
{'loss': 0.1522, 'grad_norm': 0.6283925771713257, 'learning_rate': 1.164447484450922e-06, 'epoch': 2.4}

 80%|████████  | 10660/13317 [1:50:10<27:12,  1.63it/s]
 80%|████████  | 10661/13317 [1:50:10<27:11,  1.63it/s]
                                                       
{'loss': 0.1341, 'grad_norm': 0.5490059852600098, 'learning_rate': 1.1636068240901831e-06, 'epoch': 2.4}

 80%|████████  | 10661/13317 [1:50:11<27:11,  1.63it/s]
 80%|████████  | 10662/13317 [1:50:11<27:12,  1.63it/s]
                                                       
{'loss': 0.1119, 'grad_norm': 0.4798782765865326, 'learning_rate': 1.1627664273302753e-06, 'epoch': 2.4}

 80%|████████  | 10662/13317 [1:50:11<27:12,  1.63it/s]
 80%|████████  | 10663/13317 [1:50:12<27:11,  1.63it/s]
                                                       
{'loss': 0.1172, 'grad_norm': 0.4866042733192444, 'learning_rate': 1.1619262942289438e-06, 'epoch': 2.4}

 80%|████████  | 10663/13317 [1:50:12<27:11,  1.63it/s]
 80%|████████  | 10664/13317 [1:50:12<27:10,  1.63it/s]
                                                       
{'loss': 0.1488, 'grad_norm': 0.5590972304344177, 'learning_rate': 1.1610864248439136e-06, 'epoch': 2.4}

 80%|████████  | 10664/13317 [1:50:12<27:10,  1.63it/s]
 80%|████████  | 10665/13317 [1:50:13<27:09,  1.63it/s]
                                                       
{'loss': 0.1444, 'grad_norm': 0.6018319129943848, 'learning_rate': 1.1602468192328936e-06, 'epoch': 2.4}

 80%|████████  | 10665/13317 [1:50:13<27:09,  1.63it/s]
 80%|████████  | 10666/13317 [1:50:14<27:08,  1.63it/s]
                                                       
{'loss': 0.13, 'grad_norm': 0.5252519845962524, 'learning_rate': 1.1594074774535724e-06, 'epoch': 2.4}

 80%|████████  | 10666/13317 [1:50:14<27:08,  1.63it/s]
 80%|████████  | 10667/13317 [1:50:14<27:06,  1.63it/s]
                                                       
{'loss': 0.1527, 'grad_norm': 0.5793480277061462, 'learning_rate': 1.1585683995636232e-06, 'epoch': 2.4}

 80%|████████  | 10667/13317 [1:50:14<27:06,  1.63it/s]
 80%|████████  | 10668/13317 [1:50:15<27:05,  1.63it/s]
                                                       
{'loss': 0.1303, 'grad_norm': 0.49126553535461426, 'learning_rate': 1.1577295856206994e-06, 'epoch': 2.4}

 80%|████████  | 10668/13317 [1:50:15<27:05,  1.63it/s]
 80%|████████  | 10669/13317 [1:50:15<27:04,  1.63it/s]
                                                       
{'loss': 0.1586, 'grad_norm': 0.6418523192405701, 'learning_rate': 1.1568910356824336e-06, 'epoch': 2.4}

 80%|████████  | 10669/13317 [1:50:15<27:04,  1.63it/s]
 80%|████████  | 10670/13317 [1:50:16<27:05,  1.63it/s]
                                                       
{'loss': 0.1581, 'grad_norm': 0.5981714129447937, 'learning_rate': 1.1560527498064466e-06, 'epoch': 2.4}

 80%|████████  | 10670/13317 [1:50:16<27:05,  1.63it/s]
 80%|████████  | 10671/13317 [1:50:17<27:04,  1.63it/s]
                                                       
{'loss': 0.181, 'grad_norm': 0.9720470309257507, 'learning_rate': 1.1552147280503345e-06, 'epoch': 2.4}

 80%|████████  | 10671/13317 [1:50:17<27:04,  1.63it/s]
 80%|████████  | 10672/13317 [1:50:17<27:03,  1.63it/s]
                                                       
{'loss': 0.1518, 'grad_norm': 0.516390860080719, 'learning_rate': 1.1543769704716817e-06, 'epoch': 2.4}

 80%|████████  | 10672/13317 [1:50:17<27:03,  1.63it/s]
 80%|████████  | 10673/13317 [1:50:18<27:02,  1.63it/s]
                                                       
{'loss': 0.1196, 'grad_norm': 0.4808087944984436, 'learning_rate': 1.1535394771280479e-06, 'epoch': 2.4}

 80%|████████  | 10673/13317 [1:50:18<27:02,  1.63it/s]
 80%|████████  | 10674/13317 [1:50:18<27:02,  1.63it/s]
                                                       
{'loss': 0.1402, 'grad_norm': 0.5076280832290649, 'learning_rate': 1.1527022480769801e-06, 'epoch': 2.4}

 80%|████████  | 10674/13317 [1:50:19<27:02,  1.63it/s]
 80%|████████  | 10675/13317 [1:50:19<27:02,  1.63it/s]
                                                       
{'loss': 0.1387, 'grad_norm': 0.5710874795913696, 'learning_rate': 1.1518652833760035e-06, 'epoch': 2.4}

 80%|████████  | 10675/13317 [1:50:19<27:02,  1.63it/s]
 80%|████████  | 10676/13317 [1:50:20<27:01,  1.63it/s]
                                                       
{'loss': 0.1409, 'grad_norm': 0.5612186789512634, 'learning_rate': 1.1510285830826252e-06, 'epoch': 2.41}

 80%|████████  | 10676/13317 [1:50:20<27:01,  1.63it/s]
 80%|████████  | 10677/13317 [1:50:20<26:59,  1.63it/s]
                                                       
{'loss': 0.1633, 'grad_norm': 0.5944460034370422, 'learning_rate': 1.1501921472543382e-06, 'epoch': 2.41}

 80%|████████  | 10677/13317 [1:50:20<26:59,  1.63it/s]
 80%|████████  | 10678/13317 [1:50:21<26:59,  1.63it/s]
                                                       
{'loss': 0.1471, 'grad_norm': 0.5420814752578735, 'learning_rate': 1.1493559759486123e-06, 'epoch': 2.41}

 80%|████████  | 10678/13317 [1:50:21<26:59,  1.63it/s]
 80%|████████  | 10679/13317 [1:50:21<26:59,  1.63it/s]
                                                       
{'loss': 0.1677, 'grad_norm': 0.5391279458999634, 'learning_rate': 1.1485200692229026e-06, 'epoch': 2.41}

 80%|████████  | 10679/13317 [1:50:22<26:59,  1.63it/s]
 80%|████████  | 10680/13317 [1:50:22<27:01,  1.63it/s]
                                                       
{'loss': 0.1501, 'grad_norm': 0.537607729434967, 'learning_rate': 1.147684427134645e-06, 'epoch': 2.41}

 80%|████████  | 10680/13317 [1:50:22<27:01,  1.63it/s]
 80%|████████  | 10681/13317 [1:50:23<26:59,  1.63it/s]
                                                       
{'loss': 0.1474, 'grad_norm': 0.5973088145256042, 'learning_rate': 1.146849049741255e-06, 'epoch': 2.41}

 80%|████████  | 10681/13317 [1:50:23<26:59,  1.63it/s]
 80%|████████  | 10682/13317 [1:50:23<26:58,  1.63it/s]
                                                       
{'loss': 0.1759, 'grad_norm': 0.6252503991127014, 'learning_rate': 1.1460139371001339e-06, 'epoch': 2.41}

 80%|████████  | 10682/13317 [1:50:23<26:58,  1.63it/s]
 80%|████████  | 10683/13317 [1:50:24<26:58,  1.63it/s]
                                                       
{'loss': 0.1161, 'grad_norm': 0.4462050795555115, 'learning_rate': 1.1451790892686603e-06, 'epoch': 2.41}

 80%|████████  | 10683/13317 [1:50:24<26:58,  1.63it/s]
 80%|████████  | 10684/13317 [1:50:25<26:56,  1.63it/s]
                                                       
{'loss': 0.1941, 'grad_norm': 0.669657289981842, 'learning_rate': 1.1443445063042007e-06, 'epoch': 2.41}

 80%|████████  | 10684/13317 [1:50:25<26:56,  1.63it/s]
 80%|████████  | 10685/13317 [1:50:25<26:55,  1.63it/s]
                                                       
{'loss': 0.1645, 'grad_norm': 0.6271459460258484, 'learning_rate': 1.1435101882640964e-06, 'epoch': 2.41}

 80%|████████  | 10685/13317 [1:50:25<26:55,  1.63it/s]
 80%|████████  | 10686/13317 [1:50:26<26:54,  1.63it/s]
                                                       
{'loss': 0.1436, 'grad_norm': 0.5147145986557007, 'learning_rate': 1.142676135205677e-06, 'epoch': 2.41}

 80%|████████  | 10686/13317 [1:50:26<26:54,  1.63it/s]
 80%|████████  | 10687/13317 [1:50:26<26:53,  1.63it/s]
                                                       
{'loss': 0.1292, 'grad_norm': 0.463798463344574, 'learning_rate': 1.1418423471862488e-06, 'epoch': 2.41}

 80%|████████  | 10687/13317 [1:50:27<26:53,  1.63it/s]
 80%|████████  | 10688/13317 [1:50:27<26:53,  1.63it/s]
                                                       
{'loss': 0.1622, 'grad_norm': 0.5826510787010193, 'learning_rate': 1.1410088242631007e-06, 'epoch': 2.41}

 80%|████████  | 10688/13317 [1:50:27<26:53,  1.63it/s]
 80%|████████  | 10689/13317 [1:50:28<26:52,  1.63it/s]
                                                       
{'loss': 0.1829, 'grad_norm': 0.7301505208015442, 'learning_rate': 1.1401755664935076e-06, 'epoch': 2.41}

 80%|████████  | 10689/13317 [1:50:28<26:52,  1.63it/s]
 80%|████████  | 10690/13317 [1:50:28<26:52,  1.63it/s]
                                                       
{'loss': 0.1451, 'grad_norm': 0.5128793120384216, 'learning_rate': 1.1393425739347214e-06, 'epoch': 2.41}

 80%|████████  | 10690/13317 [1:50:28<26:52,  1.63it/s]
 80%|████████  | 10691/13317 [1:50:29<26:51,  1.63it/s]
                                                       
{'loss': 0.1393, 'grad_norm': 0.5357117652893066, 'learning_rate': 1.1385098466439764e-06, 'epoch': 2.41}

 80%|████████  | 10691/13317 [1:50:29<26:51,  1.63it/s]
 80%|████████  | 10692/13317 [1:50:29<26:50,  1.63it/s]
                                                       
{'loss': 0.1473, 'grad_norm': 0.5467501282691956, 'learning_rate': 1.137677384678491e-06, 'epoch': 2.41}

 80%|████████  | 10692/13317 [1:50:30<26:50,  1.63it/s]
 80%|████████  | 10693/13317 [1:50:30<26:50,  1.63it/s]
                                                       
{'loss': 0.1279, 'grad_norm': 0.5165166854858398, 'learning_rate': 1.1368451880954656e-06, 'epoch': 2.41}

 80%|████████  | 10693/13317 [1:50:30<26:50,  1.63it/s]
 80%|████████  | 10694/13317 [1:50:31<26:49,  1.63it/s]
                                                       
{'loss': 0.143, 'grad_norm': 0.5340296626091003, 'learning_rate': 1.1360132569520799e-06, 'epoch': 2.41}

 80%|████████  | 10694/13317 [1:50:31<26:49,  1.63it/s]
 80%|████████  | 10695/13317 [1:50:31<26:48,  1.63it/s]
                                                       
{'loss': 0.149, 'grad_norm': 0.553214430809021, 'learning_rate': 1.1351815913054943e-06, 'epoch': 2.41}

 80%|████████  | 10695/13317 [1:50:31<26:48,  1.63it/s]
 80%|████████  | 10696/13317 [1:50:32<26:46,  1.63it/s]
                                                       
{'loss': 0.1508, 'grad_norm': 0.538861870765686, 'learning_rate': 1.1343501912128547e-06, 'epoch': 2.41}

 80%|████████  | 10696/13317 [1:50:32<26:46,  1.63it/s]
 80%|████████  | 10697/13317 [1:50:33<26:45,  1.63it/s]
                                                       
{'loss': 0.1283, 'grad_norm': 0.5649163722991943, 'learning_rate': 1.1335190567312898e-06, 'epoch': 2.41}

 80%|████████  | 10697/13317 [1:50:33<26:45,  1.63it/s]
 80%|████████  | 10698/13317 [1:50:33<26:45,  1.63it/s]
                                                       
{'loss': 0.1067, 'grad_norm': 0.4308982789516449, 'learning_rate': 1.1326881879179019e-06, 'epoch': 2.41}

 80%|████████  | 10698/13317 [1:50:33<26:45,  1.63it/s]
 80%|████████  | 10699/13317 [1:50:34<26:45,  1.63it/s]
                                                       
{'loss': 0.1775, 'grad_norm': 0.629307746887207, 'learning_rate': 1.1318575848297831e-06, 'epoch': 2.41}

 80%|████████  | 10699/13317 [1:50:34<26:45,  1.63it/s]
 80%|████████  | 10700/13317 [1:50:34<26:45,  1.63it/s]
                                                       
{'loss': 0.1318, 'grad_norm': 0.4478450417518616, 'learning_rate': 1.1310272475240065e-06, 'epoch': 2.41}

 80%|████████  | 10700/13317 [1:50:35<26:45,  1.63it/s]
 80%|████████  | 10701/13317 [1:50:35<26:44,  1.63it/s]
                                                       
{'loss': 0.1324, 'grad_norm': 0.4735277593135834, 'learning_rate': 1.130197176057623e-06, 'epoch': 2.41}

 80%|████████  | 10701/13317 [1:50:35<26:44,  1.63it/s]
 80%|████████  | 10702/13317 [1:50:36<26:42,  1.63it/s]
                                                       
{'loss': 0.1233, 'grad_norm': 0.467390239238739, 'learning_rate': 1.1293673704876662e-06, 'epoch': 2.41}

 80%|████████  | 10702/13317 [1:50:36<26:42,  1.63it/s]
 80%|████████  | 10703/13317 [1:50:36<26:44,  1.63it/s]
                                                       
{'loss': 0.1472, 'grad_norm': 0.5678412914276123, 'learning_rate': 1.1285378308711542e-06, 'epoch': 2.41}

 80%|████████  | 10703/13317 [1:50:36<26:44,  1.63it/s]
 80%|████████  | 10704/13317 [1:50:37<26:44,  1.63it/s]
                                                       
{'loss': 0.1359, 'grad_norm': 0.4896989166736603, 'learning_rate': 1.127708557265088e-06, 'epoch': 2.41}

 80%|████████  | 10704/13317 [1:50:37<26:44,  1.63it/s]
 80%|████████  | 10705/13317 [1:50:37<26:43,  1.63it/s]
                                                       
{'loss': 0.1563, 'grad_norm': 0.6600270867347717, 'learning_rate': 1.1268795497264413e-06, 'epoch': 2.41}

 80%|████████  | 10705/13317 [1:50:38<26:43,  1.63it/s]
 80%|████████  | 10706/13317 [1:50:38<26:42,  1.63it/s]
                                                       
{'loss': 0.1419, 'grad_norm': 0.5479336380958557, 'learning_rate': 1.1260508083121784e-06, 'epoch': 2.41}

 80%|████████  | 10706/13317 [1:50:38<26:42,  1.63it/s]
 80%|████████  | 10707/13317 [1:50:39<26:42,  1.63it/s]
                                                       
{'loss': 0.1528, 'grad_norm': 0.5986954569816589, 'learning_rate': 1.125222333079245e-06, 'epoch': 2.41}

 80%|████████  | 10707/13317 [1:50:39<26:42,  1.63it/s]
 80%|████████  | 10708/13317 [1:50:39<26:41,  1.63it/s]
                                                       
{'loss': 0.1446, 'grad_norm': 0.5705296993255615, 'learning_rate': 1.124394124084564e-06, 'epoch': 2.41}

 80%|████████  | 10708/13317 [1:50:39<26:41,  1.63it/s]
 80%|████████  | 10709/13317 [1:50:40<26:40,  1.63it/s]
                                                       
{'loss': 0.1551, 'grad_norm': 0.5491181015968323, 'learning_rate': 1.1235661813850406e-06, 'epoch': 2.41}

 80%|████████  | 10709/13317 [1:50:40<26:40,  1.63it/s]
 80%|████████  | 10710/13317 [1:50:41<26:39,  1.63it/s]
                                                       
{'loss': 0.1253, 'grad_norm': 0.4690971076488495, 'learning_rate': 1.122738505037565e-06, 'epoch': 2.41}

 80%|████████  | 10710/13317 [1:50:41<26:39,  1.63it/s]
 80%|████████  | 10711/13317 [1:50:41<26:38,  1.63it/s]
                                                       
{'loss': 0.1087, 'grad_norm': 0.4798584282398224, 'learning_rate': 1.1219110950990102e-06, 'epoch': 2.41}

 80%|████████  | 10711/13317 [1:50:41<26:38,  1.63it/s]
 80%|████████  | 10712/13317 [1:50:42<26:36,  1.63it/s]
                                                       
{'loss': 0.1546, 'grad_norm': 0.5975042581558228, 'learning_rate': 1.1210839516262223e-06, 'epoch': 2.41}

 80%|████████  | 10712/13317 [1:50:42<26:36,  1.63it/s]
 80%|████████  | 10713/13317 [1:50:42<26:36,  1.63it/s]
                                                       
{'loss': 0.1697, 'grad_norm': 0.5856941938400269, 'learning_rate': 1.1202570746760371e-06, 'epoch': 2.41}

 80%|████████  | 10713/13317 [1:50:42<26:36,  1.63it/s]
 80%|████████  | 10714/13317 [1:50:43<26:36,  1.63it/s]
                                                       
{'loss': 0.1376, 'grad_norm': 0.5288708806037903, 'learning_rate': 1.1194304643052717e-06, 'epoch': 2.41}

 80%|████████  | 10714/13317 [1:50:43<26:36,  1.63it/s]
 80%|████████  | 10715/13317 [1:50:44<26:35,  1.63it/s]
                                                       
{'loss': 0.1125, 'grad_norm': 0.4094899594783783, 'learning_rate': 1.118604120570722e-06, 'epoch': 2.41}

 80%|████████  | 10715/13317 [1:50:44<26:35,  1.63it/s]
 80%|████████  | 10716/13317 [1:50:44<26:35,  1.63it/s]
                                                       
{'loss': 0.1662, 'grad_norm': 0.7627487778663635, 'learning_rate': 1.1177780435291641e-06, 'epoch': 2.41}

 80%|████████  | 10716/13317 [1:50:44<26:35,  1.63it/s]
 80%|████████  | 10717/13317 [1:50:45<26:35,  1.63it/s]
                                                       
{'loss': 0.13, 'grad_norm': 0.5016361474990845, 'learning_rate': 1.1169522332373606e-06, 'epoch': 2.41}

 80%|████████  | 10717/13317 [1:50:45<26:35,  1.63it/s]
 80%|████████  | 10718/13317 [1:50:45<26:33,  1.63it/s]
                                                       
{'loss': 0.1434, 'grad_norm': 0.5186248421669006, 'learning_rate': 1.1161266897520544e-06, 'epoch': 2.41}

 80%|████████  | 10718/13317 [1:50:46<26:33,  1.63it/s]
 80%|████████  | 10719/13317 [1:50:46<26:34,  1.63it/s]
                                                       
{'loss': 0.1385, 'grad_norm': 0.5308839678764343, 'learning_rate': 1.1153014131299678e-06, 'epoch': 2.41}

 80%|████████  | 10719/13317 [1:50:46<26:34,  1.63it/s]
 80%|████████  | 10720/13317 [1:50:47<26:33,  1.63it/s]
                                                       
{'loss': 0.1371, 'grad_norm': 0.537751317024231, 'learning_rate': 1.1144764034278043e-06, 'epoch': 2.41}

 80%|████████  | 10720/13317 [1:50:47<26:33,  1.63it/s]
 81%|████████  | 10721/13317 [1:50:47<26:32,  1.63it/s]
                                                       
{'loss': 0.1399, 'grad_norm': 0.5117639899253845, 'learning_rate': 1.113651660702254e-06, 'epoch': 2.42}

 81%|████████  | 10721/13317 [1:50:47<26:32,  1.63it/s]
 81%|████████  | 10722/13317 [1:50:48<26:30,  1.63it/s]
                                                       
{'loss': 0.1379, 'grad_norm': 0.5586310029029846, 'learning_rate': 1.1128271850099842e-06, 'epoch': 2.42}

 81%|████████  | 10722/13317 [1:50:48<26:30,  1.63it/s]
 81%|████████  | 10723/13317 [1:50:48<26:31,  1.63it/s]
                                                       
{'loss': 0.1739, 'grad_norm': 0.6158810257911682, 'learning_rate': 1.112002976407643e-06, 'epoch': 2.42}

 81%|████████  | 10723/13317 [1:50:49<26:31,  1.63it/s]
 81%|████████  | 10724/13317 [1:50:49<26:29,  1.63it/s]
                                                       
{'loss': 0.158, 'grad_norm': 0.5742827653884888, 'learning_rate': 1.1111790349518658e-06, 'epoch': 2.42}

 81%|████████  | 10724/13317 [1:50:49<26:29,  1.63it/s]
 81%|████████  | 10725/13317 [1:50:50<26:30,  1.63it/s]
                                                       
{'loss': 0.1749, 'grad_norm': 0.6062498688697815, 'learning_rate': 1.1103553606992628e-06, 'epoch': 2.42}

 81%|████████  | 10725/13317 [1:50:50<26:30,  1.63it/s]
 81%|████████  | 10726/13317 [1:50:50<26:29,  1.63it/s]
                                                       
{'loss': 0.13, 'grad_norm': 0.5410491824150085, 'learning_rate': 1.109531953706432e-06, 'epoch': 2.42}

 81%|████████  | 10726/13317 [1:50:50<26:29,  1.63it/s]
 81%|████████  | 10727/13317 [1:50:51<26:28,  1.63it/s]
                                                       
{'loss': 0.1374, 'grad_norm': 0.4843725264072418, 'learning_rate': 1.108708814029948e-06, 'epoch': 2.42}

 81%|████████  | 10727/13317 [1:50:51<26:28,  1.63it/s]
 81%|████████  | 10728/13317 [1:50:52<26:27,  1.63it/s]
                                                       
{'loss': 0.1514, 'grad_norm': 0.5420644879341125, 'learning_rate': 1.107885941726371e-06, 'epoch': 2.42}

 81%|████████  | 10728/13317 [1:50:52<26:27,  1.63it/s]
 81%|████████  | 10729/13317 [1:50:52<26:26,  1.63it/s]
                                                       
{'loss': 0.1752, 'grad_norm': 0.6985176205635071, 'learning_rate': 1.1070633368522404e-06, 'epoch': 2.42}

 81%|████████  | 10729/13317 [1:50:52<26:26,  1.63it/s]
 81%|████████  | 10730/13317 [1:50:53<26:26,  1.63it/s]
                                                       
{'loss': 0.1509, 'grad_norm': 0.608939528465271, 'learning_rate': 1.1062409994640765e-06, 'epoch': 2.42}

 81%|████████  | 10730/13317 [1:50:53<26:26,  1.63it/s]
 81%|████████  | 10731/13317 [1:50:53<26:25,  1.63it/s]
                                                       
{'loss': 0.1584, 'grad_norm': 0.6335769295692444, 'learning_rate': 1.1054189296183849e-06, 'epoch': 2.42}

 81%|████████  | 10731/13317 [1:50:54<26:25,  1.63it/s]
 81%|████████  | 10732/13317 [1:50:54<26:25,  1.63it/s]
                                                       
{'loss': 0.1744, 'grad_norm': 0.6293816566467285, 'learning_rate': 1.1045971273716476e-06, 'epoch': 2.42}

 81%|████████  | 10732/13317 [1:50:54<26:25,  1.63it/s]
 81%|████████  | 10733/13317 [1:50:55<26:24,  1.63it/s]
                                                       
{'loss': 0.1463, 'grad_norm': 0.5364255309104919, 'learning_rate': 1.1037755927803345e-06, 'epoch': 2.42}

 81%|████████  | 10733/13317 [1:50:55<26:24,  1.63it/s]
 81%|████████  | 10734/13317 [1:50:55<26:24,  1.63it/s]
                                                       
{'loss': 0.138, 'grad_norm': 0.5271705389022827, 'learning_rate': 1.1029543259008908e-06, 'epoch': 2.42}

 81%|████████  | 10734/13317 [1:50:55<26:24,  1.63it/s]
 81%|████████  | 10735/13317 [1:50:56<26:23,  1.63it/s]
                                                       
{'loss': 0.1534, 'grad_norm': 0.582455039024353, 'learning_rate': 1.1021333267897487e-06, 'epoch': 2.42}

 81%|████████  | 10735/13317 [1:50:56<26:23,  1.63it/s]
 81%|████████  | 10736/13317 [1:50:56<26:22,  1.63it/s]
                                                       
{'loss': 0.1235, 'grad_norm': 0.4921170473098755, 'learning_rate': 1.1013125955033188e-06, 'epoch': 2.42}

 81%|████████  | 10736/13317 [1:50:57<26:22,  1.63it/s]
 81%|████████  | 10737/13317 [1:50:57<26:21,  1.63it/s]
                                                       
{'loss': 0.1558, 'grad_norm': 0.6543856263160706, 'learning_rate': 1.1004921320979917e-06, 'epoch': 2.42}

 81%|████████  | 10737/13317 [1:50:57<26:21,  1.63it/s]
 81%|████████  | 10738/13317 [1:50:58<26:22,  1.63it/s]
                                                       
{'loss': 0.1358, 'grad_norm': 0.5359885692596436, 'learning_rate': 1.099671936630145e-06, 'epoch': 2.42}

 81%|████████  | 10738/13317 [1:50:58<26:22,  1.63it/s]
 81%|████████  | 10739/13317 [1:50:58<26:21,  1.63it/s]
                                                       
{'loss': 0.1335, 'grad_norm': 0.5036492347717285, 'learning_rate': 1.0988520091561328e-06, 'epoch': 2.42}

 81%|████████  | 10739/13317 [1:50:58<26:21,  1.63it/s]
 81%|████████  | 10740/13317 [1:50:59<26:20,  1.63it/s]
                                                       
{'loss': 0.1276, 'grad_norm': 0.5252094864845276, 'learning_rate': 1.0980323497322941e-06, 'epoch': 2.42}

 81%|████████  | 10740/13317 [1:50:59<26:20,  1.63it/s]
 81%|████████  | 10741/13317 [1:51:00<26:20,  1.63it/s]
                                                       
{'loss': 0.145, 'grad_norm': 0.5127671957015991, 'learning_rate': 1.0972129584149482e-06, 'epoch': 2.42}

 81%|████████  | 10741/13317 [1:51:00<26:20,  1.63it/s]
 81%|████████  | 10742/13317 [1:51:00<26:19,  1.63it/s]
                                                       
{'loss': 0.1474, 'grad_norm': 0.5696055889129639, 'learning_rate': 1.0963938352603942e-06, 'epoch': 2.42}

 81%|████████  | 10742/13317 [1:51:00<26:19,  1.63it/s]
 81%|████████  | 10743/13317 [1:51:01<26:19,  1.63it/s]
                                                       
{'loss': 0.1399, 'grad_norm': 0.5257719159126282, 'learning_rate': 1.0955749803249155e-06, 'epoch': 2.42}

 81%|████████  | 10743/13317 [1:51:01<26:19,  1.63it/s]
 81%|████████  | 10744/13317 [1:51:01<26:19,  1.63it/s]
                                                       
{'loss': 0.1551, 'grad_norm': 0.5794333219528198, 'learning_rate': 1.0947563936647781e-06, 'epoch': 2.42}

 81%|████████  | 10744/13317 [1:51:02<26:19,  1.63it/s]
 81%|████████  | 10745/13317 [1:51:02<26:18,  1.63it/s]
                                                       
{'loss': 0.1842, 'grad_norm': 0.6500816941261292, 'learning_rate': 1.0939380753362255e-06, 'epoch': 2.42}

 81%|████████  | 10745/13317 [1:51:02<26:18,  1.63it/s]
 81%|████████  | 10746/13317 [1:51:03<26:18,  1.63it/s]
                                                       
{'loss': 0.1403, 'grad_norm': 0.5818924307823181, 'learning_rate': 1.0931200253954838e-06, 'epoch': 2.42}

 81%|████████  | 10746/13317 [1:51:03<26:18,  1.63it/s]
 81%|████████  | 10747/13317 [1:51:03<26:17,  1.63it/s]
                                                       
{'loss': 0.1521, 'grad_norm': 0.5959500670433044, 'learning_rate': 1.0923022438987645e-06, 'epoch': 2.42}

 81%|████████  | 10747/13317 [1:51:03<26:17,  1.63it/s]
 81%|████████  | 10748/13317 [1:51:04<26:16,  1.63it/s]
                                                       
{'loss': 0.1239, 'grad_norm': 0.5921803116798401, 'learning_rate': 1.0914847309022565e-06, 'epoch': 2.42}

 81%|████████  | 10748/13317 [1:51:04<26:16,  1.63it/s]
 81%|████████  | 10749/13317 [1:51:04<26:15,  1.63it/s]
                                                       
{'loss': 0.1707, 'grad_norm': 0.8323166370391846, 'learning_rate': 1.0906674864621297e-06, 'epoch': 2.42}

 81%|████████  | 10749/13317 [1:51:05<26:15,  1.63it/s]
 81%|████████  | 10750/13317 [1:51:05<26:16,  1.63it/s]
                                                       
{'loss': 0.1092, 'grad_norm': 0.448159396648407, 'learning_rate': 1.0898505106345397e-06, 'epoch': 2.42}

 81%|████████  | 10750/13317 [1:51:05<26:16,  1.63it/s]
 81%|████████  | 10751/13317 [1:51:06<26:15,  1.63it/s]
                                                       
{'loss': 0.1263, 'grad_norm': 0.4780685007572174, 'learning_rate': 1.0890338034756238e-06, 'epoch': 2.42}

 81%|████████  | 10751/13317 [1:51:06<26:15,  1.63it/s]
 81%|████████  | 10752/13317 [1:51:06<26:14,  1.63it/s]
                                                       
{'loss': 0.1253, 'grad_norm': 0.5269527435302734, 'learning_rate': 1.0882173650414923e-06, 'epoch': 2.42}

 81%|████████  | 10752/13317 [1:51:06<26:14,  1.63it/s]
 81%|████████  | 10753/13317 [1:51:07<26:13,  1.63it/s]
                                                       
{'loss': 0.1276, 'grad_norm': 0.49330806732177734, 'learning_rate': 1.0874011953882468e-06, 'epoch': 2.42}

 81%|████████  | 10753/13317 [1:51:07<26:13,  1.63it/s]
 81%|████████  | 10754/13317 [1:51:08<26:13,  1.63it/s]
                                                       
{'loss': 0.1786, 'grad_norm': 0.6959000825881958, 'learning_rate': 1.086585294571968e-06, 'epoch': 2.42}

 81%|████████  | 10754/13317 [1:51:08<26:13,  1.63it/s]
 81%|████████  | 10755/13317 [1:51:08<26:11,  1.63it/s]
                                                       
{'loss': 0.1023, 'grad_norm': 0.42702117562294006, 'learning_rate': 1.0857696626487151e-06, 'epoch': 2.42}

 81%|████████  | 10755/13317 [1:51:08<26:11,  1.63it/s]
 81%|████████  | 10756/13317 [1:51:09<26:11,  1.63it/s]
                                                       
{'loss': 0.1487, 'grad_norm': 0.5926663875579834, 'learning_rate': 1.0849542996745293e-06, 'epoch': 2.42}

 81%|████████  | 10756/13317 [1:51:09<26:11,  1.63it/s]
 81%|████████  | 10757/13317 [1:51:09<26:10,  1.63it/s]
                                                       
{'loss': 0.1285, 'grad_norm': 0.5014798045158386, 'learning_rate': 1.0841392057054367e-06, 'epoch': 2.42}

 81%|████████  | 10757/13317 [1:51:09<26:10,  1.63it/s]
 81%|████████  | 10758/13317 [1:51:10<26:10,  1.63it/s]
                                                       
{'loss': 0.1513, 'grad_norm': 0.553532063961029, 'learning_rate': 1.0833243807974453e-06, 'epoch': 2.42}

 81%|████████  | 10758/13317 [1:51:10<26:10,  1.63it/s]
 81%|████████  | 10759/13317 [1:51:11<26:09,  1.63it/s]
                                                       
{'loss': 0.1463, 'grad_norm': 0.5513935685157776, 'learning_rate': 1.0825098250065363e-06, 'epoch': 2.42}

 81%|████████  | 10759/13317 [1:51:11<26:09,  1.63it/s]
 81%|████████  | 10760/13317 [1:51:11<26:09,  1.63it/s]
                                                       
{'loss': 0.1712, 'grad_norm': 0.6317394375801086, 'learning_rate': 1.0816955383886813e-06, 'epoch': 2.42}

 81%|████████  | 10760/13317 [1:51:11<26:09,  1.63it/s]
 81%|████████  | 10761/13317 [1:51:12<26:08,  1.63it/s]
                                                       
{'loss': 0.1386, 'grad_norm': 0.4899536669254303, 'learning_rate': 1.080881520999832e-06, 'epoch': 2.42}

 81%|████████  | 10761/13317 [1:51:12<26:08,  1.63it/s]
 81%|████████  | 10762/13317 [1:51:12<26:08,  1.63it/s]
                                                       
{'loss': 0.171, 'grad_norm': 0.5699849128723145, 'learning_rate': 1.0800677728959186e-06, 'epoch': 2.42}

 81%|████████  | 10762/13317 [1:51:13<26:08,  1.63it/s]
 81%|████████  | 10763/13317 [1:51:13<26:07,  1.63it/s]
                                                       
{'loss': 0.1598, 'grad_norm': 0.5753113627433777, 'learning_rate': 1.0792542941328527e-06, 'epoch': 2.42}

 81%|████████  | 10763/13317 [1:51:13<26:07,  1.63it/s]
 81%|████████  | 10764/13317 [1:51:14<26:07,  1.63it/s]
                                                       
{'loss': 0.1561, 'grad_norm': 0.6031795144081116, 'learning_rate': 1.07844108476653e-06, 'epoch': 2.42}

 81%|████████  | 10764/13317 [1:51:14<26:07,  1.63it/s]
 81%|████████  | 10765/13317 [1:51:14<26:07,  1.63it/s]
                                                       
{'loss': 0.1393, 'grad_norm': 0.5446879267692566, 'learning_rate': 1.0776281448528287e-06, 'epoch': 2.43}

 81%|████████  | 10765/13317 [1:51:14<26:07,  1.63it/s]
 81%|████████  | 10766/13317 [1:51:15<26:05,  1.63it/s]
                                                       
{'loss': 0.1443, 'grad_norm': 0.5148540139198303, 'learning_rate': 1.0768154744476044e-06, 'epoch': 2.43}

 81%|████████  | 10766/13317 [1:51:15<26:05,  1.63it/s]
 81%|████████  | 10767/13317 [1:51:15<26:04,  1.63it/s]
                                                       
{'loss': 0.1332, 'grad_norm': 0.535524845123291, 'learning_rate': 1.0760030736066952e-06, 'epoch': 2.43}

 81%|████████  | 10767/13317 [1:51:16<26:04,  1.63it/s]
 81%|████████  | 10768/13317 [1:51:16<27:24,  1.55it/s]
                                                       
{'loss': 0.1618, 'grad_norm': 0.6590030193328857, 'learning_rate': 1.0751909423859235e-06, 'epoch': 2.43}

 81%|████████  | 10768/13317 [1:51:16<27:24,  1.55it/s]
 81%|████████  | 10769/13317 [1:51:17<26:59,  1.57it/s]
                                                       
{'loss': 0.1554, 'grad_norm': 0.6124571561813354, 'learning_rate': 1.0743790808410914e-06, 'epoch': 2.43}

 81%|████████  | 10769/13317 [1:51:17<26:59,  1.57it/s]
 81%|████████  | 10770/13317 [1:51:17<26:50,  1.58it/s]
                                                       
{'loss': 0.1304, 'grad_norm': 0.501405656337738, 'learning_rate': 1.0735674890279801e-06, 'epoch': 2.43}

 81%|████████  | 10770/13317 [1:51:18<26:50,  1.58it/s]
 81%|████████  | 10771/13317 [1:51:18<26:35,  1.60it/s]
                                                       
{'loss': 0.1165, 'grad_norm': 0.42876967787742615, 'learning_rate': 1.0727561670023556e-06, 'epoch': 2.43}

 81%|████████  | 10771/13317 [1:51:18<26:35,  1.60it/s]
 81%|████████  | 10772/13317 [1:51:19<26:24,  1.61it/s]
                                                       
{'loss': 0.1399, 'grad_norm': 0.5258716940879822, 'learning_rate': 1.071945114819966e-06, 'epoch': 2.43}

 81%|████████  | 10772/13317 [1:51:19<26:24,  1.61it/s]
 81%|████████  | 10773/13317 [1:51:19<26:17,  1.61it/s]
                                                       
{'loss': 0.1249, 'grad_norm': 0.47874996066093445, 'learning_rate': 1.0711343325365386e-06, 'epoch': 2.43}

 81%|████████  | 10773/13317 [1:51:19<26:17,  1.61it/s]
 81%|████████  | 10774/13317 [1:51:20<26:12,  1.62it/s]
                                                       
{'loss': 0.2163, 'grad_norm': 0.6465375423431396, 'learning_rate': 1.0703238202077803e-06, 'epoch': 2.43}

 81%|████████  | 10774/13317 [1:51:20<26:12,  1.62it/s]
 81%|████████  | 10775/13317 [1:51:21<26:10,  1.62it/s]
                                                       
{'loss': 0.2025, 'grad_norm': 0.7449777722358704, 'learning_rate': 1.0695135778893851e-06, 'epoch': 2.43}

 81%|████████  | 10775/13317 [1:51:21<26:10,  1.62it/s]
 81%|████████  | 10776/13317 [1:51:21<26:06,  1.62it/s]
                                                       
{'loss': 0.1662, 'grad_norm': 0.6270003914833069, 'learning_rate': 1.0687036056370237e-06, 'epoch': 2.43}

 81%|████████  | 10776/13317 [1:51:21<26:06,  1.62it/s]
 81%|████████  | 10777/13317 [1:51:22<26:03,  1.62it/s]
                                                       
{'loss': 0.1417, 'grad_norm': 0.5107801556587219, 'learning_rate': 1.0678939035063484e-06, 'epoch': 2.43}

 81%|████████  | 10777/13317 [1:51:22<26:03,  1.62it/s]
 81%|████████  | 10778/13317 [1:51:22<26:01,  1.63it/s]
                                                       
{'loss': 0.1782, 'grad_norm': 0.5976600050926208, 'learning_rate': 1.067084471552996e-06, 'epoch': 2.43}

 81%|████████  | 10778/13317 [1:51:22<26:01,  1.63it/s]
 81%|████████  | 10779/13317 [1:51:23<25:59,  1.63it/s]
                                                       
{'loss': 0.1129, 'grad_norm': 0.5052507519721985, 'learning_rate': 1.066275309832584e-06, 'epoch': 2.43}

 81%|████████  | 10779/13317 [1:51:23<25:59,  1.63it/s]
 81%|████████  | 10780/13317 [1:51:24<27:21,  1.55it/s]
                                                       
{'loss': 0.1385, 'grad_norm': 0.5112053155899048, 'learning_rate': 1.065466418400709e-06, 'epoch': 2.43}

 81%|████████  | 10780/13317 [1:51:24<27:21,  1.55it/s]
 81%|████████  | 10781/13317 [1:51:24<26:55,  1.57it/s]
                                                       
{'loss': 0.1614, 'grad_norm': 0.5786210298538208, 'learning_rate': 1.064657797312949e-06, 'epoch': 2.43}

 81%|████████  | 10781/13317 [1:51:24<26:55,  1.57it/s]
 81%|████████  | 10782/13317 [1:51:25<26:36,  1.59it/s]
                                                       
{'loss': 0.1781, 'grad_norm': 0.6056848168373108, 'learning_rate': 1.0638494466248684e-06, 'epoch': 2.43}

 81%|████████  | 10782/13317 [1:51:25<26:36,  1.59it/s]
 81%|████████  | 10783/13317 [1:51:26<26:22,  1.60it/s]
                                                       
{'loss': 0.1258, 'grad_norm': 0.5541137456893921, 'learning_rate': 1.063041366392007e-06, 'epoch': 2.43}

 81%|████████  | 10783/13317 [1:51:26<26:22,  1.60it/s]
 81%|████████  | 10784/13317 [1:51:26<26:13,  1.61it/s]
                                                       
{'loss': 0.1528, 'grad_norm': 0.5614636540412903, 'learning_rate': 1.0622335566698878e-06, 'epoch': 2.43}

 81%|████████  | 10784/13317 [1:51:26<26:13,  1.61it/s]
 81%|████████  | 10785/13317 [1:51:27<26:09,  1.61it/s]
                                                       
{'loss': 0.14, 'grad_norm': 0.5583823919296265, 'learning_rate': 1.0614260175140179e-06, 'epoch': 2.43}

 81%|████████  | 10785/13317 [1:51:27<26:09,  1.61it/s]
 81%|████████  | 10786/13317 [1:51:27<26:04,  1.62it/s]
                                                       
{'loss': 0.1118, 'grad_norm': 0.4782375991344452, 'learning_rate': 1.060618748979882e-06, 'epoch': 2.43}

 81%|████████  | 10786/13317 [1:51:28<26:04,  1.62it/s]
 81%|████████  | 10787/13317 [1:51:28<26:01,  1.62it/s]
                                                       
{'loss': 0.1607, 'grad_norm': 0.6872501969337463, 'learning_rate': 1.05981175112295e-06, 'epoch': 2.43}

 81%|████████  | 10787/13317 [1:51:28<26:01,  1.62it/s]
 81%|████████  | 10788/13317 [1:51:29<25:59,  1.62it/s]
                                                       
{'loss': 0.1566, 'grad_norm': 0.5567161440849304, 'learning_rate': 1.0590050239986688e-06, 'epoch': 2.43}

 81%|████████  | 10788/13317 [1:51:29<25:59,  1.62it/s]
 81%|████████  | 10789/13317 [1:51:29<25:55,  1.62it/s]
                                                       
{'loss': 0.1454, 'grad_norm': 0.5549672842025757, 'learning_rate': 1.0581985676624717e-06, 'epoch': 2.43}

 81%|████████  | 10789/13317 [1:51:29<25:55,  1.62it/s]
 81%|████████  | 10790/13317 [1:51:30<25:56,  1.62it/s]
                                                       
{'loss': 0.1763, 'grad_norm': 0.6390795707702637, 'learning_rate': 1.0573923821697697e-06, 'epoch': 2.43}

 81%|████████  | 10790/13317 [1:51:30<25:56,  1.62it/s]
 81%|████████  | 10791/13317 [1:51:30<25:54,  1.62it/s]
                                                       
{'loss': 0.2186, 'grad_norm': 0.8539546132087708, 'learning_rate': 1.0565864675759546e-06, 'epoch': 2.43}

 81%|████████  | 10791/13317 [1:51:31<25:54,  1.62it/s]
 81%|████████  | 10792/13317 [1:51:31<25:53,  1.63it/s]
                                                       
{'loss': 0.1509, 'grad_norm': 0.5338110327720642, 'learning_rate': 1.0557808239364037e-06, 'epoch': 2.43}

 81%|████████  | 10792/13317 [1:51:31<25:53,  1.63it/s]
 81%|████████  | 10793/13317 [1:51:32<25:51,  1.63it/s]
                                                       
{'loss': 0.1572, 'grad_norm': 0.5659125447273254, 'learning_rate': 1.0549754513064714e-06, 'epoch': 2.43}

 81%|████████  | 10793/13317 [1:51:32<25:51,  1.63it/s]
 81%|████████  | 10794/13317 [1:51:32<25:49,  1.63it/s]
                                                       
{'loss': 0.1753, 'grad_norm': 0.6494248509407043, 'learning_rate': 1.0541703497414974e-06, 'epoch': 2.43}

 81%|████████  | 10794/13317 [1:51:32<25:49,  1.63it/s]
 81%|████████  | 10795/13317 [1:51:33<25:51,  1.63it/s]
                                                       
{'loss': 0.1543, 'grad_norm': 0.6011121273040771, 'learning_rate': 1.053365519296798e-06, 'epoch': 2.43}

 81%|████████  | 10795/13317 [1:51:33<25:51,  1.63it/s]
 81%|████████  | 10796/13317 [1:51:34<25:49,  1.63it/s]
                                                       
{'loss': 0.1607, 'grad_norm': 0.625028133392334, 'learning_rate': 1.0525609600276765e-06, 'epoch': 2.43}

 81%|████████  | 10796/13317 [1:51:34<25:49,  1.63it/s]
 81%|████████  | 10797/13317 [1:51:34<25:47,  1.63it/s]
                                                       
{'loss': 0.1968, 'grad_norm': 0.654750645160675, 'learning_rate': 1.0517566719894119e-06, 'epoch': 2.43}

 81%|████████  | 10797/13317 [1:51:34<25:47,  1.63it/s]
 81%|████████  | 10798/13317 [1:51:35<25:46,  1.63it/s]
                                                       
{'loss': 0.142, 'grad_norm': 0.574424147605896, 'learning_rate': 1.0509526552372695e-06, 'epoch': 2.43}

 81%|████████  | 10798/13317 [1:51:35<25:46,  1.63it/s]
 81%|████████  | 10799/13317 [1:51:35<25:45,  1.63it/s]
                                                       
{'loss': 0.1465, 'grad_norm': 0.5851528644561768, 'learning_rate': 1.0501489098264932e-06, 'epoch': 2.43}

 81%|████████  | 10799/13317 [1:51:35<25:45,  1.63it/s]
 81%|████████  | 10800/13317 [1:51:36<25:45,  1.63it/s]
                                                       
{'loss': 0.1295, 'grad_norm': 0.5081824660301208, 'learning_rate': 1.0493454358123074e-06, 'epoch': 2.43}

 81%|████████  | 10800/13317 [1:51:36<25:45,  1.63it/s]
 81%|████████  | 10801/13317 [1:51:37<25:44,  1.63it/s]
                                                       
{'loss': 0.1461, 'grad_norm': 0.5390105247497559, 'learning_rate': 1.0485422332499213e-06, 'epoch': 2.43}

 81%|████████  | 10801/13317 [1:51:37<25:44,  1.63it/s]
 81%|████████  | 10802/13317 [1:51:37<25:43,  1.63it/s]
                                                       
{'loss': 0.1605, 'grad_norm': 0.5933002233505249, 'learning_rate': 1.0477393021945226e-06, 'epoch': 2.43}

 81%|████████  | 10802/13317 [1:51:37<25:43,  1.63it/s]
 81%|████████  | 10803/13317 [1:51:38<25:42,  1.63it/s]
                                                       
{'loss': 0.1318, 'grad_norm': 0.49994343519210815, 'learning_rate': 1.04693664270128e-06, 'epoch': 2.43}

 81%|████████  | 10803/13317 [1:51:38<25:42,  1.63it/s]
 81%|████████  | 10804/13317 [1:51:38<25:42,  1.63it/s]
                                                       
{'loss': 0.2006, 'grad_norm': 0.6363340616226196, 'learning_rate': 1.046134254825345e-06, 'epoch': 2.43}

 81%|████████  | 10804/13317 [1:51:39<25:42,  1.63it/s]
 81%|████████  | 10805/13317 [1:51:39<25:42,  1.63it/s]
                                                       
{'loss': 0.167, 'grad_norm': 0.674987256526947, 'learning_rate': 1.045332138621853e-06, 'epoch': 2.43}

 81%|████████  | 10805/13317 [1:51:39<25:42,  1.63it/s]
 81%|████████  | 10806/13317 [1:51:40<25:42,  1.63it/s]
                                                       
{'loss': 0.1505, 'grad_norm': 0.7858486771583557, 'learning_rate': 1.0445302941459157e-06, 'epoch': 2.43}

 81%|████████  | 10806/13317 [1:51:40<25:42,  1.63it/s]
 81%|████████  | 10807/13317 [1:51:40<25:40,  1.63it/s]
                                                       
{'loss': 0.1619, 'grad_norm': 0.6471409797668457, 'learning_rate': 1.0437287214526264e-06, 'epoch': 2.43}

 81%|████████  | 10807/13317 [1:51:40<25:40,  1.63it/s]
 81%|████████  | 10808/13317 [1:51:41<25:39,  1.63it/s]
                                                       
{'loss': 0.135, 'grad_norm': 0.507137656211853, 'learning_rate': 1.0429274205970658e-06, 'epoch': 2.43}

 81%|████████  | 10808/13317 [1:51:41<25:39,  1.63it/s]
 81%|████████  | 10809/13317 [1:51:42<25:39,  1.63it/s]
                                                       
{'loss': 0.2025, 'grad_norm': 0.6776361465454102, 'learning_rate': 1.0421263916342888e-06, 'epoch': 2.44}

 81%|████████  | 10809/13317 [1:51:42<25:39,  1.63it/s]
 81%|████████  | 10810/13317 [1:51:42<25:41,  1.63it/s]
                                                       
{'loss': 0.1323, 'grad_norm': 0.5263006091117859, 'learning_rate': 1.0413256346193345e-06, 'epoch': 2.44}

 81%|████████  | 10810/13317 [1:51:42<25:41,  1.63it/s]
 81%|████████  | 10811/13317 [1:51:43<25:39,  1.63it/s]
                                                       
{'loss': 0.1415, 'grad_norm': 0.595582127571106, 'learning_rate': 1.0405251496072245e-06, 'epoch': 2.44}

 81%|████████  | 10811/13317 [1:51:43<25:39,  1.63it/s]
 81%|████████  | 10812/13317 [1:51:43<25:38,  1.63it/s]
                                                       
{'loss': 0.1282, 'grad_norm': 0.4941456913948059, 'learning_rate': 1.0397249366529622e-06, 'epoch': 2.44}

 81%|████████  | 10812/13317 [1:51:43<25:38,  1.63it/s]
 81%|████████  | 10813/13317 [1:51:44<25:37,  1.63it/s]
                                                       
{'loss': 0.5258, 'grad_norm': 0.6456976532936096, 'learning_rate': 1.0389249958115267e-06, 'epoch': 2.44}

 81%|████████  | 10813/13317 [1:51:44<25:37,  1.63it/s]
 81%|████████  | 10814/13317 [1:51:45<25:35,  1.63it/s]
                                                       
{'loss': 0.2525, 'grad_norm': 0.8125528693199158, 'learning_rate': 1.0381253271378844e-06, 'epoch': 2.44}

 81%|████████  | 10814/13317 [1:51:45<25:35,  1.63it/s]
 81%|████████  | 10815/13317 [1:51:45<25:37,  1.63it/s]
                                                       
{'loss': 0.206, 'grad_norm': 0.8812544345855713, 'learning_rate': 1.037325930686982e-06, 'epoch': 2.44}

 81%|████████  | 10815/13317 [1:51:45<25:37,  1.63it/s]
 81%|████████  | 10816/13317 [1:51:46<25:36,  1.63it/s]
                                                       
{'loss': 0.1408, 'grad_norm': 0.6066811084747314, 'learning_rate': 1.0365268065137462e-06, 'epoch': 2.44}

 81%|████████  | 10816/13317 [1:51:46<25:36,  1.63it/s]
 81%|████████  | 10817/13317 [1:51:46<25:35,  1.63it/s]
                                                       
{'loss': 0.1447, 'grad_norm': 0.5850974321365356, 'learning_rate': 1.0357279546730832e-06, 'epoch': 2.44}

 81%|████████  | 10817/13317 [1:51:47<25:35,  1.63it/s]
 81%|████████  | 10818/13317 [1:51:47<25:34,  1.63it/s]
                                                       
{'loss': 0.1369, 'grad_norm': 0.49728506803512573, 'learning_rate': 1.0349293752198842e-06, 'epoch': 2.44}

 81%|████████  | 10818/13317 [1:51:47<25:34,  1.63it/s]
 81%|████████  | 10819/13317 [1:51:48<25:34,  1.63it/s]
                                                       
{'loss': 0.1447, 'grad_norm': 0.5488625764846802, 'learning_rate': 1.034131068209021e-06, 'epoch': 2.44}

 81%|████████  | 10819/13317 [1:51:48<25:34,  1.63it/s]
 81%|████████  | 10820/13317 [1:51:48<25:34,  1.63it/s]
                                                       
{'loss': 0.1205, 'grad_norm': 0.45729881525039673, 'learning_rate': 1.033333033695345e-06, 'epoch': 2.44}

 81%|████████  | 10820/13317 [1:51:48<25:34,  1.63it/s]
 81%|████████▏ | 10821/13317 [1:51:49<25:33,  1.63it/s]
                                                       
{'loss': 0.1761, 'grad_norm': 0.6629489660263062, 'learning_rate': 1.0325352717336879e-06, 'epoch': 2.44}

 81%|████████▏ | 10821/13317 [1:51:49<25:33,  1.63it/s]
 81%|████████▏ | 10822/13317 [1:51:49<25:32,  1.63it/s]
                                                       
{'loss': 0.178, 'grad_norm': 0.6556427478790283, 'learning_rate': 1.0317377823788654e-06, 'epoch': 2.44}

 81%|████████▏ | 10822/13317 [1:51:50<25:32,  1.63it/s]
 81%|████████▏ | 10823/13317 [1:51:50<25:31,  1.63it/s]
                                                       
{'loss': 0.1561, 'grad_norm': 0.6670910120010376, 'learning_rate': 1.0309405656856774e-06, 'epoch': 2.44}

 81%|████████▏ | 10823/13317 [1:51:50<25:31,  1.63it/s]
 81%|████████▏ | 10824/13317 [1:51:51<25:29,  1.63it/s]
                                                       
{'loss': 0.1135, 'grad_norm': 0.5152954459190369, 'learning_rate': 1.0301436217088944e-06, 'epoch': 2.44}

 81%|████████▏ | 10824/13317 [1:51:51<25:29,  1.63it/s]
 81%|████████▏ | 10825/13317 [1:51:51<25:37,  1.62it/s]
                                                       
{'loss': 0.1749, 'grad_norm': 0.6932466626167297, 'learning_rate': 1.0293469505032783e-06, 'epoch': 2.44}

 81%|████████▏ | 10825/13317 [1:51:51<25:37,  1.62it/s]
 81%|████████▏ | 10826/13317 [1:51:52<25:33,  1.62it/s]
                                                       
{'loss': 0.1636, 'grad_norm': 0.5810678005218506, 'learning_rate': 1.02855055212357e-06, 'epoch': 2.44}

 81%|████████▏ | 10826/13317 [1:51:52<25:33,  1.62it/s]
 81%|████████▏ | 10827/13317 [1:51:53<25:30,  1.63it/s]
                                                       
{'loss': 0.1211, 'grad_norm': 0.5395059585571289, 'learning_rate': 1.0277544266244888e-06, 'epoch': 2.44}

 81%|████████▏ | 10827/13317 [1:51:53<25:30,  1.63it/s]
 81%|████████▏ | 10828/13317 [1:51:53<25:29,  1.63it/s]
                                                       
{'loss': 0.1314, 'grad_norm': 0.5622702836990356, 'learning_rate': 1.0269585740607362e-06, 'epoch': 2.44}

 81%|████████▏ | 10828/13317 [1:51:53<25:29,  1.63it/s]
 81%|████████▏ | 10829/13317 [1:51:54<25:29,  1.63it/s]
                                                       
{'loss': 0.1701, 'grad_norm': 0.5970160961151123, 'learning_rate': 1.0261629944869982e-06, 'epoch': 2.44}

 81%|████████▏ | 10829/13317 [1:51:54<25:29,  1.63it/s]
 81%|████████▏ | 10830/13317 [1:51:54<25:30,  1.62it/s]
                                                       
{'loss': 0.104, 'grad_norm': 0.4254988133907318, 'learning_rate': 1.0253676879579383e-06, 'epoch': 2.44}

 81%|████████▏ | 10830/13317 [1:51:55<25:30,  1.62it/s]
 81%|████████▏ | 10831/13317 [1:51:55<25:29,  1.63it/s]
                                                       
{'loss': 0.1651, 'grad_norm': 0.5972644090652466, 'learning_rate': 1.0245726545282002e-06, 'epoch': 2.44}

 81%|████████▏ | 10831/13317 [1:51:55<25:29,  1.63it/s]
 81%|████████▏ | 10832/13317 [1:51:56<25:27,  1.63it/s]
                                                       
{'loss': 0.1431, 'grad_norm': 0.5952590107917786, 'learning_rate': 1.023777894252414e-06, 'epoch': 2.44}

 81%|████████▏ | 10832/13317 [1:51:56<25:27,  1.63it/s]
 81%|████████▏ | 10833/13317 [1:51:56<25:25,  1.63it/s]
                                                       
{'loss': 0.1303, 'grad_norm': 0.5013935565948486, 'learning_rate': 1.0229834071851886e-06, 'epoch': 2.44}

 81%|████████▏ | 10833/13317 [1:51:56<25:25,  1.63it/s]
 81%|████████▏ | 10834/13317 [1:51:57<25:25,  1.63it/s]
                                                       
{'loss': 0.1441, 'grad_norm': 0.5962483286857605, 'learning_rate': 1.0221891933811123e-06, 'epoch': 2.44}

 81%|████████▏ | 10834/13317 [1:51:57<25:25,  1.63it/s]
 81%|████████▏ | 10835/13317 [1:51:57<25:27,  1.62it/s]
                                                       
{'loss': 0.1377, 'grad_norm': 0.5130948424339294, 'learning_rate': 1.0213952528947552e-06, 'epoch': 2.44}

 81%|████████▏ | 10835/13317 [1:51:58<25:27,  1.62it/s]
 81%|████████▏ | 10836/13317 [1:51:58<25:25,  1.63it/s]
                                                       
{'loss': 0.1394, 'grad_norm': 0.5256103277206421, 'learning_rate': 1.0206015857806713e-06, 'epoch': 2.44}

 81%|████████▏ | 10836/13317 [1:51:58<25:25,  1.63it/s]
 81%|████████▏ | 10837/13317 [1:51:59<25:23,  1.63it/s]
                                                       
{'loss': 0.1553, 'grad_norm': 0.5346702337265015, 'learning_rate': 1.0198081920933928e-06, 'epoch': 2.44}

 81%|████████▏ | 10837/13317 [1:51:59<25:23,  1.63it/s]
 81%|████████▏ | 10838/13317 [1:51:59<25:22,  1.63it/s]
                                                       
{'loss': 0.1373, 'grad_norm': 0.5048219561576843, 'learning_rate': 1.0190150718874337e-06, 'epoch': 2.44}

 81%|████████▏ | 10838/13317 [1:51:59<25:22,  1.63it/s]
 81%|████████▏ | 10839/13317 [1:52:00<25:21,  1.63it/s]
                                                       
{'loss': 0.1504, 'grad_norm': 0.5021230578422546, 'learning_rate': 1.01822222521729e-06, 'epoch': 2.44}

 81%|████████▏ | 10839/13317 [1:52:00<25:21,  1.63it/s]
 81%|████████▏ | 10840/13317 [1:52:01<25:27,  1.62it/s]
                                                       
{'loss': 0.1672, 'grad_norm': 0.7435264587402344, 'learning_rate': 1.017429652137441e-06, 'epoch': 2.44}

 81%|████████▏ | 10840/13317 [1:52:01<25:27,  1.62it/s]
 81%|████████▏ | 10841/13317 [1:52:01<25:24,  1.62it/s]
                                                       
{'loss': 0.1403, 'grad_norm': 0.5284789204597473, 'learning_rate': 1.0166373527023427e-06, 'epoch': 2.44}

 81%|████████▏ | 10841/13317 [1:52:01<25:24,  1.62it/s]
 81%|████████▏ | 10842/13317 [1:52:02<25:21,  1.63it/s]
                                                       
{'loss': 0.1342, 'grad_norm': 0.4952075779438019, 'learning_rate': 1.0158453269664332e-06, 'epoch': 2.44}

 81%|████████▏ | 10842/13317 [1:52:02<25:21,  1.63it/s]
 81%|████████▏ | 10843/13317 [1:52:02<25:20,  1.63it/s]
                                                       
{'loss': 0.1653, 'grad_norm': 0.5979377627372742, 'learning_rate': 1.0150535749841362e-06, 'epoch': 2.44}

 81%|████████▏ | 10843/13317 [1:52:03<25:20,  1.63it/s]
 81%|████████▏ | 10844/13317 [1:52:03<25:18,  1.63it/s]
                                                       
{'loss': 0.1348, 'grad_norm': 0.5350939035415649, 'learning_rate': 1.014262096809851e-06, 'epoch': 2.44}

 81%|████████▏ | 10844/13317 [1:52:03<25:18,  1.63it/s]
 81%|████████▏ | 10845/13317 [1:52:04<25:20,  1.63it/s]
                                                       
{'loss': 0.1565, 'grad_norm': 0.5517596006393433, 'learning_rate': 1.013470892497962e-06, 'epoch': 2.44}

 81%|████████▏ | 10845/13317 [1:52:04<25:20,  1.63it/s]
 81%|████████▏ | 10846/13317 [1:52:04<25:18,  1.63it/s]
                                                       
{'loss': 0.1515, 'grad_norm': 0.6416924595832825, 'learning_rate': 1.0126799621028332e-06, 'epoch': 2.44}

 81%|████████▏ | 10846/13317 [1:52:04<25:18,  1.63it/s]
 81%|████████▏ | 10847/13317 [1:52:05<25:17,  1.63it/s]
                                                       
{'loss': 0.1657, 'grad_norm': 0.6066722273826599, 'learning_rate': 1.0118893056788082e-06, 'epoch': 2.44}

 81%|████████▏ | 10847/13317 [1:52:05<25:17,  1.63it/s]
 81%|████████▏ | 10848/13317 [1:52:05<25:15,  1.63it/s]
                                                       
{'loss': 0.1229, 'grad_norm': 0.5172845125198364, 'learning_rate': 1.011098923280216e-06, 'epoch': 2.44}

 81%|████████▏ | 10848/13317 [1:52:06<25:15,  1.63it/s]
 81%|████████▏ | 10849/13317 [1:52:06<25:14,  1.63it/s]
                                                       
{'loss': 0.1601, 'grad_norm': 0.5438947677612305, 'learning_rate': 1.010308814961361e-06, 'epoch': 2.44}

 81%|████████▏ | 10849/13317 [1:52:06<25:14,  1.63it/s]
 81%|████████▏ | 10850/13317 [1:52:07<25:21,  1.62it/s]
                                                       
{'loss': 0.1641, 'grad_norm': 0.5867887735366821, 'learning_rate': 1.009518980776536e-06, 'epoch': 2.44}

 81%|████████▏ | 10850/13317 [1:52:07<25:21,  1.62it/s]
 81%|████████▏ | 10851/13317 [1:52:07<25:18,  1.62it/s]
                                                       
{'loss': 0.1696, 'grad_norm': 0.6665441989898682, 'learning_rate': 1.008729420780007e-06, 'epoch': 2.44}

 81%|████████▏ | 10851/13317 [1:52:07<25:18,  1.62it/s]
 81%|████████▏ | 10852/13317 [1:52:08<25:15,  1.63it/s]
                                                       
{'loss': 0.1307, 'grad_norm': 0.5658749938011169, 'learning_rate': 1.0079401350260288e-06, 'epoch': 2.44}

 81%|████████▏ | 10852/13317 [1:52:08<25:15,  1.63it/s]
 81%|████████▏ | 10853/13317 [1:52:09<25:13,  1.63it/s]
                                                       
{'loss': 0.1429, 'grad_norm': 0.5515552759170532, 'learning_rate': 1.0071511235688319e-06, 'epoch': 2.44}

 81%|████████▏ | 10853/13317 [1:52:09<25:13,  1.63it/s]
 82%|████████▏ | 10854/13317 [1:52:09<25:12,  1.63it/s]
                                                       
{'loss': 0.1431, 'grad_norm': 0.5543326139450073, 'learning_rate': 1.0063623864626282e-06, 'epoch': 2.45}

 82%|████████▏ | 10854/13317 [1:52:09<25:12,  1.63it/s]
 82%|████████▏ | 10855/13317 [1:52:10<25:11,  1.63it/s]
                                                       
{'loss': 0.1286, 'grad_norm': 0.49206408858299255, 'learning_rate': 1.0055739237616153e-06, 'epoch': 2.45}

 82%|████████▏ | 10855/13317 [1:52:10<25:11,  1.63it/s]
 82%|████████▏ | 10856/13317 [1:52:10<25:12,  1.63it/s]
                                                       
{'loss': 0.1539, 'grad_norm': 0.6020524501800537, 'learning_rate': 1.0047857355199674e-06, 'epoch': 2.45}

 82%|████████▏ | 10856/13317 [1:52:11<25:12,  1.63it/s]
 82%|████████▏ | 10857/13317 [1:52:11<25:11,  1.63it/s]
                                                       
{'loss': 0.1542, 'grad_norm': 0.9253699779510498, 'learning_rate': 1.0039978217918406e-06, 'epoch': 2.45}

 82%|████████▏ | 10857/13317 [1:52:11<25:11,  1.63it/s]
 82%|████████▏ | 10858/13317 [1:52:12<25:10,  1.63it/s]
                                                       
{'loss': 0.143, 'grad_norm': 0.5316133499145508, 'learning_rate': 1.0032101826313733e-06, 'epoch': 2.45}

 82%|████████▏ | 10858/13317 [1:52:12<25:10,  1.63it/s]
 82%|████████▏ | 10859/13317 [1:52:12<25:09,  1.63it/s]
                                                       
{'loss': 0.1507, 'grad_norm': 0.5384863615036011, 'learning_rate': 1.0024228180926865e-06, 'epoch': 2.45}

 82%|████████▏ | 10859/13317 [1:52:12<25:09,  1.63it/s]
 82%|████████▏ | 10860/13317 [1:52:13<25:09,  1.63it/s]
                                                       
{'loss': 0.1427, 'grad_norm': 0.5040905475616455, 'learning_rate': 1.0016357282298793e-06, 'epoch': 2.45}

 82%|████████▏ | 10860/13317 [1:52:13<25:09,  1.63it/s]
 82%|████████▏ | 10861/13317 [1:52:13<25:08,  1.63it/s]
                                                       
{'loss': 0.1215, 'grad_norm': 0.5331257581710815, 'learning_rate': 1.0008489130970317e-06, 'epoch': 2.45}

 82%|████████▏ | 10861/13317 [1:52:14<25:08,  1.63it/s]
 82%|████████▏ | 10862/13317 [1:52:14<25:07,  1.63it/s]
                                                       
{'loss': 0.1579, 'grad_norm': 0.5536611080169678, 'learning_rate': 1.0000623727482084e-06, 'epoch': 2.45}

 82%|████████▏ | 10862/13317 [1:52:14<25:07,  1.63it/s]
 82%|████████▏ | 10863/13317 [1:52:15<25:07,  1.63it/s]
                                                       
{'loss': 0.1835, 'grad_norm': 0.6852261424064636, 'learning_rate': 9.992761072374523e-07, 'epoch': 2.45}

 82%|████████▏ | 10863/13317 [1:52:15<25:07,  1.63it/s]
 82%|████████▏ | 10864/13317 [1:52:15<25:06,  1.63it/s]
                                                       
{'loss': 0.1408, 'grad_norm': 0.5350165367126465, 'learning_rate': 9.984901166187866e-07, 'epoch': 2.45}

 82%|████████▏ | 10864/13317 [1:52:15<25:06,  1.63it/s]
 82%|████████▏ | 10865/13317 [1:52:16<25:07,  1.63it/s]
                                                       
{'loss': 0.1281, 'grad_norm': 0.5027930736541748, 'learning_rate': 9.977044009462189e-07, 'epoch': 2.45}

 82%|████████▏ | 10865/13317 [1:52:16<25:07,  1.63it/s]
 82%|████████▏ | 10866/13317 [1:52:17<25:05,  1.63it/s]
                                                       
{'loss': 0.134, 'grad_norm': 0.5654463171958923, 'learning_rate': 9.969189602737366e-07, 'epoch': 2.45}

 82%|████████▏ | 10866/13317 [1:52:17<25:05,  1.63it/s]
 82%|████████▏ | 10867/13317 [1:52:17<25:03,  1.63it/s]
                                                       
{'loss': 0.1501, 'grad_norm': 0.5852257609367371, 'learning_rate': 9.96133794655308e-07, 'epoch': 2.45}

 82%|████████▏ | 10867/13317 [1:52:17<25:03,  1.63it/s]
 82%|████████▏ | 10868/13317 [1:52:18<25:03,  1.63it/s]
                                                       
{'loss': 0.1395, 'grad_norm': 0.6325884461402893, 'learning_rate': 9.953489041448788e-07, 'epoch': 2.45}

 82%|████████▏ | 10868/13317 [1:52:18<25:03,  1.63it/s]
 82%|████████▏ | 10869/13317 [1:52:18<25:01,  1.63it/s]
                                                       
{'loss': 0.1615, 'grad_norm': 0.6698064804077148, 'learning_rate': 9.945642887963842e-07, 'epoch': 2.45}

 82%|████████▏ | 10869/13317 [1:52:19<25:01,  1.63it/s]
 82%|████████▏ | 10870/13317 [1:52:19<25:02,  1.63it/s]
                                                       
{'loss': 0.173, 'grad_norm': 0.6797378063201904, 'learning_rate': 9.937799486637323e-07, 'epoch': 2.45}

 82%|████████▏ | 10870/13317 [1:52:19<25:02,  1.63it/s]
 82%|████████▏ | 10871/13317 [1:52:20<25:01,  1.63it/s]
                                                       
{'loss': 0.182, 'grad_norm': 0.6802011728286743, 'learning_rate': 9.929958838008157e-07, 'epoch': 2.45}

 82%|████████▏ | 10871/13317 [1:52:20<25:01,  1.63it/s]
 82%|████████▏ | 10872/13317 [1:52:20<25:01,  1.63it/s]
                                                       
{'loss': 0.1621, 'grad_norm': 0.6475291848182678, 'learning_rate': 9.922120942615081e-07, 'epoch': 2.45}

 82%|████████▏ | 10872/13317 [1:52:20<25:01,  1.63it/s]
 82%|████████▏ | 10873/13317 [1:52:21<25:00,  1.63it/s]
                                                       
{'loss': 0.1366, 'grad_norm': 0.5131414532661438, 'learning_rate': 9.91428580099667e-07, 'epoch': 2.45}

 82%|████████▏ | 10873/13317 [1:52:21<25:00,  1.63it/s]
 82%|████████▏ | 10874/13317 [1:52:21<24:59,  1.63it/s]
                                                       
{'loss': 0.1432, 'grad_norm': 0.5703161358833313, 'learning_rate': 9.906453413691252e-07, 'epoch': 2.45}

 82%|████████▏ | 10874/13317 [1:52:22<24:59,  1.63it/s]
 82%|████████▏ | 10875/13317 [1:52:22<25:01,  1.63it/s]
                                                       
{'loss': 0.1637, 'grad_norm': 0.6593710780143738, 'learning_rate': 9.898623781236994e-07, 'epoch': 2.45}

 82%|████████▏ | 10875/13317 [1:52:22<25:01,  1.63it/s]
 82%|████████▏ | 10876/13317 [1:52:23<25:00,  1.63it/s]
                                                       
{'loss': 0.1326, 'grad_norm': 0.48617029190063477, 'learning_rate': 9.890796904171872e-07, 'epoch': 2.45}

 82%|████████▏ | 10876/13317 [1:52:23<25:00,  1.63it/s]
 82%|████████▏ | 10877/13317 [1:52:23<24:58,  1.63it/s]
                                                       
{'loss': 0.1549, 'grad_norm': 0.5651014447212219, 'learning_rate': 9.882972783033723e-07, 'epoch': 2.45}

 82%|████████▏ | 10877/13317 [1:52:23<24:58,  1.63it/s]
 82%|████████▏ | 10878/13317 [1:52:24<24:57,  1.63it/s]
                                                       
{'loss': 0.1348, 'grad_norm': 0.468183308839798, 'learning_rate': 9.87515141836008e-07, 'epoch': 2.45}

 82%|████████▏ | 10878/13317 [1:52:24<24:57,  1.63it/s]
 82%|████████▏ | 10879/13317 [1:52:25<24:56,  1.63it/s]
                                                       
{'loss': 0.1727, 'grad_norm': 0.5900207757949829, 'learning_rate': 9.867332810688379e-07, 'epoch': 2.45}

 82%|████████▏ | 10879/13317 [1:52:25<24:56,  1.63it/s]
 82%|████████▏ | 10880/13317 [1:52:25<24:58,  1.63it/s]
                                                       
{'loss': 0.1444, 'grad_norm': 0.5656198263168335, 'learning_rate': 9.859516960555864e-07, 'epoch': 2.45}

 82%|████████▏ | 10880/13317 [1:52:25<24:58,  1.63it/s]
 82%|████████▏ | 10881/13317 [1:52:26<24:56,  1.63it/s]
                                                       
{'loss': 0.1277, 'grad_norm': 0.5044808387756348, 'learning_rate': 9.851703868499546e-07, 'epoch': 2.45}

 82%|████████▏ | 10881/13317 [1:52:26<24:56,  1.63it/s]
 82%|████████▏ | 10882/13317 [1:52:26<24:55,  1.63it/s]
                                                       
{'loss': 0.1583, 'grad_norm': 0.5790465474128723, 'learning_rate': 9.84389353505626e-07, 'epoch': 2.45}

 82%|████████▏ | 10882/13317 [1:52:26<24:55,  1.63it/s]
 82%|████████▏ | 10883/13317 [1:52:27<24:54,  1.63it/s]
                                                       
{'loss': 0.1551, 'grad_norm': 0.7103495597839355, 'learning_rate': 9.836085960762665e-07, 'epoch': 2.45}

 82%|████████▏ | 10883/13317 [1:52:27<24:54,  1.63it/s]
 82%|████████▏ | 10884/13317 [1:52:28<24:53,  1.63it/s]
                                                       
{'loss': 0.1298, 'grad_norm': 0.5251473188400269, 'learning_rate': 9.82828114615525e-07, 'epoch': 2.45}

 82%|████████▏ | 10884/13317 [1:52:28<24:53,  1.63it/s]
 82%|████████▏ | 10885/13317 [1:52:28<24:53,  1.63it/s]
                                                       
{'loss': 0.1602, 'grad_norm': 0.6300309300422668, 'learning_rate': 9.82047909177024e-07, 'epoch': 2.45}

 82%|████████▏ | 10885/13317 [1:52:28<24:53,  1.63it/s]
 82%|████████▏ | 10886/13317 [1:52:29<24:52,  1.63it/s]
                                                       
{'loss': 0.1427, 'grad_norm': 0.5396043062210083, 'learning_rate': 9.812679798143748e-07, 'epoch': 2.45}

 82%|████████▏ | 10886/13317 [1:52:29<24:52,  1.63it/s]
 82%|████████▏ | 10887/13317 [1:52:29<24:51,  1.63it/s]
                                                       
{'loss': 0.1473, 'grad_norm': 0.5466325879096985, 'learning_rate': 9.80488326581167e-07, 'epoch': 2.45}

 82%|████████▏ | 10887/13317 [1:52:30<24:51,  1.63it/s]
 82%|████████▏ | 10888/13317 [1:52:30<24:51,  1.63it/s]
                                                       
{'loss': 0.1254, 'grad_norm': 0.45710885524749756, 'learning_rate': 9.79708949530971e-07, 'epoch': 2.45}

 82%|████████▏ | 10888/13317 [1:52:30<24:51,  1.63it/s]
 82%|████████▏ | 10889/13317 [1:52:31<24:50,  1.63it/s]
                                                       
{'loss': 0.1346, 'grad_norm': 0.5517166256904602, 'learning_rate': 9.789298487173364e-07, 'epoch': 2.45}

 82%|████████▏ | 10889/13317 [1:52:31<24:50,  1.63it/s]
 82%|████████▏ | 10890/13317 [1:52:31<24:52,  1.63it/s]
                                                       
{'loss': 0.168, 'grad_norm': 0.5905390381813049, 'learning_rate': 9.78151024193798e-07, 'epoch': 2.45}

 82%|████████▏ | 10890/13317 [1:52:31<24:52,  1.63it/s]
 82%|████████▏ | 10891/13317 [1:52:32<24:50,  1.63it/s]
                                                       
{'loss': 0.1347, 'grad_norm': 0.5038378834724426, 'learning_rate': 9.773724760138687e-07, 'epoch': 2.45}

 82%|████████▏ | 10891/13317 [1:52:32<24:50,  1.63it/s]
 82%|████████▏ | 10892/13317 [1:52:33<24:48,  1.63it/s]
                                                       
{'loss': 0.2339, 'grad_norm': 0.6497934460639954, 'learning_rate': 9.765942042310406e-07, 'epoch': 2.45}

 82%|████████▏ | 10892/13317 [1:52:33<24:48,  1.63it/s]
 82%|████████▏ | 10893/13317 [1:52:33<24:47,  1.63it/s]
                                                       
{'loss': 0.1364, 'grad_norm': 0.5460111498832703, 'learning_rate': 9.758162088987915e-07, 'epoch': 2.45}

 82%|████████▏ | 10893/13317 [1:52:33<24:47,  1.63it/s]
 82%|████████▏ | 10894/13317 [1:52:34<24:47,  1.63it/s]
                                                       
{'loss': 0.126, 'grad_norm': 0.48599255084991455, 'learning_rate': 9.750384900705789e-07, 'epoch': 2.45}

 82%|████████▏ | 10894/13317 [1:52:34<24:47,  1.63it/s]
 82%|████████▏ | 10895/13317 [1:52:34<24:54,  1.62it/s]
                                                       
{'loss': 0.1582, 'grad_norm': 0.6202681064605713, 'learning_rate': 9.742610477998394e-07, 'epoch': 2.45}

 82%|████████▏ | 10895/13317 [1:52:34<24:54,  1.62it/s]
 82%|████████▏ | 10896/13317 [1:52:35<24:51,  1.62it/s]
                                                       
{'loss': 0.1316, 'grad_norm': 0.5332838892936707, 'learning_rate': 9.73483882139989e-07, 'epoch': 2.45}

 82%|████████▏ | 10896/13317 [1:52:35<24:51,  1.62it/s]
 82%|████████▏ | 10897/13317 [1:52:36<24:49,  1.62it/s]
                                                       
{'loss': 0.1729, 'grad_norm': 0.6212320923805237, 'learning_rate': 9.72706993144431e-07, 'epoch': 2.45}

 82%|████████▏ | 10897/13317 [1:52:36<24:49,  1.62it/s]
 82%|████████▏ | 10898/13317 [1:52:36<24:46,  1.63it/s]
                                                       
{'loss': 0.1651, 'grad_norm': 0.649350643157959, 'learning_rate': 9.719303808665431e-07, 'epoch': 2.46}

 82%|████████▏ | 10898/13317 [1:52:36<24:46,  1.63it/s]
 82%|████████▏ | 10899/13317 [1:52:37<24:46,  1.63it/s]
                                                       
{'loss': 0.1206, 'grad_norm': 0.4936501979827881, 'learning_rate': 9.711540453596891e-07, 'epoch': 2.46}

 82%|████████▏ | 10899/13317 [1:52:37<24:46,  1.63it/s]
 82%|████████▏ | 10900/13317 [1:52:37<24:45,  1.63it/s]
                                                       
{'loss': 0.1446, 'grad_norm': 0.5855985879898071, 'learning_rate': 9.70377986677209e-07, 'epoch': 2.46}

 82%|████████▏ | 10900/13317 [1:52:38<24:45,  1.63it/s]
 82%|████████▏ | 10901/13317 [1:52:38<24:44,  1.63it/s]
                                                       
{'loss': 0.1615, 'grad_norm': 0.6664443612098694, 'learning_rate': 9.69602204872429e-07, 'epoch': 2.46}

 82%|████████▏ | 10901/13317 [1:52:38<24:44,  1.63it/s]
 82%|████████▏ | 10902/13317 [1:52:39<24:42,  1.63it/s]
                                                       
{'loss': 0.1653, 'grad_norm': 0.6105704307556152, 'learning_rate': 9.688266999986518e-07, 'epoch': 2.46}

 82%|████████▏ | 10902/13317 [1:52:39<24:42,  1.63it/s]
 82%|████████▏ | 10903/13317 [1:52:39<24:41,  1.63it/s]
                                                       
{'loss': 0.1219, 'grad_norm': 0.4778488576412201, 'learning_rate': 9.680514721091623e-07, 'epoch': 2.46}

 82%|████████▏ | 10903/13317 [1:52:39<24:41,  1.63it/s]
 82%|████████▏ | 10904/13317 [1:52:40<24:41,  1.63it/s]
                                                       
{'loss': 0.1435, 'grad_norm': 0.5512971878051758, 'learning_rate': 9.672765212572283e-07, 'epoch': 2.46}

 82%|████████▏ | 10904/13317 [1:52:40<24:41,  1.63it/s]
 82%|████████▏ | 10905/13317 [1:52:40<24:40,  1.63it/s]
                                                       
{'loss': 0.16, 'grad_norm': 0.5928704738616943, 'learning_rate': 9.665018474960951e-07, 'epoch': 2.46}

 82%|████████▏ | 10905/13317 [1:52:41<24:40,  1.63it/s]
 82%|████████▏ | 10906/13317 [1:52:41<24:40,  1.63it/s]
                                                       
{'loss': 0.1241, 'grad_norm': 0.5037896037101746, 'learning_rate': 9.657274508789938e-07, 'epoch': 2.46}

 82%|████████▏ | 10906/13317 [1:52:41<24:40,  1.63it/s]
 82%|████████▏ | 10907/13317 [1:52:42<24:38,  1.63it/s]
                                                       
{'loss': 0.1513, 'grad_norm': 0.6376506090164185, 'learning_rate': 9.649533314591326e-07, 'epoch': 2.46}

 82%|████████▏ | 10907/13317 [1:52:42<24:38,  1.63it/s]
 82%|████████▏ | 10908/13317 [1:52:42<24:38,  1.63it/s]
                                                       
{'loss': 0.1345, 'grad_norm': 0.554567277431488, 'learning_rate': 9.641794892896995e-07, 'epoch': 2.46}

 82%|████████▏ | 10908/13317 [1:52:42<24:38,  1.63it/s]
 82%|████████▏ | 10909/13317 [1:52:43<24:37,  1.63it/s]
                                                       
{'loss': 0.178, 'grad_norm': 0.6230900883674622, 'learning_rate': 9.634059244238691e-07, 'epoch': 2.46}

 82%|████████▏ | 10909/13317 [1:52:43<24:37,  1.63it/s]
 82%|████████▏ | 10910/13317 [1:52:44<24:37,  1.63it/s]
                                                       
{'loss': 0.1508, 'grad_norm': 0.5478876829147339, 'learning_rate': 9.626326369147904e-07, 'epoch': 2.46}

 82%|████████▏ | 10910/13317 [1:52:44<24:37,  1.63it/s]
 82%|████████▏ | 10911/13317 [1:52:44<24:36,  1.63it/s]
                                                       
{'loss': 0.1522, 'grad_norm': 0.596072256565094, 'learning_rate': 9.618596268155988e-07, 'epoch': 2.46}

 82%|████████▏ | 10911/13317 [1:52:44<24:36,  1.63it/s]
 82%|████████▏ | 10912/13317 [1:52:45<24:35,  1.63it/s]
                                                       
{'loss': 0.1827, 'grad_norm': 0.6413159370422363, 'learning_rate': 9.610868941794065e-07, 'epoch': 2.46}

 82%|████████▏ | 10912/13317 [1:52:45<24:35,  1.63it/s]
 82%|████████▏ | 10913/13317 [1:52:45<24:34,  1.63it/s]
                                                       
{'loss': 0.1127, 'grad_norm': 0.4294220805168152, 'learning_rate': 9.603144390593106e-07, 'epoch': 2.46}

 82%|████████▏ | 10913/13317 [1:52:46<24:34,  1.63it/s]
 82%|████████▏ | 10914/13317 [1:52:46<24:33,  1.63it/s]
                                                       
{'loss': 0.1206, 'grad_norm': 0.4867851734161377, 'learning_rate': 9.59542261508385e-07, 'epoch': 2.46}

 82%|████████▏ | 10914/13317 [1:52:46<24:33,  1.63it/s]
 82%|████████▏ | 10915/13317 [1:52:47<24:34,  1.63it/s]
                                                       
{'loss': 0.1421, 'grad_norm': 0.4718650281429291, 'learning_rate': 9.587703615796863e-07, 'epoch': 2.46}

 82%|████████▏ | 10915/13317 [1:52:47<24:34,  1.63it/s]
 82%|████████▏ | 10916/13317 [1:52:47<24:33,  1.63it/s]
                                                       
{'loss': 0.1306, 'grad_norm': 0.48713210225105286, 'learning_rate': 9.579987393262542e-07, 'epoch': 2.46}

 82%|████████▏ | 10916/13317 [1:52:47<24:33,  1.63it/s]
 82%|████████▏ | 10917/13317 [1:52:48<24:32,  1.63it/s]
                                                       
{'loss': 0.1436, 'grad_norm': 0.5817557573318481, 'learning_rate': 9.572273948011064e-07, 'epoch': 2.46}

 82%|████████▏ | 10917/13317 [1:52:48<24:32,  1.63it/s]
 82%|████████▏ | 10918/13317 [1:52:48<24:31,  1.63it/s]
                                                       
{'loss': 0.1524, 'grad_norm': 0.5705589056015015, 'learning_rate': 9.564563280572402e-07, 'epoch': 2.46}

 82%|████████▏ | 10918/13317 [1:52:49<24:31,  1.63it/s]
 82%|████████▏ | 10919/13317 [1:52:49<24:30,  1.63it/s]
                                                       
{'loss': 0.1575, 'grad_norm': 0.6306602358818054, 'learning_rate': 9.556855391476382e-07, 'epoch': 2.46}

 82%|████████▏ | 10919/13317 [1:52:49<24:30,  1.63it/s]
 82%|████████▏ | 10920/13317 [1:52:50<24:32,  1.63it/s]
                                                       
{'loss': 0.1293, 'grad_norm': 0.591397762298584, 'learning_rate': 9.549150281252633e-07, 'epoch': 2.46}

 82%|████████▏ | 10920/13317 [1:52:50<24:32,  1.63it/s]
 82%|████████▏ | 10921/13317 [1:52:50<24:31,  1.63it/s]
                                                       
{'loss': 0.1392, 'grad_norm': 0.5810797810554504, 'learning_rate': 9.541447950430554e-07, 'epoch': 2.46}

 82%|████████▏ | 10921/13317 [1:52:50<24:31,  1.63it/s]
 82%|████████▏ | 10922/13317 [1:52:51<24:29,  1.63it/s]
                                                       
{'loss': 0.1266, 'grad_norm': 0.4957957863807678, 'learning_rate': 9.533748399539378e-07, 'epoch': 2.46}

 82%|████████▏ | 10922/13317 [1:52:51<24:29,  1.63it/s]
 82%|████████▏ | 10923/13317 [1:52:52<24:28,  1.63it/s]
                                                       
{'loss': 0.1345, 'grad_norm': 0.662629246711731, 'learning_rate': 9.526051629108141e-07, 'epoch': 2.46}

 82%|████████▏ | 10923/13317 [1:52:52<24:28,  1.63it/s]
 82%|████████▏ | 10924/13317 [1:52:52<24:27,  1.63it/s]
                                                       
{'loss': 0.1482, 'grad_norm': 0.5274332761764526, 'learning_rate': 9.518357639665732e-07, 'epoch': 2.46}

 82%|████████▏ | 10924/13317 [1:52:52<24:27,  1.63it/s]
 82%|████████▏ | 10925/13317 [1:52:53<24:29,  1.63it/s]
                                                       
{'loss': 0.1703, 'grad_norm': 0.6960853934288025, 'learning_rate': 9.510666431740757e-07, 'epoch': 2.46}

 82%|████████▏ | 10925/13317 [1:52:53<24:29,  1.63it/s]
 82%|████████▏ | 10926/13317 [1:52:53<24:32,  1.62it/s]
                                                       
{'loss': 0.1411, 'grad_norm': 0.6253756880760193, 'learning_rate': 9.502978005861702e-07, 'epoch': 2.46}

 82%|████████▏ | 10926/13317 [1:52:54<24:32,  1.62it/s]
 82%|████████▏ | 10927/13317 [1:52:54<24:29,  1.63it/s]
                                                       
{'loss': 0.1716, 'grad_norm': 0.7579858303070068, 'learning_rate': 9.495292362556863e-07, 'epoch': 2.46}

 82%|████████▏ | 10927/13317 [1:52:54<24:29,  1.63it/s]
 82%|████████▏ | 10928/13317 [1:52:55<24:27,  1.63it/s]
                                                       
{'loss': 0.1313, 'grad_norm': 0.5243228673934937, 'learning_rate': 9.487609502354306e-07, 'epoch': 2.46}

 82%|████████▏ | 10928/13317 [1:52:55<24:27,  1.63it/s]
 82%|████████▏ | 10929/13317 [1:52:55<24:26,  1.63it/s]
                                                       
{'loss': 0.238, 'grad_norm': 0.5452648997306824, 'learning_rate': 9.47992942578192e-07, 'epoch': 2.46}

 82%|████████▏ | 10929/13317 [1:52:55<24:26,  1.63it/s]
 82%|████████▏ | 10930/13317 [1:52:56<24:27,  1.63it/s]
                                                       
{'loss': 0.2044, 'grad_norm': 0.771108090877533, 'learning_rate': 9.47225213336741e-07, 'epoch': 2.46}

 82%|████████▏ | 10930/13317 [1:52:56<24:27,  1.63it/s]
 82%|████████▏ | 10931/13317 [1:52:56<24:25,  1.63it/s]
                                                       
{'loss': 0.1535, 'grad_norm': 0.7624255418777466, 'learning_rate': 9.464577625638322e-07, 'epoch': 2.46}

 82%|████████▏ | 10931/13317 [1:52:57<24:25,  1.63it/s]
 82%|████████▏ | 10932/13317 [1:52:57<24:24,  1.63it/s]
                                                       
{'loss': 0.1507, 'grad_norm': 0.605989396572113, 'learning_rate': 9.456905903121922e-07, 'epoch': 2.46}

 82%|████████▏ | 10932/13317 [1:52:57<24:24,  1.63it/s]
 82%|████████▏ | 10933/13317 [1:52:58<24:23,  1.63it/s]
                                                       
{'loss': 0.1417, 'grad_norm': 0.524791419506073, 'learning_rate': 9.449236966345366e-07, 'epoch': 2.46}

 82%|████████▏ | 10933/13317 [1:52:58<24:23,  1.63it/s]
 82%|████████▏ | 10934/13317 [1:52:58<24:23,  1.63it/s]
                                                       
{'loss': 0.1849, 'grad_norm': 0.6624841094017029, 'learning_rate': 9.441570815835594e-07, 'epoch': 2.46}

 82%|████████▏ | 10934/13317 [1:52:58<24:23,  1.63it/s]
 82%|████████▏ | 10935/13317 [1:52:59<24:23,  1.63it/s]
                                                       
{'loss': 0.2074, 'grad_norm': 0.7997151613235474, 'learning_rate': 9.433907452119351e-07, 'epoch': 2.46}

 82%|████████▏ | 10935/13317 [1:52:59<24:23,  1.63it/s]
 82%|████████▏ | 10936/13317 [1:53:00<24:22,  1.63it/s]
                                                       
{'loss': 0.179, 'grad_norm': 0.6953026652336121, 'learning_rate': 9.426246875723178e-07, 'epoch': 2.46}

 82%|████████▏ | 10936/13317 [1:53:00<24:22,  1.63it/s]
 82%|████████▏ | 10937/13317 [1:53:00<24:20,  1.63it/s]
                                                       
{'loss': 0.1633, 'grad_norm': 0.5847195386886597, 'learning_rate': 9.418589087173441e-07, 'epoch': 2.46}

 82%|████████▏ | 10937/13317 [1:53:00<24:20,  1.63it/s]
 82%|████████▏ | 10938/13317 [1:53:01<24:20,  1.63it/s]
                                                       
{'loss': 0.1263, 'grad_norm': 0.5628577470779419, 'learning_rate': 9.410934086996343e-07, 'epoch': 2.46}

 82%|████████▏ | 10938/13317 [1:53:01<24:20,  1.63it/s]
 82%|████████▏ | 10939/13317 [1:53:01<24:18,  1.63it/s]
                                                       
{'loss': 0.145, 'grad_norm': 0.5604367256164551, 'learning_rate': 9.403281875717812e-07, 'epoch': 2.46}

 82%|████████▏ | 10939/13317 [1:53:01<24:18,  1.63it/s]
 82%|████████▏ | 10940/13317 [1:53:02<24:18,  1.63it/s]
                                                       
{'loss': 0.1156, 'grad_norm': 0.44971171021461487, 'learning_rate': 9.395632453863657e-07, 'epoch': 2.46}

 82%|████████▏ | 10940/13317 [1:53:02<24:18,  1.63it/s]
 82%|████████▏ | 10941/13317 [1:53:03<24:18,  1.63it/s]
                                                       
{'loss': 0.1412, 'grad_norm': 0.5122118592262268, 'learning_rate': 9.3879858219595e-07, 'epoch': 2.46}

 82%|████████▏ | 10941/13317 [1:53:03<24:18,  1.63it/s]
 82%|████████▏ | 10942/13317 [1:53:03<24:17,  1.63it/s]
                                                       
{'loss': 0.1741, 'grad_norm': 0.6843433976173401, 'learning_rate': 9.380341980530716e-07, 'epoch': 2.46}

 82%|████████▏ | 10942/13317 [1:53:03<24:17,  1.63it/s]
 82%|████████▏ | 10943/13317 [1:53:04<24:17,  1.63it/s]
                                                       
{'loss': 0.1728, 'grad_norm': 0.6344714760780334, 'learning_rate': 9.372700930102519e-07, 'epoch': 2.47}

 82%|████████▏ | 10943/13317 [1:53:04<24:17,  1.63it/s]
 82%|████████▏ | 10944/13317 [1:53:04<24:16,  1.63it/s]
                                                       
{'loss': 0.1451, 'grad_norm': 0.5962556004524231, 'learning_rate': 9.365062671199931e-07, 'epoch': 2.47}

 82%|████████▏ | 10944/13317 [1:53:05<24:16,  1.63it/s]
 82%|████████▏ | 10945/13317 [1:53:05<24:15,  1.63it/s]
                                                       
{'loss': 0.1359, 'grad_norm': 0.5293915271759033, 'learning_rate': 9.357427204347802e-07, 'epoch': 2.47}

 82%|████████▏ | 10945/13317 [1:53:05<24:15,  1.63it/s]
 82%|████████▏ | 10946/13317 [1:53:06<24:15,  1.63it/s]
                                                       
{'loss': 0.1445, 'grad_norm': 0.6062310338020325, 'learning_rate': 9.349794530070755e-07, 'epoch': 2.47}

 82%|████████▏ | 10946/13317 [1:53:06<24:15,  1.63it/s]
 82%|████████▏ | 10947/13317 [1:53:06<24:14,  1.63it/s]
                                                       
{'loss': 0.152, 'grad_norm': 0.539785623550415, 'learning_rate': 9.342164648893227e-07, 'epoch': 2.47}

 82%|████████▏ | 10947/13317 [1:53:06<24:14,  1.63it/s]